BioAssignmentNotebook.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 297,
   "metadata": {},
   "outputs": [],
   "source": [
    "#import os\n",
    "#os.chdir(\"C:\\\\Users\\\\User\\\\Dropbox\\\\MSc_UCL\\\\CURRENT COURSES\\\\Term 2\\\\Bioinformatics\\\\Assignment\\\\code\")\n",
    "#os.chdir('/home/davide/Dropbox/MSc_UCL/CURRENT COURSES/Term 2/Bioinformatics/Assignment/code')\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import Bio\n",
    "import seaborn as sns\n",
    "import scipy\n",
    "import re\n",
    "import random\n",
    "import itertools\n",
    "\n",
    "from load_data import load_data\n",
    "from preprocess_data import preprocess_data\n",
    "from boxplot_features import boxplot_features\n",
    "from correlation_plot import correlation_plot\n",
    "from knn import knn\n",
    "from svm import svm\n",
    "\n",
    "from Bio import SeqIO\n",
    "from Bio.SeqUtils.ProtParam import ProteinAnalysis\n",
    "from Bio.SeqUtils import ProtParamData\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn import preprocessing\n",
    "from sklearn.preprocessing import StandardScaler, label_binarize\n",
    "from sklearn.utils import shuffle\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "\n",
    "import xgboost as xgb\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "from sklearn.metrics import roc_auc_score, confusion_matrix, classification_report\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "from sklearn.feature_selection import RFE\n",
    "from sklearn import tree\n",
    "from sklearn.ensemble import RandomForestClassifier, AdaBoostClassifier, GradientBoostingClassifier, VotingClassifier\n",
    "from sklearn.linear_model import RidgeCV, LassoCV, Ridge, Lasso\n",
    "from sklearn.model_selection import KFold, cross_val_predict, cross_val_score\n",
    "from sklearn.svm import SVC"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Raw Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_data(text_file, subcellular_location, drop_long_seq = False, inc_aa_perc=True):\n",
    "    if inc_aa_perc == True:\n",
    "        df = pd.DataFrame(columns=('ID',  \n",
    "                                   'Sequence_Length', \n",
    "                                   'Molar_Extinction_Coefficient',\n",
    "                                   'Isoelectric_Point', \n",
    "                                   'Aromaticity',\n",
    "                                   'Secondary_Structure_Helix',\n",
    "                                   'Secondary_Structure_Turn',\n",
    "                                   'Secondary_Structure_Sheet',\n",
    "                                   'Mean_Flexibility',\n",
    "                                   'Gravy',\n",
    "                                   'Molecular_Weight',\n",
    "                                   'Instability_Index',\n",
    "                                   'Ag', 'Cg', 'Dg', 'Eg', 'Fg',\n",
    "                                   'Gg', 'Hg', 'Ig', 'Kg', 'Lg',\n",
    "                                   'Mg', 'Ng', 'Pg', 'Qg', 'Rg',\n",
    "                                   'Sg', 'Tg', 'Vg', 'Wg', 'Yg',\n",
    "                                   'As', 'Cs', 'Ds', 'Es', 'Fs',\n",
    "                                   'Gs', 'Hs', 'Is', 'Ks', 'Ls',\n",
    "                                   'Ms', 'Ns', 'Ps', 'Qs', 'Rs',\n",
    "                                   'Ss', 'Ts', 'Vs', 'Ws', 'Ys',\n",
    "                                   'Ae', 'Ce', 'De', 'Ee', 'Fe',\n",
    "                                   'Ge', 'He', 'Ie', 'Ke', 'Le',\n",
    "                                   'Me', 'Ne', 'Pe', 'Qe', 'Re',\n",
    "                                   'Se', 'Te', 'Ve', 'We', 'Ye',\n",
    "                                   'Subcellular_Location'))\n",
    "    if inc_aa_perc == False:\n",
    "        df = pd.DataFrame(columns=('ID',  \n",
    "                                   'Sequence_Length', \n",
    "                                   'Molar_Extinction_Coefficient',\n",
    "                                   'Isoelectric_Point', \n",
    "                                   'Aromaticity',\n",
    "                                   'Secondary_Structure_Helix',\n",
    "                                   'Secondary_Structure_Turn',\n",
    "                                   'Secondary_Structure_Sheet',\n",
    "                                   'Mean_Flexibility',\n",
    "                                   'Gravy',\n",
    "                                   'Molecular_Weight',\n",
    "                                   'Instability_Index',\n",
    "                                   'Subcellular_Location'))\n",
    "    bad_seq = []\n",
    "    length_list = []\n",
    "    i=0 \n",
    "    for seq_record in SeqIO.parse(text_file, \"fasta\"):\n",
    "        # get sequence id\n",
    "        seq_id = seq_record.id\n",
    "        \n",
    "        # get species\n",
    "        #_, species = seq_id.split(\"_\", 1)\n",
    "        # get sequence length\n",
    "        seq_len = len(seq_record)\n",
    "        length_list.append(seq_len)\n",
    "        if drop_long_seq:\n",
    "            if seq_len > 2000:\n",
    "                continue\n",
    "        \n",
    "        \n",
    "        seq_rep = str(seq_record.seq)\n",
    "        \n",
    "        \n",
    "        # locate offending characters\n",
    "        B_list = [pos for pos, char in enumerate(seq_rep) if char == 'B']\n",
    "        Z_list = [pos for pos, char in enumerate(seq_rep) if char == 'Z']\n",
    "        X_list = [pos for pos, char in enumerate(seq_rep) if char == 'X']\n",
    "        U_list = [pos for pos, char in enumerate(seq_rep) if char == 'U']\n",
    "        O_list = [pos for pos, char in enumerate(seq_rep) if char == 'O']\n",
    "        \n",
    "        if len(O_list) != 0:\n",
    "            #print(\"O found at index {} in example {} of {}\".format(O_list,i,text_file))\n",
    "            for j in range(len(O_list)):\n",
    "                seq_rep = seq_rep[:O_list[j]] + 'C' + seq_rep[O_list[j]+1:]\n",
    "        \n",
    "        \n",
    "        # replace \"B\" by asparagine (\"N\")/aspartic acid (\"D\")\n",
    "        if len(B_list) != 0:\n",
    "            #print(\"B found at index {} in example {} of {}\".format(B_list,i,text_file))\n",
    "            for j in range(len(B_list)):\n",
    "                flip = random.randint(0, 1)\n",
    "                if flip == 1:\n",
    "                    seq_rep = seq_rep[:B_list[j]] + 'N' + seq_rep[B_list[j]+1:]\n",
    "                else:\n",
    "                    seq_rep = seq_rep[:B_list[j]] + 'D' + seq_rep[B_list[j]+1:]\n",
    "        # replace \"Z\" by glutamine (\"Q\")/glutamic acid (\"E\")\n",
    "        if len(Z_list) != 0:\n",
    "            #print(\"Z found at index {} in example {} of {}\".format(Z_list,i,text_file))\n",
    "            for j in range(len(Z_list)):\n",
    "                flip = random.randint(0, 1)\n",
    "                if flip == 1:\n",
    "                    seq_rep = seq_rep[:Z_list[j]] + 'Q' + seq_rep[Z_list[j]+1:]\n",
    "                else:\n",
    "                    seq_rep = seq_rep[:Z_list[j]] + 'E' + seq_rep[Z_list[j]+1:]\n",
    "                    \n",
    "        # replace \"U\" by cysteine (\"C\")\n",
    "        if len(U_list) != 0:\n",
    "            #print(\"U found at index {} in example {} of {}\".format(U_list,i,text_file))\n",
    "            for j in range(len(U_list)):\n",
    "                seq_rep = seq_rep[:U_list[j]] + 'C' + seq_rep[U_list[j]+1:]\n",
    "                \n",
    "        # replace \"X\" by a random aminoa acid\n",
    "        AA_alphabet = ['A','C','D','E','F','G','H','I','K','L','M','N','P','Q','R','S','T','V','W','Y']\n",
    "        if len(X_list) != 0:\n",
    "            #print(\"X found at index {} in example {} of {}\".format(X_list,i,text_file))\n",
    "            for j in range(len(X_list)):\n",
    "                flip = random.randint(0, 19)\n",
    "                seq_rep = seq_rep[:X_list[j]] + AA_alphabet[flip] + seq_rep[X_list[j]+1:]\n",
    "            \n",
    "      \n",
    "        \n",
    "        # get analysed sequence and global count       \n",
    "        analysed_seq = ProteinAnalysis(seq_rep)\n",
    "        global_count = analysed_seq.count_amino_acids()\n",
    "        \n",
    "        # get N and C terminal amino acid sequences and counts\n",
    "        analysed_start = ProteinAnalysis(seq_rep[:50])\n",
    "        analysed_end = ProteinAnalysis(seq_rep[-50:])\n",
    "        aa_start = analysed_start.count_amino_acids()\n",
    "        aa_end = analysed_end.count_amino_acids()\n",
    "        \n",
    "        # molar extinction coefficient\n",
    "        molar = np.mean(analysed_seq.molar_extinction_coefficient())\n",
    "\n",
    "        \n",
    "        # isoelectric point\n",
    "        isoelectric = analysed_seq.isoelectric_point() \n",
    "        # aromaticity\n",
    "        aromaticity = analysed_seq.aromaticity()\n",
    "        # secondary structure fraction\n",
    "        helix, turn, sheet = analysed_seq.secondary_structure_fraction()    \n",
    "        \n",
    "        \"\"\"Calculations below only possible if sequence doesn't have X so use try-except to set to Nan if there is an X\"\"\"   \n",
    "        \n",
    "        # gravy (grand average hydropathy)\n",
    "        try:\n",
    "            mean_flexibility = np.mean(analysed_seq.flexibility())\n",
    "        except:\n",
    "            bad_seq.append(seq_rep)\n",
    "            mean_flexibility = float('NaN')\n",
    "        # gravy (grand average hydropathy)\n",
    "        try:\n",
    "            gravy = analysed_seq.gravy()\n",
    "        except:\n",
    "            gravy = float('NaN')\n",
    "        # molecular weight\n",
    "        try:\n",
    "            molecular_weight = analysed_seq.molecular_weight()\n",
    "        except:\n",
    "            molecular_weight = float('NaN')\n",
    "        # instability index\n",
    "        try:\n",
    "            instability_idx = analysed_seq.instability_index()\n",
    "        except:\n",
    "            instability_idx = float('NaN')\n",
    "            \n",
    "\n",
    "            \n",
    "\n",
    "            \n",
    "        \n",
    "        if inc_aa_perc == True:\n",
    "            df.loc[i] = [seq_id,\n",
    "                  seq_len,\n",
    "                  molar,\n",
    "                  isoelectric,\n",
    "                  aromaticity,\n",
    "                  helix,\n",
    "                  turn,\n",
    "                  sheet,\n",
    "                  mean_flexibility,\n",
    "                  gravy,\n",
    "                  molecular_weight,\n",
    "                  instability_idx,\n",
    "                  global_count['A'],global_count['C'],global_count['D'],global_count['E'],global_count['F'],\n",
    "                  global_count['G'],global_count['H'],global_count['I'],global_count['K'],global_count['L'],\n",
    "                  global_count['M'],global_count['N'],global_count['P'],global_count['Q'],global_count['R'], \n",
    "                  global_count['S'],global_count['T'],global_count['V'],global_count['W'],global_count['Y'], \n",
    "                  aa_start['A'],aa_start['C'],aa_start['D'],aa_start['E'],aa_start['F'],\n",
    "                  aa_start['G'],aa_start['H'],aa_start['I'],aa_start['K'],aa_start['L'],\n",
    "                  aa_start['M'],aa_start['N'],aa_start['P'],aa_start['Q'],aa_start['R'], \n",
    "                  aa_start['S'],aa_start['T'],aa_start['V'],aa_start['W'],aa_start['Y'], \n",
    "                  aa_end['A'],aa_end['C'],aa_end['D'],aa_end['E'],aa_end['F'],\n",
    "                  aa_end['G'],aa_end['H'],aa_end['I'],aa_end['K'],aa_end['L'],\n",
    "                  aa_end['M'],aa_end['N'],aa_end['P'],aa_end['Q'],aa_end['R'], \n",
    "                  aa_end['S'],aa_end['T'],aa_end['V'],aa_end['W'],aa_end['Y'],\n",
    "                  subcellular_location]\n",
    " \n",
    "        if inc_aa_perc == False:\n",
    "            df.loc[i] = [seq_id,\n",
    "                  seq_len,\n",
    "                  molar,\n",
    "                  isoelectric,\n",
    "                  aromaticity,\n",
    "                  helix,\n",
    "                  turn,\n",
    "                  sheet,\n",
    "                  mean_flexibility,\n",
    "                  gravy,\n",
    "                  molecular_weight,\n",
    "                  instability_idx,\n",
    "                  subcellular_location]\n",
    "        i+=1\n",
    "        \n",
    "        # ensure the sequence length column is of dtype int\n",
    "        df.Sequence_Length = df.Sequence_Length.astype(int)\n",
    "    print(\"---------- {} ----------\".format(text_file))\n",
    "    print(\"Number of records = {}\".format(i))\n",
    "    print(\"Max string length = {}\".format(max(length_list)))\n",
    "    print(\"Min string length = {}\".format(min(length_list)))\n",
    "    print(\"Mean string length = {}\".format(np.mean(length_list)))\n",
    "\n",
    "    return df, length_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "---------- cyto2.txt ----------\n",
      "Number of records = 2924\n",
      "Max string length = 7393\n",
      "Min string length = 19\n",
      "Mean string length = 664.6654460719042\n",
      "---------- secreted2.txt ----------\n",
      "Number of records = 1586\n",
      "Max string length = 13100\n",
      "Min string length = 11\n",
      "Mean string length = 304.7102803738318\n",
      "---------- nuclear2.txt ----------\n",
      "Number of records = 3248\n",
      "Max string length = 5596\n",
      "Min string length = 35\n",
      "Mean string length = 624.4643934821968\n",
      "---------- mito2.txt ----------\n",
      "Number of records = 1297\n",
      "Max string length = 2628\n",
      "Min string length = 19\n",
      "Mean string length = 375.56966897613546\n",
      "------------------------------\n",
      "Number cystolic records = 2924\n",
      "Number of secreted records = 1586\n",
      "Number of nuclear records = 3248\n",
      "Number of mitochondrial records = 1297\n",
      "Total number of records = 9055\n"
     ]
    }
   ],
   "source": [
    "### Set random seed ###\n",
    "seed = 2020\n",
    "\n",
    "### Load data from each subcellular location ###\n",
    "\n",
    "drop_long_seq = True\n",
    "inc_aa_count_feature = False\n",
    "\n",
    "df_cystolic, c_list = load_data('cyto2.txt','cystolic',drop_long_seq,inc_aa_count_feature)\n",
    "df_secreted, s_list = load_data('secreted2.txt','secreted',drop_long_seq,inc_aa_count_feature)\n",
    "df_nuclear, n_list = load_data('nuclear2.txt','nuclear',drop_long_seq,inc_aa_count_feature)\n",
    "df_mitochondrial, m_list = load_data('mito2.txt','mitochondrial',drop_long_seq,inc_aa_count_feature)\n",
    "\n",
    "### Concatenate data and shuffle ###\n",
    "df_data = pd.concat([df_cystolic,df_secreted,df_nuclear,df_mitochondrial])\n",
    "num_records = len(df_data)\n",
    "print(\"------------------------------\")\n",
    "print(\"Number cystolic records = {}\".format(len(df_cystolic)))\n",
    "print(\"Number of secreted records = {}\".format(len(df_secreted)))\n",
    "print(\"Number of nuclear records = {}\".format(len(df_nuclear)))\n",
    "print(\"Number of mitochondrial records = {}\".format(len(df_mitochondrial)))\n",
    "print(\"Total number of records = {}\".format(num_records))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Investigate The Raw Data and Sequence Lengths"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>Sequence_Length</th>\n",
       "      <th>Molar_Extinction_Coefficient</th>\n",
       "      <th>Isoelectric_Point</th>\n",
       "      <th>Aromaticity</th>\n",
       "      <th>Secondary_Structure_Helix</th>\n",
       "      <th>Secondary_Structure_Turn</th>\n",
       "      <th>Secondary_Structure_Sheet</th>\n",
       "      <th>Mean_Flexibility</th>\n",
       "      <th>Gravy</th>\n",
       "      <th>Molecular_Weight</th>\n",
       "      <th>Instability_Index</th>\n",
       "      <th>Subcellular_Location</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>sp|Q4JIM5|ABL2_MOUSE</td>\n",
       "      <td>1182</td>\n",
       "      <td>119680.0</td>\n",
       "      <td>7.965393</td>\n",
       "      <td>0.067682</td>\n",
       "      <td>0.241963</td>\n",
       "      <td>0.296108</td>\n",
       "      <td>0.252961</td>\n",
       "      <td>1.008064</td>\n",
       "      <td>-0.509983</td>\n",
       "      <td>128194.9812</td>\n",
       "      <td>52.288164</td>\n",
       "      <td>cystolic</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>sp|P15891|ABP1_YEAST</td>\n",
       "      <td>592</td>\n",
       "      <td>60390.0</td>\n",
       "      <td>4.589172</td>\n",
       "      <td>0.060811</td>\n",
       "      <td>0.197635</td>\n",
       "      <td>0.285473</td>\n",
       "      <td>0.282095</td>\n",
       "      <td>1.025372</td>\n",
       "      <td>-1.042061</td>\n",
       "      <td>65575.3795</td>\n",
       "      <td>61.758615</td>\n",
       "      <td>cystolic</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>sp|P35609|ACTN2_HUMAN</td>\n",
       "      <td>894</td>\n",
       "      <td>127052.5</td>\n",
       "      <td>5.307678</td>\n",
       "      <td>0.078300</td>\n",
       "      <td>0.282998</td>\n",
       "      <td>0.164430</td>\n",
       "      <td>0.313199</td>\n",
       "      <td>1.003840</td>\n",
       "      <td>-0.597315</td>\n",
       "      <td>103852.5990</td>\n",
       "      <td>43.180884</td>\n",
       "      <td>cystolic</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>sp|P05095|ACTNA_DICDI</td>\n",
       "      <td>861</td>\n",
       "      <td>89060.0</td>\n",
       "      <td>5.001526</td>\n",
       "      <td>0.073171</td>\n",
       "      <td>0.298490</td>\n",
       "      <td>0.188153</td>\n",
       "      <td>0.295006</td>\n",
       "      <td>1.007373</td>\n",
       "      <td>-0.466783</td>\n",
       "      <td>97356.5559</td>\n",
       "      <td>41.535075</td>\n",
       "      <td>cystolic</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>sp|Q9Y2D8|ADIP_HUMAN</td>\n",
       "      <td>614</td>\n",
       "      <td>66787.5</td>\n",
       "      <td>6.029358</td>\n",
       "      <td>0.061889</td>\n",
       "      <td>0.250814</td>\n",
       "      <td>0.214984</td>\n",
       "      <td>0.268730</td>\n",
       "      <td>1.012336</td>\n",
       "      <td>-0.882410</td>\n",
       "      <td>71235.4260</td>\n",
       "      <td>52.350000</td>\n",
       "      <td>cystolic</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1292</td>\n",
       "      <td>sp|P38851|YSP1_YEAST</td>\n",
       "      <td>1228</td>\n",
       "      <td>177502.5</td>\n",
       "      <td>8.361145</td>\n",
       "      <td>0.118078</td>\n",
       "      <td>0.350163</td>\n",
       "      <td>0.205212</td>\n",
       "      <td>0.236156</td>\n",
       "      <td>1.000323</td>\n",
       "      <td>-0.366775</td>\n",
       "      <td>143581.8566</td>\n",
       "      <td>42.238363</td>\n",
       "      <td>mitochondrial</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1293</td>\n",
       "      <td>sp|Q9HGM3|YTA12_SCHPO</td>\n",
       "      <td>773</td>\n",
       "      <td>42352.5</td>\n",
       "      <td>9.393005</td>\n",
       "      <td>0.078913</td>\n",
       "      <td>0.296248</td>\n",
       "      <td>0.253558</td>\n",
       "      <td>0.245796</td>\n",
       "      <td>1.003549</td>\n",
       "      <td>-0.305175</td>\n",
       "      <td>85367.5693</td>\n",
       "      <td>36.004411</td>\n",
       "      <td>mitochondrial</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1294</td>\n",
       "      <td>sp|P40341|YTA12_YEAST</td>\n",
       "      <td>825</td>\n",
       "      <td>69915.0</td>\n",
       "      <td>7.220154</td>\n",
       "      <td>0.077576</td>\n",
       "      <td>0.286061</td>\n",
       "      <td>0.223030</td>\n",
       "      <td>0.271515</td>\n",
       "      <td>1.009159</td>\n",
       "      <td>-0.570788</td>\n",
       "      <td>93275.1513</td>\n",
       "      <td>40.193115</td>\n",
       "      <td>mitochondrial</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1295</td>\n",
       "      <td>sp|P42844|ZIM17_YEAST</td>\n",
       "      <td>174</td>\n",
       "      <td>11647.5</td>\n",
       "      <td>9.491272</td>\n",
       "      <td>0.057471</td>\n",
       "      <td>0.258621</td>\n",
       "      <td>0.189655</td>\n",
       "      <td>0.212644</td>\n",
       "      <td>1.001064</td>\n",
       "      <td>-0.547126</td>\n",
       "      <td>19854.8131</td>\n",
       "      <td>31.549425</td>\n",
       "      <td>mitochondrial</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1296</td>\n",
       "      <td>sp|Q9VKD7|ZUC_DROME</td>\n",
       "      <td>253</td>\n",
       "      <td>29700.0</td>\n",
       "      <td>9.346497</td>\n",
       "      <td>0.071146</td>\n",
       "      <td>0.343874</td>\n",
       "      <td>0.197628</td>\n",
       "      <td>0.229249</td>\n",
       "      <td>0.992044</td>\n",
       "      <td>0.149012</td>\n",
       "      <td>28519.2049</td>\n",
       "      <td>29.012648</td>\n",
       "      <td>mitochondrial</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>9055 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                         ID  Sequence_Length  Molar_Extinction_Coefficient  \\\n",
       "0      sp|Q4JIM5|ABL2_MOUSE             1182                      119680.0   \n",
       "1      sp|P15891|ABP1_YEAST              592                       60390.0   \n",
       "2     sp|P35609|ACTN2_HUMAN              894                      127052.5   \n",
       "3     sp|P05095|ACTNA_DICDI              861                       89060.0   \n",
       "4      sp|Q9Y2D8|ADIP_HUMAN              614                       66787.5   \n",
       "...                     ...              ...                           ...   \n",
       "1292   sp|P38851|YSP1_YEAST             1228                      177502.5   \n",
       "1293  sp|Q9HGM3|YTA12_SCHPO              773                       42352.5   \n",
       "1294  sp|P40341|YTA12_YEAST              825                       69915.0   \n",
       "1295  sp|P42844|ZIM17_YEAST              174                       11647.5   \n",
       "1296    sp|Q9VKD7|ZUC_DROME              253                       29700.0   \n",
       "\n",
       "      Isoelectric_Point  Aromaticity  Secondary_Structure_Helix  \\\n",
       "0              7.965393     0.067682                   0.241963   \n",
       "1              4.589172     0.060811                   0.197635   \n",
       "2              5.307678     0.078300                   0.282998   \n",
       "3              5.001526     0.073171                   0.298490   \n",
       "4              6.029358     0.061889                   0.250814   \n",
       "...                 ...          ...                        ...   \n",
       "1292           8.361145     0.118078                   0.350163   \n",
       "1293           9.393005     0.078913                   0.296248   \n",
       "1294           7.220154     0.077576                   0.286061   \n",
       "1295           9.491272     0.057471                   0.258621   \n",
       "1296           9.346497     0.071146                   0.343874   \n",
       "\n",
       "      Secondary_Structure_Turn  Secondary_Structure_Sheet  Mean_Flexibility  \\\n",
       "0                     0.296108                   0.252961          1.008064   \n",
       "1                     0.285473                   0.282095          1.025372   \n",
       "2                     0.164430                   0.313199          1.003840   \n",
       "3                     0.188153                   0.295006          1.007373   \n",
       "4                     0.214984                   0.268730          1.012336   \n",
       "...                        ...                        ...               ...   \n",
       "1292                  0.205212                   0.236156          1.000323   \n",
       "1293                  0.253558                   0.245796          1.003549   \n",
       "1294                  0.223030                   0.271515          1.009159   \n",
       "1295                  0.189655                   0.212644          1.001064   \n",
       "1296                  0.197628                   0.229249          0.992044   \n",
       "\n",
       "         Gravy  Molecular_Weight  Instability_Index Subcellular_Location  \n",
       "0    -0.509983       128194.9812          52.288164             cystolic  \n",
       "1    -1.042061        65575.3795          61.758615             cystolic  \n",
       "2    -0.597315       103852.5990          43.180884             cystolic  \n",
       "3    -0.466783        97356.5559          41.535075             cystolic  \n",
       "4    -0.882410        71235.4260          52.350000             cystolic  \n",
       "...        ...               ...                ...                  ...  \n",
       "1292 -0.366775       143581.8566          42.238363        mitochondrial  \n",
       "1293 -0.305175        85367.5693          36.004411        mitochondrial  \n",
       "1294 -0.570788        93275.1513          40.193115        mitochondrial  \n",
       "1295 -0.547126        19854.8131          31.549425        mitochondrial  \n",
       "1296  0.149012        28519.2049          29.012648        mitochondrial  \n",
       "\n",
       "[9055 rows x 13 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>Sequence_Length</th>\n",
       "      <th>Molar_Extinction_Coefficient</th>\n",
       "      <th>Isoelectric_Point</th>\n",
       "      <th>Aromaticity</th>\n",
       "      <th>Secondary_Structure_Helix</th>\n",
       "      <th>Secondary_Structure_Turn</th>\n",
       "      <th>Secondary_Structure_Sheet</th>\n",
       "      <th>Mean_Flexibility</th>\n",
       "      <th>Gravy</th>\n",
       "      <th>Molecular_Weight</th>\n",
       "      <th>Instability_Index</th>\n",
       "      <th>Subcellular_Location</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [ID, Sequence_Length, Molar_Extinction_Coefficient, Isoelectric_Point, Aromaticity, Secondary_Structure_Helix, Secondary_Structure_Turn, Secondary_Structure_Sheet, Mean_Flexibility, Gravy, Molecular_Weight, Instability_Index, Subcellular_Location]\n",
       "Index: []"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_data.loc[df_data['Sequence_Length'] == 13100]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# histogram of sequence lengths\n",
    "ax = df_data['Sequence_Length'].plot.hist(bins=200, alpha=0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n"
     ]
    }
   ],
   "source": [
    "# count number of examples with sequence greater than 2000\n",
    "count = 0\n",
    "for i in range(len(df_data['Sequence_Length'])):\n",
    "    if df_data['Sequence_Length'].iloc[i] > 2000:\n",
    "        print(df_data['Sequence_Length'].iloc[i])\n",
    "        count += 1\n",
    "print(count)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Cyto</th>\n",
       "      <th>Secreted</th>\n",
       "      <th>Nuclear</th>\n",
       "      <th>Mito</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1182.0</td>\n",
       "      <td>1474.0</td>\n",
       "      <td>546</td>\n",
       "      <td>728.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>592.0</td>\n",
       "      <td>132.0</td>\n",
       "      <td>560</td>\n",
       "      <td>678.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>894.0</td>\n",
       "      <td>339.0</td>\n",
       "      <td>382</td>\n",
       "      <td>415.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>861.0</td>\n",
       "      <td>101.0</td>\n",
       "      <td>731</td>\n",
       "      <td>621.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>614.0</td>\n",
       "      <td>188.0</td>\n",
       "      <td>442</td>\n",
       "      <td>647.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3309</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>614</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3310</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>383</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3311</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>337</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3312</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>412</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3313</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>302</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3314 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        Cyto  Secreted  Nuclear   Mito\n",
       "0     1182.0    1474.0      546  728.0\n",
       "1      592.0     132.0      560  678.0\n",
       "2      894.0     339.0      382  415.0\n",
       "3      861.0     101.0      731  621.0\n",
       "4      614.0     188.0      442  647.0\n",
       "...      ...       ...      ...    ...\n",
       "3309     NaN       NaN      614    NaN\n",
       "3310     NaN       NaN      383    NaN\n",
       "3311     NaN       NaN      337    NaN\n",
       "3312     NaN       NaN      412    NaN\n",
       "3313     NaN       NaN      302    NaN\n",
       "\n",
       "[3314 rows x 4 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cut_long_seq = False\n",
    "\n",
    "if cut_long_seq:\n",
    "    for n, i in enumerate(c_list):\n",
    "        if c_list[n] > 2000:\n",
    "            c_list[n] = float('NaN')\n",
    "    for n, i in enumerate(n_list):\n",
    "        if n_list[n] > 2000:\n",
    "            n_list[n] = float('NaN')\n",
    "    for n, i in enumerate(s_list):\n",
    "        if s_list[n] > 2000:\n",
    "            s_list[n] = float('NaN')\n",
    "    for n, i in enumerate(m_list):\n",
    "        if m_list[n] > 2000:\n",
    "            m_list[n] = float('NaN')\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "# pad c_list \n",
    "c_list_new = np.append(c_list, np.repeat(np.nan, 310))\n",
    "dfObj = pd.DataFrame(c_list_new, columns=['Cyto']) \n",
    "# pad s_list \n",
    "s_list_new = np.append(s_list, np.repeat(np.nan, 1709))\n",
    "dfObj['Secreted'] = s_list_new\n",
    "dfObj['Nuclear'] = n_list\n",
    "# pad m_list \n",
    "m_list_new = np.append(m_list, np.repeat(np.nan, 2015))\n",
    "dfObj['Mito'] = m_list_new\n",
    "dfObj"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = sns.violinplot(data=dfObj)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Preprocessing Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess_data(df,categorical_cols,seed):\n",
    "    # Impute missing values with mean of respective feature\n",
    "    print(\"-----Missing Features Before Imputation------\")\n",
    "    print(df.isna().sum())\n",
    "    df.fillna(df.mean(), inplace=True)\n",
    "    print(\"-----Missing Features After Imputation------\")\n",
    "    print(df.isna().sum())\n",
    "        \n",
    "    # split columns into categorical and continuous\n",
    "    continuous_cols = df.columns.difference(categorical_cols)\n",
    "        \n",
    "    # continuous columns need to be normalised\n",
    "    scaler = StandardScaler()\n",
    "    df[continuous_cols] = scaler.fit_transform(df[continuous_cols])\n",
    "        \n",
    "    # encode the categorical columns\n",
    "    for i in range(len(categorical_cols)):\n",
    "        le = preprocessing.LabelEncoder()\n",
    "        df[categorical_cols[i]] = le.fit_transform(df[categorical_cols[i]])\n",
    "        if categorical_cols[i] == 'Subcellular_Location':\n",
    "            le_name_mapping = dict(zip(le.classes_, le.transform(le.classes_)))\n",
    "            print(\"\\n\\n ------ INVERSE MAPPING ------\")\n",
    "            print(le_name_mapping)\n",
    "    # shuffle the data    \n",
    "    df = shuffle(df, random_state=seed)\n",
    "    df.reset_index(inplace=True, drop=True)\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-----Missing Features Before Imputation------\n",
      "ID                              0\n",
      "Sequence_Length                 0\n",
      "Molar_Extinction_Coefficient    0\n",
      "Isoelectric_Point               0\n",
      "Aromaticity                     0\n",
      "Secondary_Structure_Helix       0\n",
      "Secondary_Structure_Turn        0\n",
      "Secondary_Structure_Sheet       0\n",
      "Mean_Flexibility                0\n",
      "Gravy                           0\n",
      "Molecular_Weight                0\n",
      "Instability_Index               0\n",
      "Subcellular_Location            0\n",
      "dtype: int64\n",
      "-----Missing Features After Imputation------\n",
      "ID                              0\n",
      "Sequence_Length                 0\n",
      "Molar_Extinction_Coefficient    0\n",
      "Isoelectric_Point               0\n",
      "Aromaticity                     0\n",
      "Secondary_Structure_Helix       0\n",
      "Secondary_Structure_Turn        0\n",
      "Secondary_Structure_Sheet       0\n",
      "Mean_Flexibility                0\n",
      "Gravy                           0\n",
      "Molecular_Weight                0\n",
      "Instability_Index               0\n",
      "Subcellular_Location            0\n",
      "dtype: int64\n",
      "\n",
      "\n",
      " ------ INVERSE MAPPING ------\n",
      "{'cystolic': 0, 'mitochondrial': 1, 'nuclear': 2, 'secreted': 3}\n"
     ]
    }
   ],
   "source": [
    "categorical_cols = ['ID','Subcellular_Location']\n",
    "df_data = preprocess_data(df_data,categorical_cols,seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>Sequence_Length</th>\n",
       "      <th>Molar_Extinction_Coefficient</th>\n",
       "      <th>Isoelectric_Point</th>\n",
       "      <th>Aromaticity</th>\n",
       "      <th>Secondary_Structure_Helix</th>\n",
       "      <th>Secondary_Structure_Turn</th>\n",
       "      <th>Secondary_Structure_Sheet</th>\n",
       "      <th>Mean_Flexibility</th>\n",
       "      <th>Gravy</th>\n",
       "      <th>Molecular_Weight</th>\n",
       "      <th>Instability_Index</th>\n",
       "      <th>Subcellular_Location</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>5831</td>\n",
       "      <td>-0.467064</td>\n",
       "      <td>-0.412079</td>\n",
       "      <td>-1.087579</td>\n",
       "      <td>-0.346294</td>\n",
       "      <td>-0.618559</td>\n",
       "      <td>-0.386153</td>\n",
       "      <td>1.662142</td>\n",
       "      <td>0.031324</td>\n",
       "      <td>-0.108907</td>\n",
       "      <td>-0.482573</td>\n",
       "      <td>0.144234</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>129</td>\n",
       "      <td>-0.277188</td>\n",
       "      <td>-0.237655</td>\n",
       "      <td>-1.274681</td>\n",
       "      <td>-0.495080</td>\n",
       "      <td>-0.261606</td>\n",
       "      <td>-0.428909</td>\n",
       "      <td>0.052451</td>\n",
       "      <td>0.258900</td>\n",
       "      <td>-0.156911</td>\n",
       "      <td>-0.286089</td>\n",
       "      <td>-0.178052</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>591</td>\n",
       "      <td>1.244497</td>\n",
       "      <td>1.653182</td>\n",
       "      <td>-0.856997</td>\n",
       "      <td>1.136827</td>\n",
       "      <td>1.388790</td>\n",
       "      <td>-0.676516</td>\n",
       "      <td>0.156709</td>\n",
       "      <td>-0.434898</td>\n",
       "      <td>0.337770</td>\n",
       "      <td>1.337163</td>\n",
       "      <td>-0.341827</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>7104</td>\n",
       "      <td>1.953191</td>\n",
       "      <td>0.865456</td>\n",
       "      <td>-0.879436</td>\n",
       "      <td>-0.526125</td>\n",
       "      <td>-0.201071</td>\n",
       "      <td>-0.222576</td>\n",
       "      <td>0.857916</td>\n",
       "      <td>0.485277</td>\n",
       "      <td>-0.433857</td>\n",
       "      <td>1.991264</td>\n",
       "      <td>0.694963</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>956</td>\n",
       "      <td>-0.547294</td>\n",
       "      <td>-0.546911</td>\n",
       "      <td>1.133478</td>\n",
       "      <td>0.451617</td>\n",
       "      <td>-0.185935</td>\n",
       "      <td>-0.106073</td>\n",
       "      <td>-0.486471</td>\n",
       "      <td>1.090367</td>\n",
       "      <td>-1.090943</td>\n",
       "      <td>-0.518221</td>\n",
       "      <td>0.303567</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9050</td>\n",
       "      <td>3372</td>\n",
       "      <td>-1.242615</td>\n",
       "      <td>-1.169039</td>\n",
       "      <td>-1.631107</td>\n",
       "      <td>-1.750224</td>\n",
       "      <td>-3.088652</td>\n",
       "      <td>2.430601</td>\n",
       "      <td>-2.304548</td>\n",
       "      <td>1.023729</td>\n",
       "      <td>-1.004263</td>\n",
       "      <td>-1.249362</td>\n",
       "      <td>0.237287</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9051</td>\n",
       "      <td>7943</td>\n",
       "      <td>-0.889606</td>\n",
       "      <td>-0.789793</td>\n",
       "      <td>1.502489</td>\n",
       "      <td>0.562841</td>\n",
       "      <td>1.265248</td>\n",
       "      <td>-1.465314</td>\n",
       "      <td>0.254557</td>\n",
       "      <td>-0.937986</td>\n",
       "      <td>1.013895</td>\n",
       "      <td>-0.873977</td>\n",
       "      <td>-0.431532</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9052</td>\n",
       "      <td>8870</td>\n",
       "      <td>0.431506</td>\n",
       "      <td>1.257143</td>\n",
       "      <td>-1.004348</td>\n",
       "      <td>1.012324</td>\n",
       "      <td>1.280249</td>\n",
       "      <td>-0.480963</td>\n",
       "      <td>-0.087632</td>\n",
       "      <td>-0.731304</td>\n",
       "      <td>0.775369</td>\n",
       "      <td>0.460851</td>\n",
       "      <td>-0.162756</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9053</td>\n",
       "      <td>6051</td>\n",
       "      <td>-0.068591</td>\n",
       "      <td>-0.016959</td>\n",
       "      <td>0.890578</td>\n",
       "      <td>0.226223</td>\n",
       "      <td>2.395100</td>\n",
       "      <td>-0.109899</td>\n",
       "      <td>0.540609</td>\n",
       "      <td>-2.291556</td>\n",
       "      <td>2.966908</td>\n",
       "      <td>-0.102127</td>\n",
       "      <td>0.149125</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9054</td>\n",
       "      <td>1636</td>\n",
       "      <td>0.169423</td>\n",
       "      <td>-0.191383</td>\n",
       "      <td>-0.644659</td>\n",
       "      <td>-0.701527</td>\n",
       "      <td>-0.591606</td>\n",
       "      <td>0.375022</td>\n",
       "      <td>0.481627</td>\n",
       "      <td>0.769724</td>\n",
       "      <td>-0.559902</td>\n",
       "      <td>0.158073</td>\n",
       "      <td>0.500552</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>9055 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        ID  Sequence_Length  Molar_Extinction_Coefficient  Isoelectric_Point  \\\n",
       "0     5831        -0.467064                     -0.412079          -1.087579   \n",
       "1      129        -0.277188                     -0.237655          -1.274681   \n",
       "2      591         1.244497                      1.653182          -0.856997   \n",
       "3     7104         1.953191                      0.865456          -0.879436   \n",
       "4      956        -0.547294                     -0.546911           1.133478   \n",
       "...    ...              ...                           ...                ...   \n",
       "9050  3372        -1.242615                     -1.169039          -1.631107   \n",
       "9051  7943        -0.889606                     -0.789793           1.502489   \n",
       "9052  8870         0.431506                      1.257143          -1.004348   \n",
       "9053  6051        -0.068591                     -0.016959           0.890578   \n",
       "9054  1636         0.169423                     -0.191383          -0.644659   \n",
       "\n",
       "      Aromaticity  Secondary_Structure_Helix  Secondary_Structure_Turn  \\\n",
       "0       -0.346294                  -0.618559                 -0.386153   \n",
       "1       -0.495080                  -0.261606                 -0.428909   \n",
       "2        1.136827                   1.388790                 -0.676516   \n",
       "3       -0.526125                  -0.201071                 -0.222576   \n",
       "4        0.451617                  -0.185935                 -0.106073   \n",
       "...           ...                        ...                       ...   \n",
       "9050    -1.750224                  -3.088652                  2.430601   \n",
       "9051     0.562841                   1.265248                 -1.465314   \n",
       "9052     1.012324                   1.280249                 -0.480963   \n",
       "9053     0.226223                   2.395100                 -0.109899   \n",
       "9054    -0.701527                  -0.591606                  0.375022   \n",
       "\n",
       "      Secondary_Structure_Sheet  Mean_Flexibility     Gravy  Molecular_Weight  \\\n",
       "0                      1.662142          0.031324 -0.108907         -0.482573   \n",
       "1                      0.052451          0.258900 -0.156911         -0.286089   \n",
       "2                      0.156709         -0.434898  0.337770          1.337163   \n",
       "3                      0.857916          0.485277 -0.433857          1.991264   \n",
       "4                     -0.486471          1.090367 -1.090943         -0.518221   \n",
       "...                         ...               ...       ...               ...   \n",
       "9050                  -2.304548          1.023729 -1.004263         -1.249362   \n",
       "9051                   0.254557         -0.937986  1.013895         -0.873977   \n",
       "9052                  -0.087632         -0.731304  0.775369          0.460851   \n",
       "9053                   0.540609         -2.291556  2.966908         -0.102127   \n",
       "9054                   0.481627          0.769724 -0.559902          0.158073   \n",
       "\n",
       "      Instability_Index  Subcellular_Location  \n",
       "0              0.144234                     2  \n",
       "1             -0.178052                     2  \n",
       "2             -0.341827                     0  \n",
       "3              0.694963                     0  \n",
       "4              0.303567                     2  \n",
       "...                 ...                   ...  \n",
       "9050           0.237287                     3  \n",
       "9051          -0.431532                     3  \n",
       "9052          -0.162756                     0  \n",
       "9053           0.149125                     0  \n",
       "9054           0.500552                     2  \n",
       "\n",
       "[9055 rows x 13 columns]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "corrMatrix = df_data.corr()\n",
    "plt.figure(figsize=(10,7))\n",
    "\n",
    "sns.heatmap(abs(corrMatrix), annot=True, cmap='Blues')\n",
    "\n",
    "# fix for mpl bug that cuts off top/bottom of seaborn viz\n",
    "b, t = plt.ylim() # discover the values for bottom and top\n",
    "b += 0.5 # Add 0.5 to the bottom\n",
    "t -= 0.5 # Subtract 0.5 from the top\n",
    "plt.ylim(b, t) # update the ylim(bottom, top) values\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Boxplot of Features; ignoring AA percentages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def boxplot_features(df_data,outliers):\n",
    "        # variable names for indexing\n",
    "        id_vars=[#'Contains_X',\n",
    "                 'Sequence_Length', \n",
    "                   'Molar_Extinction_Coefficient',\n",
    "                   'Isoelectric_Point', \n",
    "                   'Aromaticity',\n",
    "                   'Secondary_Structure_Helix',\n",
    "                   'Secondary_Structure_Turn',\n",
    "                   'Secondary_Structure_Sheet',\n",
    "                   'Mean_Flexibility',\n",
    "                   'Gravy',\n",
    "                   'Molecular_Weight',\n",
    "                   'Instability_Index']\n",
    "\n",
    "        # names to appear on plot\n",
    "        short_names=[#'Contains X',\n",
    "                     'Sequence \\n Length',\n",
    "                     'Molar \\n Extinction \\n Coefficient',\n",
    "                     'Isoelectric \\n Point', \n",
    "                     'Aromaticity',\n",
    "                     'Secondary \\n Structure \\n Helix',\n",
    "                     'Secondary \\n Structure \\n Turn',\n",
    "                     'Secondary \\n Structure \\n Sheet',\n",
    "                     'Mean \\n Flexibility',\n",
    "                     'Gravy',\n",
    "                     'Molecular \\n Weight',\n",
    "                     'Instability \\n Index']\n",
    "        \n",
    "        # arrange data in long dataframe for plotting\n",
    "        # https://stackoverflow.com/questions/44552489/plotting-multiple-boxplots-in-seaborn\n",
    "        df_list=[]\n",
    "        for i in range(len(id_vars)):\n",
    "                df_temp = df_data.filter(['Subcellular_Location',id_vars[i]], axis=1)\n",
    "                experiment = i* np.ones((df_data.shape[0],1),dtype=int)\n",
    "                df_temp['Feature'] = experiment\n",
    "                df_temp = df_temp.rename(columns={id_vars[i]: \"Standardised Value\"})\n",
    "                df_list.append(df_temp)\n",
    "        cdf = pd.concat(df_list) \n",
    "        cdf = cdf.rename(columns={\"Subcellular_Location\": \"Subcellular Location\"})\n",
    "        \n",
    "        # rename features\n",
    "        for i in range(len(id_vars)):\n",
    "                cdf['Feature'][i*df_data.shape[0]:(i+1)*df_data.shape[0]] = short_names[i]\n",
    "        \n",
    "        # rename classes\n",
    "        cdf['Subcellular Location'] = cdf['Subcellular Location'].replace(to_replace = 0, value = 'Cystolic') \n",
    "        cdf['Subcellular Location'] = cdf['Subcellular Location'].replace(to_replace = 1, value = 'Secreted') \n",
    "        cdf['Subcellular Location'] = cdf['Subcellular Location'].replace(to_replace = 2, value = 'Nucelear') \n",
    "        cdf['Subcellular Location'] = cdf['Subcellular Location'].replace(to_replace = 3, value = 'Mitochondrial') \n",
    "        \n",
    "                \n",
    "        # make boxplot \n",
    "        plt.figure(figsize=(30, 15))\n",
    "        sns.set_style(\"darkgrid\")       \n",
    "        ax = sns.boxplot(x=\"Feature\", y=\"Standardised Value\", hue=\"Subcellular Location\", notch=True, data=cdf, palette='pastel',showfliers = outliers) \n",
    "        handles, _ = ax.get_legend_handles_labels()\n",
    "        ax.legend(handles, [\"Cyto\", \"Secreted\",\"Nuclear\",\"Mito\"])\n",
    "        ax.set_facecolor('w')\n",
    "        plt.savefig('boxplot.png')\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2160x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "boxplot_features(df_data,outliers=False) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Correlation Plot; ignoring AA percentages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "def correlation_plot(X,y):  \n",
    "    pears_r = []\n",
    "    X_plot = X.drop(['ID'],axis=1)\n",
    "    for column in X_plot:\n",
    "        inData = scipy.stats.pearsonr(X_plot[column], y)\n",
    "        pears_r.append(inData[0])\n",
    "        \n",
    "    pears_df = pd.Series(pears_r, index = X_plot.columns)\n",
    "    my_colors = []\n",
    "    mx = max(pears_r) \n",
    "    mn = min(pears_r)\n",
    "    plot_cvalues = []\n",
    "    for x in range(len(pears_r)):\n",
    "        plot_cvalues.append((pears_r[x] - mn)/(mx-mn))\n",
    "    for x in range(len(pears_df)):\n",
    "        my_colors.append((plot_cvalues[x], 0.1, 0.5+ 0.5*plot_cvalues[x]))\n",
    "    print(my_colors)\n",
    "    fig, ax = plt.subplots()\n",
    "    pears_df.plot.bar(figsize=(20,20),yticks=np.arange(-0.35,0.2,0.1),color=my_colors,alpha=0.8)\n",
    "    ax.set_xticklabels([#'Contains X',\n",
    "                        'Sequence \\n Length',\n",
    "                        'Molar \\n Extinction \\n Coefficient',\n",
    "                         #'Global Amino Acid \\n Composition',\n",
    "                         #'Local Amino Acid \\n Composition (Start)',\n",
    "                         #'Local Amino Acid \\n Composition (End)',\n",
    "                         'Isoelectric \\n Point', \n",
    "                         'Aromaticity',\n",
    "                         'Secondary Structure \\n Helix',\n",
    "                         'Secondary Structure \\n Turn',\n",
    "                         'Secondary Structure \\n Sheet',\n",
    "                         'Mean \\n Flexibility',\n",
    "                         'Gravy',\n",
    "                         'Molecular \\n Weight',\n",
    "                         'Instability \\n Index'])\n",
    "    ax.set_facecolor('w')\n",
    "    plt.xlabel('Feature',fontsize = 16)\n",
    "    plt.ylabel('r value',fontsize = 16)\n",
    "    plt.title('Pearson Correlation of Feature with Subcellular Location',fontsize = 20)\n",
    "    fig.subplots_adjust(bottom=0.3) \n",
    "    plt.figure(figsize=(20, 2))\n",
    "    plt.savefig('correlations.png')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[(0.013885581915272904, 0.1, 0.5069427909576365), (0.18482347426236884, 0.1, 0.5924117371311844), (0.8579572771846355, 0.1, 0.9289786385923178), (0.8055809550377807, 0.1, 0.9027904775188904), (0.30502889513142883, 0.1, 0.6525144475657144), (1.0, 0.1, 1.0), (0.060691087336030965, 0.1, 0.5303455436680155), (0.45600125063562125, 0.1, 0.7280006253178106), (0.7085772450069762, 0.1, 0.854288622503488), (0.0, 0.1, 0.5), (0.49791508728284195, 0.1, 0.748957543641421)]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 1440x144 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "correlation_plot(df_data.drop(['Subcellular_Location'], axis=1),df_data['Subcellular_Location'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Generate Amino Acid Plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "---------- cyto2.txt ----------\n",
      "Number of records = 2924\n",
      "Max string length = 7393\n",
      "Min string length = 19\n",
      "Mean string length = 664.6654460719042\n",
      "---------- secreted2.txt ----------\n",
      "Number of records = 1586\n",
      "Max string length = 13100\n",
      "Min string length = 11\n",
      "Mean string length = 304.7102803738318\n",
      "---------- nuclear2.txt ----------\n",
      "Number of records = 3248\n",
      "Max string length = 5596\n",
      "Min string length = 35\n",
      "Mean string length = 624.4643934821968\n",
      "---------- mito2.txt ----------\n",
      "Number of records = 1297\n",
      "Max string length = 2628\n",
      "Min string length = 19\n",
      "Mean string length = 375.56966897613546\n"
     ]
    }
   ],
   "source": [
    "# load data with long seq cut and all AA features included\n",
    "df_c, c_list = load_data('cyto2.txt','cystolic',True,True)\n",
    "df_s, s_list = load_data('secreted2.txt','secreted',True,True)\n",
    "df_n, n_list = load_data('nuclear2.txt','nuclear',True,True)\n",
    "df_m, m_list = load_data('mito2.txt','mitochondrial',True,True)\n",
    "df_aa_plot = pd.concat([df_c,df_s,df_n,df_m])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>Sequence_Length</th>\n",
       "      <th>Molar_Extinction_Coefficient</th>\n",
       "      <th>Isoelectric_Point</th>\n",
       "      <th>Aromaticity</th>\n",
       "      <th>Secondary_Structure_Helix</th>\n",
       "      <th>Secondary_Structure_Turn</th>\n",
       "      <th>Secondary_Structure_Sheet</th>\n",
       "      <th>Mean_Flexibility</th>\n",
       "      <th>Gravy</th>\n",
       "      <th>...</th>\n",
       "      <th>Ne</th>\n",
       "      <th>Pe</th>\n",
       "      <th>Qe</th>\n",
       "      <th>Re</th>\n",
       "      <th>Se</th>\n",
       "      <th>Te</th>\n",
       "      <th>Ve</th>\n",
       "      <th>We</th>\n",
       "      <th>Ye</th>\n",
       "      <th>Subcellular_Location</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>sp|Q4JIM5|ABL2_MOUSE</td>\n",
       "      <td>1182</td>\n",
       "      <td>119680.0</td>\n",
       "      <td>7.965393</td>\n",
       "      <td>0.067682</td>\n",
       "      <td>0.241963</td>\n",
       "      <td>0.296108</td>\n",
       "      <td>0.252961</td>\n",
       "      <td>1.008064</td>\n",
       "      <td>-0.509983</td>\n",
       "      <td>...</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>7</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>cystolic</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>sp|P15891|ABP1_YEAST</td>\n",
       "      <td>592</td>\n",
       "      <td>60390.0</td>\n",
       "      <td>4.589172</td>\n",
       "      <td>0.060811</td>\n",
       "      <td>0.197635</td>\n",
       "      <td>0.285473</td>\n",
       "      <td>0.282095</td>\n",
       "      <td>1.025372</td>\n",
       "      <td>-1.042061</td>\n",
       "      <td>...</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>cystolic</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>sp|P35609|ACTN2_HUMAN</td>\n",
       "      <td>894</td>\n",
       "      <td>127052.5</td>\n",
       "      <td>5.307678</td>\n",
       "      <td>0.078300</td>\n",
       "      <td>0.282998</td>\n",
       "      <td>0.164430</td>\n",
       "      <td>0.313199</td>\n",
       "      <td>1.003840</td>\n",
       "      <td>-0.597315</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>cystolic</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>sp|P05095|ACTNA_DICDI</td>\n",
       "      <td>861</td>\n",
       "      <td>89060.0</td>\n",
       "      <td>5.001526</td>\n",
       "      <td>0.073171</td>\n",
       "      <td>0.298490</td>\n",
       "      <td>0.188153</td>\n",
       "      <td>0.295006</td>\n",
       "      <td>1.007373</td>\n",
       "      <td>-0.466783</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>cystolic</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>sp|Q9Y2D8|ADIP_HUMAN</td>\n",
       "      <td>614</td>\n",
       "      <td>66787.5</td>\n",
       "      <td>6.029358</td>\n",
       "      <td>0.061889</td>\n",
       "      <td>0.250814</td>\n",
       "      <td>0.214984</td>\n",
       "      <td>0.268730</td>\n",
       "      <td>1.012336</td>\n",
       "      <td>-0.882410</td>\n",
       "      <td>...</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>cystolic</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1292</td>\n",
       "      <td>sp|P38851|YSP1_YEAST</td>\n",
       "      <td>1228</td>\n",
       "      <td>177502.5</td>\n",
       "      <td>8.361145</td>\n",
       "      <td>0.118078</td>\n",
       "      <td>0.350163</td>\n",
       "      <td>0.205212</td>\n",
       "      <td>0.236156</td>\n",
       "      <td>1.000323</td>\n",
       "      <td>-0.366775</td>\n",
       "      <td>...</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>mitochondrial</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1293</td>\n",
       "      <td>sp|Q9HGM3|YTA12_SCHPO</td>\n",
       "      <td>773</td>\n",
       "      <td>42352.5</td>\n",
       "      <td>9.393005</td>\n",
       "      <td>0.078913</td>\n",
       "      <td>0.296248</td>\n",
       "      <td>0.253558</td>\n",
       "      <td>0.245796</td>\n",
       "      <td>1.003549</td>\n",
       "      <td>-0.305175</td>\n",
       "      <td>...</td>\n",
       "      <td>4</td>\n",
       "      <td>9</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>mitochondrial</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1294</td>\n",
       "      <td>sp|P40341|YTA12_YEAST</td>\n",
       "      <td>825</td>\n",
       "      <td>69915.0</td>\n",
       "      <td>7.220154</td>\n",
       "      <td>0.077576</td>\n",
       "      <td>0.286061</td>\n",
       "      <td>0.223030</td>\n",
       "      <td>0.271515</td>\n",
       "      <td>1.009159</td>\n",
       "      <td>-0.570788</td>\n",
       "      <td>...</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>mitochondrial</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1295</td>\n",
       "      <td>sp|P42844|ZIM17_YEAST</td>\n",
       "      <td>174</td>\n",
       "      <td>11647.5</td>\n",
       "      <td>9.491272</td>\n",
       "      <td>0.057471</td>\n",
       "      <td>0.258621</td>\n",
       "      <td>0.189655</td>\n",
       "      <td>0.212644</td>\n",
       "      <td>1.001064</td>\n",
       "      <td>-0.547126</td>\n",
       "      <td>...</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>mitochondrial</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1296</td>\n",
       "      <td>sp|Q9VKD7|ZUC_DROME</td>\n",
       "      <td>253</td>\n",
       "      <td>29700.0</td>\n",
       "      <td>9.346497</td>\n",
       "      <td>0.071146</td>\n",
       "      <td>0.343874</td>\n",
       "      <td>0.197628</td>\n",
       "      <td>0.229249</td>\n",
       "      <td>0.992044</td>\n",
       "      <td>0.149012</td>\n",
       "      <td>...</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>mitochondrial</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>9055 rows × 73 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                         ID  Sequence_Length  Molar_Extinction_Coefficient  \\\n",
       "0      sp|Q4JIM5|ABL2_MOUSE             1182                      119680.0   \n",
       "1      sp|P15891|ABP1_YEAST              592                       60390.0   \n",
       "2     sp|P35609|ACTN2_HUMAN              894                      127052.5   \n",
       "3     sp|P05095|ACTNA_DICDI              861                       89060.0   \n",
       "4      sp|Q9Y2D8|ADIP_HUMAN              614                       66787.5   \n",
       "...                     ...              ...                           ...   \n",
       "1292   sp|P38851|YSP1_YEAST             1228                      177502.5   \n",
       "1293  sp|Q9HGM3|YTA12_SCHPO              773                       42352.5   \n",
       "1294  sp|P40341|YTA12_YEAST              825                       69915.0   \n",
       "1295  sp|P42844|ZIM17_YEAST              174                       11647.5   \n",
       "1296    sp|Q9VKD7|ZUC_DROME              253                       29700.0   \n",
       "\n",
       "      Isoelectric_Point  Aromaticity  Secondary_Structure_Helix  \\\n",
       "0              7.965393     0.067682                   0.241963   \n",
       "1              4.589172     0.060811                   0.197635   \n",
       "2              5.307678     0.078300                   0.282998   \n",
       "3              5.001526     0.073171                   0.298490   \n",
       "4              6.029358     0.061889                   0.250814   \n",
       "...                 ...          ...                        ...   \n",
       "1292           8.361145     0.118078                   0.350163   \n",
       "1293           9.393005     0.078913                   0.296248   \n",
       "1294           7.220154     0.077576                   0.286061   \n",
       "1295           9.491272     0.057471                   0.258621   \n",
       "1296           9.346497     0.071146                   0.343874   \n",
       "\n",
       "      Secondary_Structure_Turn  Secondary_Structure_Sheet  Mean_Flexibility  \\\n",
       "0                     0.296108                   0.252961          1.008064   \n",
       "1                     0.285473                   0.282095          1.025372   \n",
       "2                     0.164430                   0.313199          1.003840   \n",
       "3                     0.188153                   0.295006          1.007373   \n",
       "4                     0.214984                   0.268730          1.012336   \n",
       "...                        ...                        ...               ...   \n",
       "1292                  0.205212                   0.236156          1.000323   \n",
       "1293                  0.253558                   0.245796          1.003549   \n",
       "1294                  0.223030                   0.271515          1.009159   \n",
       "1295                  0.189655                   0.212644          1.001064   \n",
       "1296                  0.197628                   0.229249          0.992044   \n",
       "\n",
       "         Gravy  ...  Ne  Pe Qe Re Se Te Ve We Ye Subcellular_Location  \n",
       "0    -0.509983  ...   3   2  4  2  6  2  7  0  0             cystolic  \n",
       "1    -1.042061  ...   5   1  0  0  3  1  3  2  2             cystolic  \n",
       "2    -0.597315  ...   0   5  2  3  5  0  1  0  4             cystolic  \n",
       "3    -0.466783  ...   1   1  3  1  4  1  2  0  3             cystolic  \n",
       "4    -0.882410  ...   3   3  3  1  7  3  3  1  2             cystolic  \n",
       "...        ...  ...  ..  .. .. .. .. .. .. .. ..                  ...  \n",
       "1292 -0.366775  ...   3   1  2  3  2  4  1  1  3        mitochondrial  \n",
       "1293 -0.305175  ...   4   9  3  3  4  2  1  0  4        mitochondrial  \n",
       "1294 -0.570788  ...   5   4  0  5  1  2  0  0  2        mitochondrial  \n",
       "1295 -0.547126  ...   4   3  5  0  5  0  3  0  1        mitochondrial  \n",
       "1296  0.149012  ...   3   0  4  2  2  5  1  3  0        mitochondrial  \n",
       "\n",
       "[9055 rows x 73 columns]"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_aa_plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_preprocess_data(df,categorical_cols,seed):\n",
    "    # Impute missing values with mean of respective feature\n",
    "    df.fillna(df.mean(), inplace=True)\n",
    "    \n",
    "    # split columns into categorical and continuous\n",
    "    continuous_cols = df.columns.difference(categorical_cols)\n",
    "               \n",
    "    # encode the categorical columns\n",
    "    for i in range(len(categorical_cols)):\n",
    "        le = preprocessing.LabelEncoder()\n",
    "        df[categorical_cols[i]] = le.fit_transform(df[categorical_cols[i]])\n",
    "        if categorical_cols[i] == 'Subcellular_Location':\n",
    "            le_name_mapping = dict(zip(le.classes_, le.transform(le.classes_)))\n",
    "            #print(\"\\n\\n ------ INVERSE MAPPING ------\")\n",
    "            #print(le_name_mapping)\n",
    "    # shuffle the data    \n",
    "    #df = shuffle(df, random_state=seed)\n",
    "    df.reset_index(inplace=True, drop=True)\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "def aa_plot(df,subject,region):\n",
    "    if region == 'end':\n",
    "        q=df[['Ae', 'Ce', 'De', 'Ee', 'Fe','Ge', 'He', 'Ie', 'Ke', 'Le','Me', 'Ne', 'Pe', 'Qe', 'Re','Se', 'Te', 'Ve', 'We', 'Ye','Subcellular_Location']].copy()\n",
    "    if region == 'start':\n",
    "        q=df[['As', 'Cs', 'Ds', 'Es', 'Fs','Gs', 'Hs', 'Is', 'Ks', 'Ls','Ms', 'Ns', 'Ps', 'Qs', 'Rs','Ss', 'Ts', 'Vs', 'Ws', 'Ys','Subcellular_Location']].copy()\n",
    "    if region == 'global':\n",
    "        q=df[['Ag', 'Cg', 'Dg', 'Eg', 'Fg','Gg', 'Hg', 'Ig', 'Kg', 'Lg','Mg', 'Ng', 'Pg', 'Qg', 'Rg','Sg', 'Tg', 'Vg', 'Wg', 'Yg','Subcellular_Location']].copy()\n",
    "       \n",
    "    \n",
    "    \n",
    "    q=plot_preprocess_data(q,['Subcellular_Location'],1)\n",
    "    \n",
    "\n",
    "    \n",
    "    r=q.groupby('Subcellular_Location').mean()\n",
    "    s=r.transpose()\n",
    "    \n",
    "    # what to plot\n",
    "    if subject == 'count':\n",
    "        s.plot.bar(width=0.8, figsize=(20,10),align='edge')\n",
    "        l = plt.legend(('Cyto','Mito','Nuclear','Secreted'), loc='best')\n",
    "        plt.xlabel('Amino Acid')\n",
    "        plt.ylabel('Count across Locations')\n",
    "    if subject == 'std':\n",
    "        plt.figure(figsize=(10,7))\n",
    "        t=s.std(axis=1)\n",
    "        t.plot.bar()\n",
    "        plt.xlabel('Amino Acid')\n",
    "        plt.ylabel('Standard Deviation across Locations')\n",
    "        \n",
    "        \n",
    "        \n",
    "  #  corrMatrix = q.corr()\n",
    "  #  plt.figure(figsize=(20,20))\n",
    "\n",
    "#    sns.heatmap(abs(corrMatrix), annot=True, cmap='Blues')\n",
    "\n",
    "    # fix for mpl bug that cuts off top/bottom of seaborn viz\n",
    "#    b, t = plt.ylim() # discover the values for bottom and top\n",
    "#    b += 0.5 # Add 0.5 to the bottom\n",
    "#    t -= 0.5 # Subtract 0.5 from the top\n",
    "#    plt.ylim(b, t) # update the ylim(bottom, top) values\n",
    "\n",
    "#    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "aa_plot(df_aa_plot,'std','start')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Reload and Preprocess Data Including AA Percentages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "---------- cyto2.txt ----------\n",
      "Number of records = 2924\n",
      "Max string length = 7393\n",
      "Min string length = 19\n",
      "Mean string length = 664.6654460719042\n",
      "---------- secreted2.txt ----------\n",
      "Number of records = 1586\n",
      "Max string length = 13100\n",
      "Min string length = 11\n",
      "Mean string length = 304.7102803738318\n",
      "---------- nuclear2.txt ----------\n",
      "Number of records = 3248\n",
      "Max string length = 5596\n",
      "Min string length = 35\n",
      "Mean string length = 624.4643934821968\n",
      "---------- mito2.txt ----------\n",
      "Number of records = 1297\n",
      "Max string length = 2628\n",
      "Min string length = 19\n",
      "Mean string length = 375.56966897613546\n",
      "Number cystolic records = 2924\n",
      "Number of secreted records = 1586\n",
      "Number of nuclear records = 3248\n",
      "Number of mitochondrial records = 1297\n",
      "Total number of records = 9055\n",
      "-----Missing Features Before Imputation------\n",
      "ID                              0\n",
      "Sequence_Length                 0\n",
      "Molar_Extinction_Coefficient    0\n",
      "Isoelectric_Point               0\n",
      "Aromaticity                     0\n",
      "                               ..\n",
      "Te                              0\n",
      "Ve                              0\n",
      "We                              0\n",
      "Ye                              0\n",
      "Subcellular_Location            0\n",
      "Length: 73, dtype: int64\n",
      "-----Missing Features After Imputation------\n",
      "ID                              0\n",
      "Sequence_Length                 0\n",
      "Molar_Extinction_Coefficient    0\n",
      "Isoelectric_Point               0\n",
      "Aromaticity                     0\n",
      "                               ..\n",
      "Te                              0\n",
      "Ve                              0\n",
      "We                              0\n",
      "Ye                              0\n",
      "Subcellular_Location            0\n",
      "Length: 73, dtype: int64\n",
      "\n",
      "\n",
      " ------ INVERSE MAPPING ------\n",
      "{'cystolic': 0, 'mitochondrial': 1, 'nuclear': 2, 'secreted': 3}\n"
     ]
    }
   ],
   "source": [
    "### Set random seed ###\n",
    "seed = 2020\n",
    "\n",
    "### Load data from each subcellular location ###\n",
    "\n",
    "cut_long_seq = True\n",
    "inc_aa_count_feature = True\n",
    "\n",
    "df_cystolic, _ = load_data('cyto2.txt','cystolic',cut_long_seq,inc_aa_count_feature)\n",
    "df_secreted, _ = load_data('secreted2.txt','secreted',cut_long_seq,inc_aa_count_feature)\n",
    "df_nuclear, _ = load_data('nuclear2.txt','nuclear',cut_long_seq,inc_aa_count_feature)\n",
    "df_mitochondrial, _ = load_data('mito2.txt','mitochondrial',cut_long_seq,inc_aa_count_feature)\n",
    "\n",
    "### Concatenate data and shuffle ###\n",
    "df_data = pd.concat([df_cystolic,df_secreted,df_nuclear,df_mitochondrial])\n",
    "num_records = len(df_data)\n",
    "print(\"Number cystolic records = {}\".format(len(df_cystolic)))\n",
    "print(\"Number of secreted records = {}\".format(len(df_secreted)))\n",
    "print(\"Number of nuclear records = {}\".format(len(df_nuclear)))\n",
    "print(\"Number of mitochondrial records = {}\".format(len(df_mitochondrial)))\n",
    "print(\"Total number of records = {}\".format(num_records))\n",
    "\n",
    "\n",
    "categorical_cols = ['ID','Subcellular_Location']\n",
    "df_data = preprocess_data(df_data,categorical_cols,seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>Sequence_Length</th>\n",
       "      <th>Molar_Extinction_Coefficient</th>\n",
       "      <th>Isoelectric_Point</th>\n",
       "      <th>Aromaticity</th>\n",
       "      <th>Secondary_Structure_Helix</th>\n",
       "      <th>Secondary_Structure_Turn</th>\n",
       "      <th>Secondary_Structure_Sheet</th>\n",
       "      <th>Mean_Flexibility</th>\n",
       "      <th>Gravy</th>\n",
       "      <th>...</th>\n",
       "      <th>Ne</th>\n",
       "      <th>Pe</th>\n",
       "      <th>Qe</th>\n",
       "      <th>Re</th>\n",
       "      <th>Se</th>\n",
       "      <th>Te</th>\n",
       "      <th>Ve</th>\n",
       "      <th>We</th>\n",
       "      <th>Ye</th>\n",
       "      <th>Subcellular_Location</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>5831</td>\n",
       "      <td>-0.467064</td>\n",
       "      <td>-0.412122</td>\n",
       "      <td>-1.087001</td>\n",
       "      <td>-0.345612</td>\n",
       "      <td>-0.618647</td>\n",
       "      <td>-0.387055</td>\n",
       "      <td>1.659587</td>\n",
       "      <td>0.031473</td>\n",
       "      <td>-0.108118</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.682757</td>\n",
       "      <td>-0.708496</td>\n",
       "      <td>-0.093184</td>\n",
       "      <td>1.376535</td>\n",
       "      <td>-0.399516</td>\n",
       "      <td>-1.410233</td>\n",
       "      <td>0.680395</td>\n",
       "      <td>0.478102</td>\n",
       "      <td>-0.284615</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>129</td>\n",
       "      <td>-0.277188</td>\n",
       "      <td>-0.237691</td>\n",
       "      <td>-1.274085</td>\n",
       "      <td>-0.494124</td>\n",
       "      <td>-0.261659</td>\n",
       "      <td>-0.429878</td>\n",
       "      <td>0.051924</td>\n",
       "      <td>0.258738</td>\n",
       "      <td>-0.156050</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.682757</td>\n",
       "      <td>-1.147969</td>\n",
       "      <td>-1.114098</td>\n",
       "      <td>0.071213</td>\n",
       "      <td>0.328668</td>\n",
       "      <td>-0.290431</td>\n",
       "      <td>-0.946194</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>591</td>\n",
       "      <td>1.244497</td>\n",
       "      <td>1.653226</td>\n",
       "      <td>-0.856441</td>\n",
       "      <td>1.134772</td>\n",
       "      <td>1.388896</td>\n",
       "      <td>-0.677865</td>\n",
       "      <td>0.156050</td>\n",
       "      <td>-0.434112</td>\n",
       "      <td>0.337899</td>\n",
       "      <td>...</td>\n",
       "      <td>0.946182</td>\n",
       "      <td>0.170451</td>\n",
       "      <td>-0.603641</td>\n",
       "      <td>-0.363895</td>\n",
       "      <td>-0.035424</td>\n",
       "      <td>-0.290431</td>\n",
       "      <td>0.138199</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>7104</td>\n",
       "      <td>1.953191</td>\n",
       "      <td>0.865467</td>\n",
       "      <td>-0.878878</td>\n",
       "      <td>-0.525111</td>\n",
       "      <td>-0.201118</td>\n",
       "      <td>-0.223227</td>\n",
       "      <td>0.856375</td>\n",
       "      <td>0.484807</td>\n",
       "      <td>-0.432588</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.139778</td>\n",
       "      <td>-0.708496</td>\n",
       "      <td>0.927729</td>\n",
       "      <td>-0.363895</td>\n",
       "      <td>0.692759</td>\n",
       "      <td>-0.290431</td>\n",
       "      <td>-0.946194</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.988957</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>956</td>\n",
       "      <td>-0.547294</td>\n",
       "      <td>-0.546960</td>\n",
       "      <td>1.133840</td>\n",
       "      <td>0.450826</td>\n",
       "      <td>-0.185981</td>\n",
       "      <td>-0.106545</td>\n",
       "      <td>-0.486319</td>\n",
       "      <td>1.089070</td>\n",
       "      <td>-1.088703</td>\n",
       "      <td>...</td>\n",
       "      <td>0.403202</td>\n",
       "      <td>0.170451</td>\n",
       "      <td>-0.603641</td>\n",
       "      <td>0.071213</td>\n",
       "      <td>0.328668</td>\n",
       "      <td>-0.850332</td>\n",
       "      <td>0.680395</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9050</td>\n",
       "      <td>3372</td>\n",
       "      <td>-1.242615</td>\n",
       "      <td>-1.169114</td>\n",
       "      <td>-1.630477</td>\n",
       "      <td>-1.746953</td>\n",
       "      <td>-3.088980</td>\n",
       "      <td>2.434029</td>\n",
       "      <td>-2.302105</td>\n",
       "      <td>1.022523</td>\n",
       "      <td>-1.002151</td>\n",
       "      <td>...</td>\n",
       "      <td>2.032141</td>\n",
       "      <td>1.049397</td>\n",
       "      <td>0.927729</td>\n",
       "      <td>-0.799002</td>\n",
       "      <td>-0.763607</td>\n",
       "      <td>0.829371</td>\n",
       "      <td>-0.403998</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9051</td>\n",
       "      <td>7943</td>\n",
       "      <td>-0.889606</td>\n",
       "      <td>-0.789852</td>\n",
       "      <td>1.502815</td>\n",
       "      <td>0.561845</td>\n",
       "      <td>1.265343</td>\n",
       "      <td>-1.467876</td>\n",
       "      <td>0.253775</td>\n",
       "      <td>-0.936514</td>\n",
       "      <td>1.013025</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.682757</td>\n",
       "      <td>-1.147969</td>\n",
       "      <td>-0.603641</td>\n",
       "      <td>0.506320</td>\n",
       "      <td>1.420943</td>\n",
       "      <td>-0.850332</td>\n",
       "      <td>-1.488391</td>\n",
       "      <td>0.478102</td>\n",
       "      <td>0.419727</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9052</td>\n",
       "      <td>8870</td>\n",
       "      <td>0.431506</td>\n",
       "      <td>1.257170</td>\n",
       "      <td>-1.003778</td>\n",
       "      <td>1.010499</td>\n",
       "      <td>1.280346</td>\n",
       "      <td>-0.482012</td>\n",
       "      <td>-0.087983</td>\n",
       "      <td>-0.730113</td>\n",
       "      <td>0.774852</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.682757</td>\n",
       "      <td>0.170451</td>\n",
       "      <td>-1.114098</td>\n",
       "      <td>0.071213</td>\n",
       "      <td>-1.127699</td>\n",
       "      <td>0.269470</td>\n",
       "      <td>0.680395</td>\n",
       "      <td>1.645006</td>\n",
       "      <td>-0.284615</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9053</td>\n",
       "      <td>6051</td>\n",
       "      <td>-0.068591</td>\n",
       "      <td>-0.016986</td>\n",
       "      <td>0.890963</td>\n",
       "      <td>0.225848</td>\n",
       "      <td>2.395305</td>\n",
       "      <td>-0.110377</td>\n",
       "      <td>0.539467</td>\n",
       "      <td>-2.288235</td>\n",
       "      <td>2.963153</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.139778</td>\n",
       "      <td>-0.269023</td>\n",
       "      <td>-0.093184</td>\n",
       "      <td>-1.234110</td>\n",
       "      <td>0.692759</td>\n",
       "      <td>0.829371</td>\n",
       "      <td>0.138199</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9054</td>\n",
       "      <td>1636</td>\n",
       "      <td>0.169423</td>\n",
       "      <td>-0.191417</td>\n",
       "      <td>-0.644124</td>\n",
       "      <td>-0.700191</td>\n",
       "      <td>-0.591691</td>\n",
       "      <td>0.375289</td>\n",
       "      <td>0.480559</td>\n",
       "      <td>0.768865</td>\n",
       "      <td>-0.558446</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.225737</td>\n",
       "      <td>-0.269023</td>\n",
       "      <td>0.927729</td>\n",
       "      <td>0.506320</td>\n",
       "      <td>1.785034</td>\n",
       "      <td>1.949173</td>\n",
       "      <td>-0.403998</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>9055 rows × 73 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        ID  Sequence_Length  Molar_Extinction_Coefficient  Isoelectric_Point  \\\n",
       "0     5831        -0.467064                     -0.412122          -1.087001   \n",
       "1      129        -0.277188                     -0.237691          -1.274085   \n",
       "2      591         1.244497                      1.653226          -0.856441   \n",
       "3     7104         1.953191                      0.865467          -0.878878   \n",
       "4      956        -0.547294                     -0.546960           1.133840   \n",
       "...    ...              ...                           ...                ...   \n",
       "9050  3372        -1.242615                     -1.169114          -1.630477   \n",
       "9051  7943        -0.889606                     -0.789852           1.502815   \n",
       "9052  8870         0.431506                      1.257170          -1.003778   \n",
       "9053  6051        -0.068591                     -0.016986           0.890963   \n",
       "9054  1636         0.169423                     -0.191417          -0.644124   \n",
       "\n",
       "      Aromaticity  Secondary_Structure_Helix  Secondary_Structure_Turn  \\\n",
       "0       -0.345612                  -0.618647                 -0.387055   \n",
       "1       -0.494124                  -0.261659                 -0.429878   \n",
       "2        1.134772                   1.388896                 -0.677865   \n",
       "3       -0.525111                  -0.201118                 -0.223227   \n",
       "4        0.450826                  -0.185981                 -0.106545   \n",
       "...           ...                        ...                       ...   \n",
       "9050    -1.746953                  -3.088980                  2.434029   \n",
       "9051     0.561845                   1.265343                 -1.467876   \n",
       "9052     1.010499                   1.280346                 -0.482012   \n",
       "9053     0.225848                   2.395305                 -0.110377   \n",
       "9054    -0.700191                  -0.591691                  0.375289   \n",
       "\n",
       "      Secondary_Structure_Sheet  Mean_Flexibility     Gravy  ...        Ne  \\\n",
       "0                      1.659587          0.031473 -0.108118  ... -0.682757   \n",
       "1                      0.051924          0.258738 -0.156050  ... -0.682757   \n",
       "2                      0.156050         -0.434112  0.337899  ...  0.946182   \n",
       "3                      0.856375          0.484807 -0.432588  ... -0.139778   \n",
       "4                     -0.486319          1.089070 -1.088703  ...  0.403202   \n",
       "...                         ...               ...       ...  ...       ...   \n",
       "9050                  -2.302105          1.022523 -1.002151  ...  2.032141   \n",
       "9051                   0.253775         -0.936514  1.013025  ... -0.682757   \n",
       "9052                  -0.087983         -0.730113  0.774852  ... -0.682757   \n",
       "9053                   0.539467         -2.288235  2.963153  ... -0.139778   \n",
       "9054                   0.480559          0.768865 -0.558446  ... -1.225737   \n",
       "\n",
       "            Pe        Qe        Re        Se        Te        Ve        We  \\\n",
       "0    -0.708496 -0.093184  1.376535 -0.399516 -1.410233  0.680395  0.478102   \n",
       "1    -1.147969 -1.114098  0.071213  0.328668 -0.290431 -0.946194 -0.688802   \n",
       "2     0.170451 -0.603641 -0.363895 -0.035424 -0.290431  0.138199 -0.688802   \n",
       "3    -0.708496  0.927729 -0.363895  0.692759 -0.290431 -0.946194 -0.688802   \n",
       "4     0.170451 -0.603641  0.071213  0.328668 -0.850332  0.680395 -0.688802   \n",
       "...        ...       ...       ...       ...       ...       ...       ...   \n",
       "9050  1.049397  0.927729 -0.799002 -0.763607  0.829371 -0.403998 -0.688802   \n",
       "9051 -1.147969 -0.603641  0.506320  1.420943 -0.850332 -1.488391  0.478102   \n",
       "9052  0.170451 -1.114098  0.071213 -1.127699  0.269470  0.680395  1.645006   \n",
       "9053 -0.269023 -0.093184 -1.234110  0.692759  0.829371  0.138199 -0.688802   \n",
       "9054 -0.269023  0.927729  0.506320  1.785034  1.949173 -0.403998 -0.688802   \n",
       "\n",
       "            Ye  Subcellular_Location  \n",
       "0    -0.284615                     2  \n",
       "1    -0.284615                     2  \n",
       "2    -0.284615                     0  \n",
       "3    -0.988957                     0  \n",
       "4    -0.284615                     2  \n",
       "...        ...                   ...  \n",
       "9050 -0.284615                     3  \n",
       "9051  0.419727                     3  \n",
       "9052 -0.284615                     0  \n",
       "9053 -0.284615                     0  \n",
       "9054 -0.284615                     2  \n",
       "\n",
       "[9055 rows x 73 columns]"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Split Features and Labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "y = df_data['Subcellular_Location']\n",
    "X = df_data.drop(['Subcellular_Location'], axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Drop ID Tag \n",
    "This is definitely unrelated to subcellular location"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Sequence_Length</th>\n",
       "      <th>Molar_Extinction_Coefficient</th>\n",
       "      <th>Isoelectric_Point</th>\n",
       "      <th>Aromaticity</th>\n",
       "      <th>Secondary_Structure_Helix</th>\n",
       "      <th>Secondary_Structure_Turn</th>\n",
       "      <th>Secondary_Structure_Sheet</th>\n",
       "      <th>Mean_Flexibility</th>\n",
       "      <th>Gravy</th>\n",
       "      <th>Molecular_Weight</th>\n",
       "      <th>...</th>\n",
       "      <th>Me</th>\n",
       "      <th>Ne</th>\n",
       "      <th>Pe</th>\n",
       "      <th>Qe</th>\n",
       "      <th>Re</th>\n",
       "      <th>Se</th>\n",
       "      <th>Te</th>\n",
       "      <th>Ve</th>\n",
       "      <th>We</th>\n",
       "      <th>Ye</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>-0.467064</td>\n",
       "      <td>-0.412122</td>\n",
       "      <td>-1.087001</td>\n",
       "      <td>-0.345612</td>\n",
       "      <td>-0.618647</td>\n",
       "      <td>-0.387055</td>\n",
       "      <td>1.659587</td>\n",
       "      <td>0.031473</td>\n",
       "      <td>-0.108118</td>\n",
       "      <td>-0.482574</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.899821</td>\n",
       "      <td>-0.682757</td>\n",
       "      <td>-0.708496</td>\n",
       "      <td>-0.093184</td>\n",
       "      <td>1.376535</td>\n",
       "      <td>-0.399516</td>\n",
       "      <td>-1.410233</td>\n",
       "      <td>0.680395</td>\n",
       "      <td>0.478102</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>-0.277188</td>\n",
       "      <td>-0.237691</td>\n",
       "      <td>-1.274085</td>\n",
       "      <td>-0.494124</td>\n",
       "      <td>-0.261659</td>\n",
       "      <td>-0.429878</td>\n",
       "      <td>0.051924</td>\n",
       "      <td>0.258738</td>\n",
       "      <td>-0.156050</td>\n",
       "      <td>-0.286089</td>\n",
       "      <td>...</td>\n",
       "      <td>0.001593</td>\n",
       "      <td>-0.682757</td>\n",
       "      <td>-1.147969</td>\n",
       "      <td>-1.114098</td>\n",
       "      <td>0.071213</td>\n",
       "      <td>0.328668</td>\n",
       "      <td>-0.290431</td>\n",
       "      <td>-0.946194</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>1.244497</td>\n",
       "      <td>1.653226</td>\n",
       "      <td>-0.856441</td>\n",
       "      <td>1.134772</td>\n",
       "      <td>1.388896</td>\n",
       "      <td>-0.677865</td>\n",
       "      <td>0.156050</td>\n",
       "      <td>-0.434112</td>\n",
       "      <td>0.337899</td>\n",
       "      <td>1.337163</td>\n",
       "      <td>...</td>\n",
       "      <td>0.903007</td>\n",
       "      <td>0.946182</td>\n",
       "      <td>0.170451</td>\n",
       "      <td>-0.603641</td>\n",
       "      <td>-0.363895</td>\n",
       "      <td>-0.035424</td>\n",
       "      <td>-0.290431</td>\n",
       "      <td>0.138199</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>1.953191</td>\n",
       "      <td>0.865467</td>\n",
       "      <td>-0.878878</td>\n",
       "      <td>-0.525111</td>\n",
       "      <td>-0.201118</td>\n",
       "      <td>-0.223227</td>\n",
       "      <td>0.856375</td>\n",
       "      <td>0.484807</td>\n",
       "      <td>-0.432588</td>\n",
       "      <td>1.991264</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.899821</td>\n",
       "      <td>-0.139778</td>\n",
       "      <td>-0.708496</td>\n",
       "      <td>0.927729</td>\n",
       "      <td>-0.363895</td>\n",
       "      <td>0.692759</td>\n",
       "      <td>-0.290431</td>\n",
       "      <td>-0.946194</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.988957</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>-0.547294</td>\n",
       "      <td>-0.546960</td>\n",
       "      <td>1.133840</td>\n",
       "      <td>0.450826</td>\n",
       "      <td>-0.185981</td>\n",
       "      <td>-0.106545</td>\n",
       "      <td>-0.486319</td>\n",
       "      <td>1.089070</td>\n",
       "      <td>-1.088703</td>\n",
       "      <td>-0.518221</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.899821</td>\n",
       "      <td>0.403202</td>\n",
       "      <td>0.170451</td>\n",
       "      <td>-0.603641</td>\n",
       "      <td>0.071213</td>\n",
       "      <td>0.328668</td>\n",
       "      <td>-0.850332</td>\n",
       "      <td>0.680395</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9050</td>\n",
       "      <td>-1.242615</td>\n",
       "      <td>-1.169114</td>\n",
       "      <td>-1.630477</td>\n",
       "      <td>-1.746953</td>\n",
       "      <td>-3.088980</td>\n",
       "      <td>2.434029</td>\n",
       "      <td>-2.302105</td>\n",
       "      <td>1.022523</td>\n",
       "      <td>-1.002151</td>\n",
       "      <td>-1.249362</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.899821</td>\n",
       "      <td>2.032141</td>\n",
       "      <td>1.049397</td>\n",
       "      <td>0.927729</td>\n",
       "      <td>-0.799002</td>\n",
       "      <td>-0.763607</td>\n",
       "      <td>0.829371</td>\n",
       "      <td>-0.403998</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9051</td>\n",
       "      <td>-0.889606</td>\n",
       "      <td>-0.789852</td>\n",
       "      <td>1.502815</td>\n",
       "      <td>0.561845</td>\n",
       "      <td>1.265343</td>\n",
       "      <td>-1.467876</td>\n",
       "      <td>0.253775</td>\n",
       "      <td>-0.936514</td>\n",
       "      <td>1.013025</td>\n",
       "      <td>-0.873977</td>\n",
       "      <td>...</td>\n",
       "      <td>0.903007</td>\n",
       "      <td>-0.682757</td>\n",
       "      <td>-1.147969</td>\n",
       "      <td>-0.603641</td>\n",
       "      <td>0.506320</td>\n",
       "      <td>1.420943</td>\n",
       "      <td>-0.850332</td>\n",
       "      <td>-1.488391</td>\n",
       "      <td>0.478102</td>\n",
       "      <td>0.419727</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9052</td>\n",
       "      <td>0.431506</td>\n",
       "      <td>1.257170</td>\n",
       "      <td>-1.003778</td>\n",
       "      <td>1.010499</td>\n",
       "      <td>1.280346</td>\n",
       "      <td>-0.482012</td>\n",
       "      <td>-0.087983</td>\n",
       "      <td>-0.730113</td>\n",
       "      <td>0.774852</td>\n",
       "      <td>0.460851</td>\n",
       "      <td>...</td>\n",
       "      <td>0.903007</td>\n",
       "      <td>-0.682757</td>\n",
       "      <td>0.170451</td>\n",
       "      <td>-1.114098</td>\n",
       "      <td>0.071213</td>\n",
       "      <td>-1.127699</td>\n",
       "      <td>0.269470</td>\n",
       "      <td>0.680395</td>\n",
       "      <td>1.645006</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9053</td>\n",
       "      <td>-0.068591</td>\n",
       "      <td>-0.016986</td>\n",
       "      <td>0.890963</td>\n",
       "      <td>0.225848</td>\n",
       "      <td>2.395305</td>\n",
       "      <td>-0.110377</td>\n",
       "      <td>0.539467</td>\n",
       "      <td>-2.288235</td>\n",
       "      <td>2.963153</td>\n",
       "      <td>-0.102127</td>\n",
       "      <td>...</td>\n",
       "      <td>0.903007</td>\n",
       "      <td>-0.139778</td>\n",
       "      <td>-0.269023</td>\n",
       "      <td>-0.093184</td>\n",
       "      <td>-1.234110</td>\n",
       "      <td>0.692759</td>\n",
       "      <td>0.829371</td>\n",
       "      <td>0.138199</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9054</td>\n",
       "      <td>0.169423</td>\n",
       "      <td>-0.191417</td>\n",
       "      <td>-0.644124</td>\n",
       "      <td>-0.700191</td>\n",
       "      <td>-0.591691</td>\n",
       "      <td>0.375289</td>\n",
       "      <td>0.480559</td>\n",
       "      <td>0.768865</td>\n",
       "      <td>-0.558446</td>\n",
       "      <td>0.158073</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.899821</td>\n",
       "      <td>-1.225737</td>\n",
       "      <td>-0.269023</td>\n",
       "      <td>0.927729</td>\n",
       "      <td>0.506320</td>\n",
       "      <td>1.785034</td>\n",
       "      <td>1.949173</td>\n",
       "      <td>-0.403998</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>9055 rows × 71 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      Sequence_Length  Molar_Extinction_Coefficient  Isoelectric_Point  \\\n",
       "0           -0.467064                     -0.412122          -1.087001   \n",
       "1           -0.277188                     -0.237691          -1.274085   \n",
       "2            1.244497                      1.653226          -0.856441   \n",
       "3            1.953191                      0.865467          -0.878878   \n",
       "4           -0.547294                     -0.546960           1.133840   \n",
       "...               ...                           ...                ...   \n",
       "9050        -1.242615                     -1.169114          -1.630477   \n",
       "9051        -0.889606                     -0.789852           1.502815   \n",
       "9052         0.431506                      1.257170          -1.003778   \n",
       "9053        -0.068591                     -0.016986           0.890963   \n",
       "9054         0.169423                     -0.191417          -0.644124   \n",
       "\n",
       "      Aromaticity  Secondary_Structure_Helix  Secondary_Structure_Turn  \\\n",
       "0       -0.345612                  -0.618647                 -0.387055   \n",
       "1       -0.494124                  -0.261659                 -0.429878   \n",
       "2        1.134772                   1.388896                 -0.677865   \n",
       "3       -0.525111                  -0.201118                 -0.223227   \n",
       "4        0.450826                  -0.185981                 -0.106545   \n",
       "...           ...                        ...                       ...   \n",
       "9050    -1.746953                  -3.088980                  2.434029   \n",
       "9051     0.561845                   1.265343                 -1.467876   \n",
       "9052     1.010499                   1.280346                 -0.482012   \n",
       "9053     0.225848                   2.395305                 -0.110377   \n",
       "9054    -0.700191                  -0.591691                  0.375289   \n",
       "\n",
       "      Secondary_Structure_Sheet  Mean_Flexibility     Gravy  Molecular_Weight  \\\n",
       "0                      1.659587          0.031473 -0.108118         -0.482574   \n",
       "1                      0.051924          0.258738 -0.156050         -0.286089   \n",
       "2                      0.156050         -0.434112  0.337899          1.337163   \n",
       "3                      0.856375          0.484807 -0.432588          1.991264   \n",
       "4                     -0.486319          1.089070 -1.088703         -0.518221   \n",
       "...                         ...               ...       ...               ...   \n",
       "9050                  -2.302105          1.022523 -1.002151         -1.249362   \n",
       "9051                   0.253775         -0.936514  1.013025         -0.873977   \n",
       "9052                  -0.087983         -0.730113  0.774852          0.460851   \n",
       "9053                   0.539467         -2.288235  2.963153         -0.102127   \n",
       "9054                   0.480559          0.768865 -0.558446          0.158073   \n",
       "\n",
       "      ...        Me        Ne        Pe        Qe        Re        Se  \\\n",
       "0     ... -0.899821 -0.682757 -0.708496 -0.093184  1.376535 -0.399516   \n",
       "1     ...  0.001593 -0.682757 -1.147969 -1.114098  0.071213  0.328668   \n",
       "2     ...  0.903007  0.946182  0.170451 -0.603641 -0.363895 -0.035424   \n",
       "3     ... -0.899821 -0.139778 -0.708496  0.927729 -0.363895  0.692759   \n",
       "4     ... -0.899821  0.403202  0.170451 -0.603641  0.071213  0.328668   \n",
       "...   ...       ...       ...       ...       ...       ...       ...   \n",
       "9050  ... -0.899821  2.032141  1.049397  0.927729 -0.799002 -0.763607   \n",
       "9051  ...  0.903007 -0.682757 -1.147969 -0.603641  0.506320  1.420943   \n",
       "9052  ...  0.903007 -0.682757  0.170451 -1.114098  0.071213 -1.127699   \n",
       "9053  ...  0.903007 -0.139778 -0.269023 -0.093184 -1.234110  0.692759   \n",
       "9054  ... -0.899821 -1.225737 -0.269023  0.927729  0.506320  1.785034   \n",
       "\n",
       "            Te        Ve        We        Ye  \n",
       "0    -1.410233  0.680395  0.478102 -0.284615  \n",
       "1    -0.290431 -0.946194 -0.688802 -0.284615  \n",
       "2    -0.290431  0.138199 -0.688802 -0.284615  \n",
       "3    -0.290431 -0.946194 -0.688802 -0.988957  \n",
       "4    -0.850332  0.680395 -0.688802 -0.284615  \n",
       "...        ...       ...       ...       ...  \n",
       "9050  0.829371 -0.403998 -0.688802 -0.284615  \n",
       "9051 -0.850332 -1.488391  0.478102  0.419727  \n",
       "9052  0.269470  0.680395  1.645006 -0.284615  \n",
       "9053  0.829371  0.138199 -0.688802 -0.284615  \n",
       "9054  1.949173 -0.403998 -0.688802 -0.284615  \n",
       "\n",
       "[9055 rows x 71 columns]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = X.drop(columns=['ID'])\n",
    "X"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Feature Selection (Recursive Feature Elimination)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x396 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fd22fef88d0>"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from yellowbrick.model_selection import RFECV\n",
    "\n",
    "lr = True\n",
    "\n",
    "if lr == False:\n",
    "    svc = SVC(kernel='linear',probability=True,random_state=1)\n",
    "    visualizer = RFECV(svc)\n",
    "\n",
    "\n",
    "if lr == True:\n",
    "    #visualizer = RFECV(LogisticRegression(class_weight='balanced'))\n",
    "    visualizer = RFECV(LogisticRegression())\n",
    "    \n",
    "visualizer.fit(X,y)\n",
    "visualizer.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Remove Features That Aren't Useful"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 291,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Redundant features = Index(['Wg', 'Cs', 'Fs', 'Is', 'Vs', 'Ws'], dtype='object')\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Sequence_Length</th>\n",
       "      <th>Molar_Extinction_Coefficient</th>\n",
       "      <th>Isoelectric_Point</th>\n",
       "      <th>Aromaticity</th>\n",
       "      <th>Secondary_Structure_Helix</th>\n",
       "      <th>Secondary_Structure_Turn</th>\n",
       "      <th>Secondary_Structure_Sheet</th>\n",
       "      <th>Mean_Flexibility</th>\n",
       "      <th>Gravy</th>\n",
       "      <th>Molecular_Weight</th>\n",
       "      <th>...</th>\n",
       "      <th>Me</th>\n",
       "      <th>Ne</th>\n",
       "      <th>Pe</th>\n",
       "      <th>Qe</th>\n",
       "      <th>Re</th>\n",
       "      <th>Se</th>\n",
       "      <th>Te</th>\n",
       "      <th>Ve</th>\n",
       "      <th>We</th>\n",
       "      <th>Ye</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>-0.467064</td>\n",
       "      <td>-0.412122</td>\n",
       "      <td>-1.087001</td>\n",
       "      <td>-0.345612</td>\n",
       "      <td>-0.618647</td>\n",
       "      <td>-0.387055</td>\n",
       "      <td>1.659587</td>\n",
       "      <td>0.031473</td>\n",
       "      <td>-0.108118</td>\n",
       "      <td>-0.482574</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.899821</td>\n",
       "      <td>-0.682757</td>\n",
       "      <td>-0.708496</td>\n",
       "      <td>-0.093184</td>\n",
       "      <td>1.376535</td>\n",
       "      <td>-0.399516</td>\n",
       "      <td>-1.410233</td>\n",
       "      <td>0.680395</td>\n",
       "      <td>0.478102</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>-0.277188</td>\n",
       "      <td>-0.237691</td>\n",
       "      <td>-1.274085</td>\n",
       "      <td>-0.494124</td>\n",
       "      <td>-0.261659</td>\n",
       "      <td>-0.429878</td>\n",
       "      <td>0.051924</td>\n",
       "      <td>0.258738</td>\n",
       "      <td>-0.156050</td>\n",
       "      <td>-0.286089</td>\n",
       "      <td>...</td>\n",
       "      <td>0.001593</td>\n",
       "      <td>-0.682757</td>\n",
       "      <td>-1.147969</td>\n",
       "      <td>-1.114098</td>\n",
       "      <td>0.071213</td>\n",
       "      <td>0.328668</td>\n",
       "      <td>-0.290431</td>\n",
       "      <td>-0.946194</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>1.244497</td>\n",
       "      <td>1.653226</td>\n",
       "      <td>-0.856441</td>\n",
       "      <td>1.134772</td>\n",
       "      <td>1.388896</td>\n",
       "      <td>-0.677865</td>\n",
       "      <td>0.156050</td>\n",
       "      <td>-0.434112</td>\n",
       "      <td>0.337899</td>\n",
       "      <td>1.337163</td>\n",
       "      <td>...</td>\n",
       "      <td>0.903007</td>\n",
       "      <td>0.946182</td>\n",
       "      <td>0.170451</td>\n",
       "      <td>-0.603641</td>\n",
       "      <td>-0.363895</td>\n",
       "      <td>-0.035424</td>\n",
       "      <td>-0.290431</td>\n",
       "      <td>0.138199</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>1.953191</td>\n",
       "      <td>0.865467</td>\n",
       "      <td>-0.878878</td>\n",
       "      <td>-0.525111</td>\n",
       "      <td>-0.201118</td>\n",
       "      <td>-0.223227</td>\n",
       "      <td>0.856375</td>\n",
       "      <td>0.484807</td>\n",
       "      <td>-0.432588</td>\n",
       "      <td>1.991264</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.899821</td>\n",
       "      <td>-0.139778</td>\n",
       "      <td>-0.708496</td>\n",
       "      <td>0.927729</td>\n",
       "      <td>-0.363895</td>\n",
       "      <td>0.692759</td>\n",
       "      <td>-0.290431</td>\n",
       "      <td>-0.946194</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.988957</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>-0.547294</td>\n",
       "      <td>-0.546960</td>\n",
       "      <td>1.133840</td>\n",
       "      <td>0.450826</td>\n",
       "      <td>-0.185981</td>\n",
       "      <td>-0.106545</td>\n",
       "      <td>-0.486319</td>\n",
       "      <td>1.089070</td>\n",
       "      <td>-1.088703</td>\n",
       "      <td>-0.518221</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.899821</td>\n",
       "      <td>0.403202</td>\n",
       "      <td>0.170451</td>\n",
       "      <td>-0.603641</td>\n",
       "      <td>0.071213</td>\n",
       "      <td>0.328668</td>\n",
       "      <td>-0.850332</td>\n",
       "      <td>0.680395</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9050</td>\n",
       "      <td>-1.242615</td>\n",
       "      <td>-1.169114</td>\n",
       "      <td>-1.630477</td>\n",
       "      <td>-1.746953</td>\n",
       "      <td>-3.088980</td>\n",
       "      <td>2.434029</td>\n",
       "      <td>-2.302105</td>\n",
       "      <td>1.022523</td>\n",
       "      <td>-1.002151</td>\n",
       "      <td>-1.249362</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.899821</td>\n",
       "      <td>2.032141</td>\n",
       "      <td>1.049397</td>\n",
       "      <td>0.927729</td>\n",
       "      <td>-0.799002</td>\n",
       "      <td>-0.763607</td>\n",
       "      <td>0.829371</td>\n",
       "      <td>-0.403998</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9051</td>\n",
       "      <td>-0.889606</td>\n",
       "      <td>-0.789852</td>\n",
       "      <td>1.502815</td>\n",
       "      <td>0.561845</td>\n",
       "      <td>1.265343</td>\n",
       "      <td>-1.467876</td>\n",
       "      <td>0.253775</td>\n",
       "      <td>-0.936514</td>\n",
       "      <td>1.013025</td>\n",
       "      <td>-0.873977</td>\n",
       "      <td>...</td>\n",
       "      <td>0.903007</td>\n",
       "      <td>-0.682757</td>\n",
       "      <td>-1.147969</td>\n",
       "      <td>-0.603641</td>\n",
       "      <td>0.506320</td>\n",
       "      <td>1.420943</td>\n",
       "      <td>-0.850332</td>\n",
       "      <td>-1.488391</td>\n",
       "      <td>0.478102</td>\n",
       "      <td>0.419727</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9052</td>\n",
       "      <td>0.431506</td>\n",
       "      <td>1.257170</td>\n",
       "      <td>-1.003778</td>\n",
       "      <td>1.010499</td>\n",
       "      <td>1.280346</td>\n",
       "      <td>-0.482012</td>\n",
       "      <td>-0.087983</td>\n",
       "      <td>-0.730113</td>\n",
       "      <td>0.774852</td>\n",
       "      <td>0.460851</td>\n",
       "      <td>...</td>\n",
       "      <td>0.903007</td>\n",
       "      <td>-0.682757</td>\n",
       "      <td>0.170451</td>\n",
       "      <td>-1.114098</td>\n",
       "      <td>0.071213</td>\n",
       "      <td>-1.127699</td>\n",
       "      <td>0.269470</td>\n",
       "      <td>0.680395</td>\n",
       "      <td>1.645006</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9053</td>\n",
       "      <td>-0.068591</td>\n",
       "      <td>-0.016986</td>\n",
       "      <td>0.890963</td>\n",
       "      <td>0.225848</td>\n",
       "      <td>2.395305</td>\n",
       "      <td>-0.110377</td>\n",
       "      <td>0.539467</td>\n",
       "      <td>-2.288235</td>\n",
       "      <td>2.963153</td>\n",
       "      <td>-0.102127</td>\n",
       "      <td>...</td>\n",
       "      <td>0.903007</td>\n",
       "      <td>-0.139778</td>\n",
       "      <td>-0.269023</td>\n",
       "      <td>-0.093184</td>\n",
       "      <td>-1.234110</td>\n",
       "      <td>0.692759</td>\n",
       "      <td>0.829371</td>\n",
       "      <td>0.138199</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9054</td>\n",
       "      <td>0.169423</td>\n",
       "      <td>-0.191417</td>\n",
       "      <td>-0.644124</td>\n",
       "      <td>-0.700191</td>\n",
       "      <td>-0.591691</td>\n",
       "      <td>0.375289</td>\n",
       "      <td>0.480559</td>\n",
       "      <td>0.768865</td>\n",
       "      <td>-0.558446</td>\n",
       "      <td>0.158073</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.899821</td>\n",
       "      <td>-1.225737</td>\n",
       "      <td>-0.269023</td>\n",
       "      <td>0.927729</td>\n",
       "      <td>0.506320</td>\n",
       "      <td>1.785034</td>\n",
       "      <td>1.949173</td>\n",
       "      <td>-0.403998</td>\n",
       "      <td>-0.688802</td>\n",
       "      <td>-0.284615</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>9055 rows × 65 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      Sequence_Length  Molar_Extinction_Coefficient  Isoelectric_Point  \\\n",
       "0           -0.467064                     -0.412122          -1.087001   \n",
       "1           -0.277188                     -0.237691          -1.274085   \n",
       "2            1.244497                      1.653226          -0.856441   \n",
       "3            1.953191                      0.865467          -0.878878   \n",
       "4           -0.547294                     -0.546960           1.133840   \n",
       "...               ...                           ...                ...   \n",
       "9050        -1.242615                     -1.169114          -1.630477   \n",
       "9051        -0.889606                     -0.789852           1.502815   \n",
       "9052         0.431506                      1.257170          -1.003778   \n",
       "9053        -0.068591                     -0.016986           0.890963   \n",
       "9054         0.169423                     -0.191417          -0.644124   \n",
       "\n",
       "      Aromaticity  Secondary_Structure_Helix  Secondary_Structure_Turn  \\\n",
       "0       -0.345612                  -0.618647                 -0.387055   \n",
       "1       -0.494124                  -0.261659                 -0.429878   \n",
       "2        1.134772                   1.388896                 -0.677865   \n",
       "3       -0.525111                  -0.201118                 -0.223227   \n",
       "4        0.450826                  -0.185981                 -0.106545   \n",
       "...           ...                        ...                       ...   \n",
       "9050    -1.746953                  -3.088980                  2.434029   \n",
       "9051     0.561845                   1.265343                 -1.467876   \n",
       "9052     1.010499                   1.280346                 -0.482012   \n",
       "9053     0.225848                   2.395305                 -0.110377   \n",
       "9054    -0.700191                  -0.591691                  0.375289   \n",
       "\n",
       "      Secondary_Structure_Sheet  Mean_Flexibility     Gravy  Molecular_Weight  \\\n",
       "0                      1.659587          0.031473 -0.108118         -0.482574   \n",
       "1                      0.051924          0.258738 -0.156050         -0.286089   \n",
       "2                      0.156050         -0.434112  0.337899          1.337163   \n",
       "3                      0.856375          0.484807 -0.432588          1.991264   \n",
       "4                     -0.486319          1.089070 -1.088703         -0.518221   \n",
       "...                         ...               ...       ...               ...   \n",
       "9050                  -2.302105          1.022523 -1.002151         -1.249362   \n",
       "9051                   0.253775         -0.936514  1.013025         -0.873977   \n",
       "9052                  -0.087983         -0.730113  0.774852          0.460851   \n",
       "9053                   0.539467         -2.288235  2.963153         -0.102127   \n",
       "9054                   0.480559          0.768865 -0.558446          0.158073   \n",
       "\n",
       "      ...        Me        Ne        Pe        Qe        Re        Se  \\\n",
       "0     ... -0.899821 -0.682757 -0.708496 -0.093184  1.376535 -0.399516   \n",
       "1     ...  0.001593 -0.682757 -1.147969 -1.114098  0.071213  0.328668   \n",
       "2     ...  0.903007  0.946182  0.170451 -0.603641 -0.363895 -0.035424   \n",
       "3     ... -0.899821 -0.139778 -0.708496  0.927729 -0.363895  0.692759   \n",
       "4     ... -0.899821  0.403202  0.170451 -0.603641  0.071213  0.328668   \n",
       "...   ...       ...       ...       ...       ...       ...       ...   \n",
       "9050  ... -0.899821  2.032141  1.049397  0.927729 -0.799002 -0.763607   \n",
       "9051  ...  0.903007 -0.682757 -1.147969 -0.603641  0.506320  1.420943   \n",
       "9052  ...  0.903007 -0.682757  0.170451 -1.114098  0.071213 -1.127699   \n",
       "9053  ...  0.903007 -0.139778 -0.269023 -0.093184 -1.234110  0.692759   \n",
       "9054  ... -0.899821 -1.225737 -0.269023  0.927729  0.506320  1.785034   \n",
       "\n",
       "            Te        Ve        We        Ye  \n",
       "0    -1.410233  0.680395  0.478102 -0.284615  \n",
       "1    -0.290431 -0.946194 -0.688802 -0.284615  \n",
       "2    -0.290431  0.138199 -0.688802 -0.284615  \n",
       "3    -0.290431 -0.946194 -0.688802 -0.988957  \n",
       "4    -0.850332  0.680395 -0.688802 -0.284615  \n",
       "...        ...       ...       ...       ...  \n",
       "9050  0.829371 -0.403998 -0.688802 -0.284615  \n",
       "9051 -0.850332 -1.488391  0.478102  0.419727  \n",
       "9052  0.269470  0.680395  1.645006 -0.284615  \n",
       "9053  0.829371  0.138199 -0.688802 -0.284615  \n",
       "9054  1.949173 -0.403998 -0.688802 -0.284615  \n",
       "\n",
       "[9055 rows x 65 columns]"
      ]
     },
     "execution_count": 291,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# drop redundant features\n",
    "temp = pd.Series(visualizer.support_, index=list(X.columns))\n",
    "deselect = temp[temp==False].index\n",
    "\n",
    "#deselect = ['Cs', 'Fs', 'Is', 'Vs', 'Ws', 'Wg']\n",
    "\n",
    "X_reduced = X.drop(columns=deselect)\n",
    "print(\"Redundant features = {}\".format(deselect))\n",
    "X_reduced"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train/Test Split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X_reduced, y, test_size=0.2, random_state=2020)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train and Cross-Validate Different Models\n",
    "Note we use 10-fold CV on (X_train,y_train) to grid search the best parameters.\n",
    "\n",
    "We then run each mclassifer (with its best parameters) once on (X_test,y_test) to get a score. We will use these test scores to compare the different classifiers and identify which is best for making the blind predictions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_bin = label_binarize(y, classes=[0,1,2,3])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 0, 1, 0],\n",
       "       [0, 0, 1, 0],\n",
       "       [1, 0, 0, 0],\n",
       "       ...,\n",
       "       [1, 0, 0, 0],\n",
       "       [1, 0, 0, 0],\n",
       "       [0, 0, 1, 0]])"
      ]
     },
     "execution_count": 200,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_bin"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 294,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running 10-fold CV\n",
      "C = 0.1, mean acc = 0.6194164680602453, best acc = 0\n",
      "C = 0.5, mean acc = 0.6486684132230469, best acc = 0.6194164680602453\n",
      "C = 1.0, mean acc = 0.6535036383438138, best acc = 0.6486684132230469\n",
      "C = 5.0, mean acc = 0.6493666373235735, best acc = 0.6535036383438138\n",
      "C = 10.0, mean acc = 0.6409492024358391, best acc = 0.6535036383438138\n",
      "C = 20.0, mean acc = 0.6344555667074501, best acc = 0.6535036383438138\n",
      "10-fold CV complete \n",
      " best C = 1.0\n",
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.56884   0.56987   0.56936       551\n",
      "        mito    0.60684   0.80682   0.69268       264\n",
      "     nuclear    0.69611   0.60522   0.64749       651\n",
      "    secreted    0.82164   0.81449   0.81805       345\n",
      "\n",
      "    accuracy                        0.66372      1811\n",
      "   macro avg    0.67336   0.69910   0.68190      1811\n",
      "weighted avg    0.66829   0.66372   0.66280      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    "### SVM rbf ###\n",
    "best_acc = 0\n",
    "best_C_idx = 0\n",
    "C_list = [0.1,0.5,1.0,5.0,10.0,20.0]\n",
    "C_num = len(C_list)\n",
    "\n",
    "print(\"Running 10-fold CV\")\n",
    "for i in range(C_num):\n",
    "    clf = SVC(kernel='rbf', C=C_list[i], class_weight='balanced',decision_function_shape='ovr')\n",
    "    #clf_acc = cross_val_score(clf, X_reduced, y, cv=10)\n",
    "    clf_acc = cross_val_score(clf, X_train, y_train, cv=10)\n",
    "    print(\"C = {}, mean acc = {}, best acc = {}\".format(C_list[i],clf_acc.mean(), best_acc))\n",
    "    if clf_acc.mean() > best_acc:\n",
    "        best_acc = clf_acc.mean()\n",
    "        best_C_idx_rbf = i\n",
    "print(\"10-fold CV complete \\n best C = {}\".format(C_list[best_C_idx_rbf]))\n",
    "\n",
    "#best_C_idx_rbf = 2\n",
    "\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_svmr = SVC(kernel='rbf', C=C_list[best_C_idx_rbf],probability=True, class_weight='balanced',decision_function_shape='ovr')\n",
    "#clf_svmr_acc = cross_val_score(clf_svmr, X_test, y_test, cv=10)\n",
    "clf_svmr.fit(X_train,y_train)\n",
    "y_pred_svmr = clf_svmr.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_svmr, target_names=target_names,digits=5))\n",
    "    \n",
    "#SVMrbf_results = pd.DataFrame({'Model': 'SVM rbf', 'Accuracy': clf_svmr_acc.mean(), 'Acc_Std': clf_svmr_acc.std()}, index=[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 339,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running 10-fold CV\n",
      "C = 1.0, deg = 2, mean acc = 0.5821244951506097, best acc = 0\n",
      "C = 1.0, deg = 3, mean acc = 0.6141741109978207, best acc = 0.5821244951506097\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-339-31c85faef27c>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m     12\u001b[0m         \u001b[0mclf\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mSVC\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mkernel\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'poly'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mC\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mC_list\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdegree\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdeg_list\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mj\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mclass_weight\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'balanced'\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mdecision_function_shape\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'ovr'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     13\u001b[0m         \u001b[0;31m#clf_acc = cross_val_score(clf, X_reduced, y, cv=10)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 14\u001b[0;31m         \u001b[0mclf_acc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcross_val_score\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mclf\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mX_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcv\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m10\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     15\u001b[0m         \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"C = {}, deg = {}, mean acc = {}, best acc = {}\"\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mformat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mC_list\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mdeg_list\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mj\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mclf_acc\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmean\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbest_acc\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     16\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mclf_acc\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmean\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m>\u001b[0m \u001b[0mbest_acc\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/sklearn/model_selection/_validation.py\u001b[0m in \u001b[0;36mcross_val_score\u001b[0;34m(estimator, X, y, groups, scoring, cv, n_jobs, verbose, fit_params, pre_dispatch, error_score)\u001b[0m\n\u001b[1;32m    389\u001b[0m                                 \u001b[0mfit_params\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    390\u001b[0m                                 \u001b[0mpre_dispatch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mpre_dispatch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 391\u001b[0;31m                                 error_score=error_score)\n\u001b[0m\u001b[1;32m    392\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mcv_results\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'test_score'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    393\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/sklearn/model_selection/_validation.py\u001b[0m in \u001b[0;36mcross_validate\u001b[0;34m(estimator, X, y, groups, scoring, cv, n_jobs, verbose, fit_params, pre_dispatch, return_train_score, return_estimator, error_score)\u001b[0m\n\u001b[1;32m    230\u001b[0m             \u001b[0mreturn_times\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mreturn_estimator\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mreturn_estimator\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    231\u001b[0m             error_score=error_score)\n\u001b[0;32m--> 232\u001b[0;31m         for train, test in cv.split(X, y, groups))\n\u001b[0m\u001b[1;32m    233\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    234\u001b[0m     \u001b[0mzipped_scores\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mscores\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/joblib/parallel.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, iterable)\u001b[0m\n\u001b[1;32m    922\u001b[0m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_iterating\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_iterator\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    923\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 924\u001b[0;31m             \u001b[0;32mwhile\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdispatch_one_batch\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0miterator\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    925\u001b[0m                 \u001b[0;32mpass\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    926\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/joblib/parallel.py\u001b[0m in \u001b[0;36mdispatch_one_batch\u001b[0;34m(self, iterator)\u001b[0m\n\u001b[1;32m    757\u001b[0m                 \u001b[0;32mreturn\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    758\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 759\u001b[0;31m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_dispatch\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtasks\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    760\u001b[0m                 \u001b[0;32mreturn\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    761\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/joblib/parallel.py\u001b[0m in \u001b[0;36m_dispatch\u001b[0;34m(self, batch)\u001b[0m\n\u001b[1;32m    714\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_lock\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    715\u001b[0m             \u001b[0mjob_idx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_jobs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 716\u001b[0;31m             \u001b[0mjob\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_backend\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mapply_async\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbatch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcb\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    717\u001b[0m             \u001b[0;31m# A job can complete so quickly than its callback is\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    718\u001b[0m             \u001b[0;31m# called before we get here, causing self._jobs to\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/joblib/_parallel_backends.py\u001b[0m in \u001b[0;36mapply_async\u001b[0;34m(self, func, callback)\u001b[0m\n\u001b[1;32m    180\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mapply_async\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallback\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    181\u001b[0m         \u001b[0;34m\"\"\"Schedule a func to be run\"\"\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 182\u001b[0;31m         \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mImmediateResult\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfunc\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    183\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mcallback\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    184\u001b[0m             \u001b[0mcallback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/joblib/_parallel_backends.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, batch)\u001b[0m\n\u001b[1;32m    547\u001b[0m         \u001b[0;31m# Don't delay the application, to avoid keeping the input\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    548\u001b[0m         \u001b[0;31m# arguments in memory\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 549\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mbatch\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    550\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    551\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/joblib/parallel.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    223\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mparallel_backend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_backend\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_jobs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_n_jobs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    224\u001b[0m             return [func(*args, **kwargs)\n\u001b[0;32m--> 225\u001b[0;31m                     for func, args, kwargs in self.items]\n\u001b[0m\u001b[1;32m    226\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    227\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__len__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/joblib/parallel.py\u001b[0m in \u001b[0;36m<listcomp>\u001b[0;34m(.0)\u001b[0m\n\u001b[1;32m    223\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mparallel_backend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_backend\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_jobs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_n_jobs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    224\u001b[0m             return [func(*args, **kwargs)\n\u001b[0;32m--> 225\u001b[0;31m                     for func, args, kwargs in self.items]\n\u001b[0m\u001b[1;32m    226\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    227\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__len__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/sklearn/model_selection/_validation.py\u001b[0m in \u001b[0;36m_fit_and_score\u001b[0;34m(estimator, X, y, scorer, train, test, verbose, parameters, fit_params, return_train_score, return_parameters, return_n_test_samples, return_times, return_estimator, error_score)\u001b[0m\n\u001b[1;32m    514\u001b[0m             \u001b[0mestimator\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    515\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 516\u001b[0;31m             \u001b[0mestimator\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    517\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    518\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/sklearn/svm/base.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X, y, sample_weight)\u001b[0m\n\u001b[1;32m    207\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    208\u001b[0m         \u001b[0mseed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mrnd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0miinfo\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'i'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmax\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 209\u001b[0;31m         \u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msample_weight\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msolver_type\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkernel\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrandom_seed\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mseed\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    210\u001b[0m         \u001b[0;31m# see comment on the other call to np.iinfo in this file\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    211\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/lib/python3.7/site-packages/sklearn/svm/base.py\u001b[0m in \u001b[0;36m_dense_fit\u001b[0;34m(self, X, y, sample_weight, solver_type, kernel, random_seed)\u001b[0m\n\u001b[1;32m    266\u001b[0m                 \u001b[0mcache_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcache_size\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcoef0\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcoef0\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    267\u001b[0m                 \u001b[0mgamma\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_gamma\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepsilon\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mepsilon\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 268\u001b[0;31m                 max_iter=self.max_iter, random_seed=random_seed)\n\u001b[0m\u001b[1;32m    269\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    270\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_warn_from_fit_status\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "### SVM polynomial ###\n",
    "best_acc = 0\n",
    "best_C_idx_poly = 0\n",
    "C_list = [1.0,10.0,100.0]\n",
    "deg_list=[2,3,4,5]\n",
    "C_num = len(C_list)\n",
    "d_num = len(deg_list)\n",
    "\n",
    "print(\"Running 10-fold CV\")\n",
    "for i in range(C_num):\n",
    "    for j in range(d_num):\n",
    "        clf = SVC(kernel='poly', C=C_list[i], degree=deg_list[j], class_weight='balanced',decision_function_shape='ovr')\n",
    "        #clf_acc = cross_val_score(clf, X_reduced, y, cv=10)\n",
    "        clf_acc = cross_val_score(clf, X_train, y_train, cv=10)\n",
    "        print(\"C = {}, deg = {}, mean acc = {}, best acc = {}\".format(C_list[i],deg_list[j],clf_acc.mean(), best_acc))\n",
    "        if clf_acc.mean() > best_acc:\n",
    "            best_acc = clf_acc.mean()\n",
    "            best_C_idx_poly = i\n",
    "            best_d_idx = j\n",
    "print(\"10-fold CV complete \\n best C = {}, best deg = {}\".format(C_list[best_C_idx_poly],deg_list[best_d_idx]))\n",
    "\n",
    "\n",
    "#best_C_idx_poly = 1\n",
    "#best_d_idx = 1\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_svmp = SVC(kernel='poly', C=C_list[best_C_idx_poly],degree=deg_list[best_d_idx],probability=True, class_weight='balanced',decision_function_shape='ovr')\n",
    "#clf_svmp_acc = cross_val_score(clf_svmp, X_reduced, y, cv=10)\n",
    "clf_svmp.fit(X_train,y_train)\n",
    "y_pred_svmp = clf_svmp.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_svmp, target_names=target_names,digits=5))\n",
    "\n",
    "#SVMpoly_results = pd.DataFrame({'Model': 'SVM poly', 'Accuracy': clf_svmp_acc.mean(), 'Acc_Std': clf_svmp_acc.std()}, index=[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 296,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.53208   0.51180   0.52174       551\n",
      "        mito    0.66540   0.66288   0.66414       264\n",
      "     nuclear    0.62093   0.67435   0.64654       651\n",
      "    secreted    0.81029   0.73043   0.76829       345\n",
      "\n",
      "    accuracy                        0.63390      1811\n",
      "   macro avg    0.65717   0.64486   0.65018      1811\n",
      "weighted avg    0.63645   0.63390   0.63433      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    "### Logistic Regression ###\n",
    "clf_lr = LogisticRegression(multi_class='ovr')\n",
    "#clf_lr_acc = cross_val_score(clf_lr, X_reduced, y, cv=10)\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_lr.fit(X_train,y_train)\n",
    "y_pred_lr = clf_lr.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_lr, target_names=target_names,digits=5))\n",
    "\n",
    "#LogReg_results = pd.DataFrame({'Model': 'Logistic Regression', 'Accuracy': clf_lr_acc.mean(), 'Acc_Std': clf_lr_acc.std()}, index=[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 340,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running 10-fold CV\n",
      "i = 1, mean acc = 0.5585157362851293, best acc = 0\n",
      "i = 2, mean acc = 0.5496881191350039, best acc = 0.5585157362851293\n",
      "i = 3, mean acc = 0.5727485898579031, best acc = 0.5585157362851293\n",
      "i = 4, mean acc = 0.5891755808792869, best acc = 0.5727485898579031\n",
      "i = 5, mean acc = 0.5861363274123308, best acc = 0.5891755808792869\n",
      "i = 6, mean acc = 0.5951211022415954, best acc = 0.5891755808792869\n",
      "i = 7, mean acc = 0.5992459130611776, best acc = 0.5951211022415954\n",
      "i = 8, mean acc = 0.6064309206736226, best acc = 0.5992459130611776\n",
      "i = 9, mean acc = 0.6067056464774833, best acc = 0.6064309206736226\n",
      "i = 10, mean acc = 0.6062876663046376, best acc = 0.6067056464774833\n",
      "i = 11, mean acc = 0.6029782530266864, best acc = 0.6067056464774833\n",
      "i = 12, mean acc = 0.6085002777276323, best acc = 0.6067056464774833\n",
      "i = 13, mean acc = 0.6123608430573443, best acc = 0.6085002777276323\n",
      "i = 14, mean acc = 0.6112625271134373, best acc = 0.6123608430573443\n",
      "i = 15, mean acc = 0.6086374488223762, best acc = 0.6123608430573443\n",
      "i = 16, mean acc = 0.6138834104039823, best acc = 0.6123608430573443\n",
      "i = 17, mean acc = 0.6094657991871971, best acc = 0.6138834104039823\n",
      "i = 18, mean acc = 0.6160903221844822, best acc = 0.6138834104039823\n",
      "i = 19, mean acc = 0.6116751613688111, best acc = 0.6160903221844822\n",
      "i = 20, mean acc = 0.6156816555985672, best acc = 0.6160903221844822\n",
      "i = 21, mean acc = 0.6120935315104749, best acc = 0.6160903221844822\n",
      "i = 22, mean acc = 0.6085044553609225, best acc = 0.6160903221844822\n",
      "i = 23, mean acc = 0.6116751660879535, best acc = 0.6160903221844822\n",
      "i = 24, mean acc = 0.6152623428815062, best acc = 0.6160903221844822\n",
      "10-fold CV complete \n",
      " best n = 18\n",
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.49394   0.59165   0.53840       551\n",
      "        mito    0.60839   0.65909   0.63273       264\n",
      "     nuclear    0.65009   0.58218   0.61426       651\n",
      "    secreted    0.83688   0.68406   0.75279       345\n",
      "\n",
      "    accuracy                        0.61568      1811\n",
      "   macro avg    0.64732   0.62925   0.63454      1811\n",
      "weighted avg    0.63208   0.61568   0.62026      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    "### KNN ###\n",
    "best_acc = 0\n",
    "best_neighbors = 0\n",
    "\n",
    "print(\"Running 10-fold CV\")\n",
    "for i in range(1,25):\n",
    "    clf = OneVsRestClassifier(KNeighborsClassifier(n_neighbors=i))\n",
    "    #clf_acc = cross_val_score(clf, X_reduced, y, cv=10)\n",
    "    clf_acc = cross_val_score(clf, X_train, y_train, cv=10)\n",
    "    print(\"i = {}, mean acc = {}, best acc = {}\".format(i,clf_acc.mean(), best_acc))\n",
    "    if clf_acc.mean() > best_acc:\n",
    "        best_acc = clf_acc.mean()\n",
    "        best_neighbors = i\n",
    "print(\"10-fold CV complete \\n best n = {}\".format(best_neighbors))\n",
    "\n",
    "#best_neighbors = 18\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_knn = OneVsRestClassifier(KNeighborsClassifier(best_neighbors))\n",
    "#clf_knn_acc = cross_val_score(clf_knn, X_reduced, y, cv=10)\n",
    "clf_knn.fit(X_train,y_train)\n",
    "y_pred_knn = clf_knn.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_knn, target_names=target_names,digits=5))\n",
    "\n",
    "\n",
    "#KNN_results = pd.DataFrame({'Model': 'KNN', 'Accuracy': clf_knn_acc.mean(), 'Acc_Std': clf_knn_acc.std()}, index=[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 299,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.51635   0.54446   0.53004       551\n",
      "        mito    0.64639   0.64394   0.64516       264\n",
      "     nuclear    0.62861   0.63441   0.63150       651\n",
      "    secreted    0.82903   0.74493   0.78473       345\n",
      "\n",
      "    accuracy                        0.62949      1811\n",
      "   macro avg    0.65510   0.64194   0.64786      1811\n",
      "weighted avg    0.63523   0.62949   0.63181      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    "### MLP ###\n",
    "\n",
    "clf_mlp = OneVsRestClassifier(MLPClassifier(solver='adam', activation = 'logistic', alpha=1e-5, hidden_layer_sizes=(100,100), random_state=2020))\n",
    "#clf_mlp_acc = cross_val_score(clf_mlp, X_reduced, y, cv=10)\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_mlp.fit(X_train,y_train)\n",
    "y_pred_mlp = clf_mlp.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_mlp, target_names=target_names,digits=5))\n",
    "\n",
    "\n",
    "#MLP_results = pd.DataFrame({'Model': 'MLP', 'Accuracy': clf_mlp_acc.mean(), 'Acc_Std': clf_mlp_acc.std()}, index=[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 341,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.50000   0.27042   0.35100       551\n",
      "        mito    0.58537   0.36364   0.44860       264\n",
      "     nuclear    0.58305   0.52842   0.55439       651\n",
      "    secreted    0.37022   0.81449   0.50906       345\n",
      "\n",
      "    accuracy                        0.48040      1811\n",
      "   macro avg    0.50966   0.49424   0.46576      1811\n",
      "weighted avg    0.51758   0.48040   0.46845      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    "### Decision Tree ###\n",
    "clf_dt = OneVsRestClassifier(tree.DecisionTreeClassifier(random_state=2020))\n",
    "#clf_dt_acc = cross_val_score(clf_dt, X_reduced, y, cv=10)\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_dt.fit(X_train,y_train)\n",
    "y_pred_dt = clf_dt.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_dt, target_names=target_names,digits=5))\n",
    "\n",
    "\n",
    "#DT_results = pd.DataFrame({'Model': 'Decision Tree', 'Accuracy': clf_dt_acc.mean(), 'Acc_Std': clf_dt_acc.std()}, index=[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 342,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running 10-fold CV\n",
      "max depth = None, mean acc = 0.6067020136130432, best acc = 0\n",
      "max depth = 1, mean acc = 0.48550140257381297, best acc = 0.6067020136130432\n",
      "max depth = 2, mean acc = 0.5035870031147793, best acc = 0.6067020136130432\n",
      "max depth = 3, mean acc = 0.5336815855799589, best acc = 0.6067020136130432\n",
      "max depth = 4, mean acc = 0.5662560941422757, best acc = 0.6067020136130432\n",
      "max depth = 5, mean acc = 0.5782704046775428, best acc = 0.6067020136130432\n",
      "max depth = 6, mean acc = 0.5920673866381063, best acc = 0.6067020136130432\n",
      "max depth = 7, mean acc = 0.5977349502833084, best acc = 0.6067020136130432\n",
      "max depth = 8, mean acc = 0.6087755609353964, best acc = 0.6067020136130432\n",
      "max depth = 9, mean acc = 0.6103082154999333, best acc = 0.6087755609353964\n",
      "max depth = 10, mean acc = 0.6138883522169257, best acc = 0.6103082154999333\n",
      "max depth = 11, mean acc = 0.6129218708538285, best acc = 0.6138883522169257\n",
      "max depth = 12, mean acc = 0.6234135935479583, best acc = 0.6138883522169257\n",
      "max depth = 13, mean acc = 0.6073964295589424, best acc = 0.6234135935479583\n",
      "max depth = 14, mean acc = 0.6038140129641036, best acc = 0.6234135935479583\n",
      "max depth = 15, mean acc = 0.6097412781058099, best acc = 0.6234135935479583\n",
      "max depth = 16, mean acc = 0.6098810969521633, best acc = 0.6234135935479583\n",
      "max depth = 17, mean acc = 0.6028486762938193, best acc = 0.6234135935479583\n",
      "max depth = 18, mean acc = 0.6051830460881614, best acc = 0.6234135935479583\n",
      "max depth = 19, mean acc = 0.6096103624959986, best acc = 0.6234135935479583\n",
      "max depth = 20, mean acc = 0.6025572115377015, best acc = 0.6234135935479583\n",
      "10-fold CV complete \n",
      " best max depth = 12\n",
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.50909   0.55898   0.53287       551\n",
      "        mito    0.65354   0.62879   0.64093       264\n",
      "     nuclear    0.61734   0.63441   0.62576       651\n",
      "    secreted    0.81979   0.67246   0.73885       345\n",
      "\n",
      "    accuracy                        0.61789      1811\n",
      "   macro avg    0.64994   0.62366   0.63460      1811\n",
      "weighted avg    0.62825   0.61789   0.62125      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    "### Random Forest ###\n",
    "best_acc = 0\n",
    "best_depth_idx = 0\n",
    "d_list = [None,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20]\n",
    "d_num = len(d_list)\n",
    "\n",
    "print(\"Running 10-fold CV\")\n",
    "for i in range(d_num):\n",
    "    clf_rf = OneVsRestClassifier(RandomForestClassifier(max_depth=d_list[i], random_state=2020))\n",
    "    #clf_rf_acc = cross_val_score(clf_rf, X_reduced, y, cv=10)\n",
    "    clf_rf_acc = cross_val_score(clf_rf, X_train, y_train, cv=10)\n",
    "    print(\"max depth = {}, mean acc = {}, best acc = {}\".format(d_list[i],clf_rf_acc.mean(), best_acc))\n",
    "    if clf_rf_acc.mean() > best_acc:\n",
    "        best_acc = clf_rf_acc.mean()\n",
    "        best_d_idx = i\n",
    "print(\"10-fold CV complete \\n best max depth = {}\".format(d_list[best_d_idx]))\n",
    "\n",
    "\n",
    "#best_depth_idx=12\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_rf = OneVsRestClassifier(RandomForestClassifier(max_depth=d_list[best_d_idx], random_state=2020))\n",
    "#clf_rf_acc = cross_val_score(clf_rf, X_reduced, y, cv=10)\n",
    "clf_rf.fit(X_train,y_train)\n",
    "y_pred_rf = clf_rf.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_rf, target_names=target_names,digits=5))\n",
    "\n",
    "\n",
    "#RF_results = pd.DataFrame({'Model': 'Random Forest', 'Accuracy': clf_rf_acc.mean(), 'Acc_Std': clf_rf_acc.std()}, index=[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 343,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running 10-fold CV\n",
      "max depth = None, min child weight = None, mean acc = 0.656391441650945, best acc = 0\n",
      "max depth = None, min child weight = 1, mean acc = 0.656391441650945, best acc = 0.656391441650945\n",
      "max depth = None, min child weight = 2, mean acc = 0.6558539829281335, best acc = 0.656391441650945\n",
      "max depth = None, min child weight = 3, mean acc = 0.6599932496427322, best acc = 0.656391441650945\n",
      "max depth = None, min child weight = 4, mean acc = 0.66799520674314, best acc = 0.6599932496427322\n",
      "max depth = None, min child weight = 5, mean acc = 0.660412383911928, best acc = 0.66799520674314\n",
      "max depth = None, min child weight = 6, mean acc = 0.6579233273905245, best acc = 0.66799520674314\n",
      "max depth = None, min child weight = 7, mean acc = 0.6557122536782527, best acc = 0.66799520674314\n",
      "max depth = 1, min child weight = None, mean acc = 0.6274225996753564, best acc = 0.66799520674314\n",
      "max depth = 1, min child weight = 1, mean acc = 0.6274225996753564, best acc = 0.66799520674314\n",
      "max depth = 1, min child weight = 2, mean acc = 0.6271471175816925, best acc = 0.66799520674314\n",
      "max depth = 1, min child weight = 3, mean acc = 0.626316861562891, best acc = 0.66799520674314\n",
      "max depth = 1, min child weight = 4, mean acc = 0.6263161005626323, best acc = 0.66799520674314\n",
      "max depth = 1, min child weight = 5, mean acc = 0.6263164810627616, best acc = 0.66799520674314\n",
      "max depth = 1, min child weight = 6, mean acc = 0.6264542189519162, best acc = 0.66799520674314\n",
      "max depth = 1, min child weight = 7, mean acc = 0.6264570735096797, best acc = 0.66799520674314\n",
      "max depth = 2, min child weight = None, mean acc = 0.6489480950407126, best acc = 0.66799520674314\n",
      "max depth = 2, min child weight = 1, mean acc = 0.6489480950407126, best acc = 0.66799520674314\n",
      "max depth = 2, min child weight = 2, mean acc = 0.6483918022770784, best acc = 0.66799520674314\n",
      "max depth = 2, min child weight = 3, mean acc = 0.64770557738116, best acc = 0.66799520674314\n",
      "max depth = 2, min child weight = 4, mean acc = 0.6486741349432228, best acc = 0.66799520674314\n",
      "max depth = 2, min child weight = 5, mean acc = 0.6482620470181641, best acc = 0.66799520674314\n",
      "max depth = 2, min child weight = 6, mean acc = 0.6490831552411302, best acc = 0.66799520674314\n",
      "max depth = 2, min child weight = 7, mean acc = 0.6482523371340305, best acc = 0.66799520674314\n",
      "max depth = 3, min child weight = None, mean acc = 0.6535032562605023, best acc = 0.66799520674314\n",
      "max depth = 3, min child weight = 1, mean acc = 0.6535032562605023, best acc = 0.66799520674314\n",
      "max depth = 3, min child weight = 2, mean acc = 0.6514329582142773, best acc = 0.66799520674314\n",
      "max depth = 3, min child weight = 3, mean acc = 0.6548844922333016, best acc = 0.66799520674314\n",
      "max depth = 3, min child weight = 4, mean acc = 0.6514310541130746, best acc = 0.66799520674314\n",
      "max depth = 3, min child weight = 5, mean acc = 0.6515786866322092, best acc = 0.66799520674314\n",
      "max depth = 3, min child weight = 6, mean acc = 0.6515769814733708, best acc = 0.66799520674314\n",
      "max depth = 3, min child weight = 7, mean acc = 0.6543243818116034, best acc = 0.66799520674314\n",
      "max depth = 4, min child weight = None, mean acc = 0.650325481830795, best acc = 0.66799520674314\n",
      "max depth = 4, min child weight = 1, mean acc = 0.650325481830795, best acc = 0.66799520674314\n",
      "max depth = 4, min child weight = 2, mean acc = 0.6572211182620002, best acc = 0.66799520674314\n",
      "max depth = 4, min child weight = 3, mean acc = 0.6556970257745409, best acc = 0.66799520674314\n",
      "max depth = 4, min child weight = 4, mean acc = 0.649637176012425, best acc = 0.66799520674314\n",
      "max depth = 4, min child weight = 5, mean acc = 0.6554356348467774, best acc = 0.66799520674314\n",
      "max depth = 4, min child weight = 6, mean acc = 0.6544712454406291, best acc = 0.66799520674314\n",
      "max depth = 4, min child weight = 7, mean acc = 0.6573706690267345, best acc = 0.66799520674314\n",
      "max depth = 5, min child weight = None, mean acc = 0.6569536377359544, best acc = 0.66799520674314\n",
      "max depth = 5, min child weight = 1, mean acc = 0.6569536377359544, best acc = 0.66799520674314\n",
      "max depth = 5, min child weight = 2, mean acc = 0.6555817484752943, best acc = 0.66799520674314\n",
      "max depth = 5, min child weight = 3, mean acc = 0.6573651273506365, best acc = 0.66799520674314\n",
      "max depth = 5, min child weight = 4, mean acc = 0.653359237720551, best acc = 0.66799520674314\n",
      "max depth = 5, min child weight = 5, mean acc = 0.6602661013346388, best acc = 0.66799520674314\n",
      "max depth = 5, min child weight = 6, mean acc = 0.653502301041365, best acc = 0.66799520674314\n",
      "max depth = 5, min child weight = 7, mean acc = 0.6546091775312884, best acc = 0.66799520674314\n",
      "max depth = 6, min child weight = None, mean acc = 0.656391441650945, best acc = 0.66799520674314\n",
      "max depth = 6, min child weight = 1, mean acc = 0.656391441650945, best acc = 0.66799520674314\n",
      "max depth = 6, min child weight = 2, mean acc = 0.6558539829281335, best acc = 0.66799520674314\n",
      "max depth = 6, min child weight = 3, mean acc = 0.6599932496427322, best acc = 0.66799520674314\n",
      "max depth = 6, min child weight = 4, mean acc = 0.66799520674314, best acc = 0.66799520674314\n",
      "max depth = 6, min child weight = 5, mean acc = 0.660412383911928, best acc = 0.66799520674314\n",
      "max depth = 6, min child weight = 6, mean acc = 0.6579233273905245, best acc = 0.66799520674314\n",
      "max depth = 6, min child weight = 7, mean acc = 0.6557122536782527, best acc = 0.66799520674314\n",
      "max depth = 7, min child weight = None, mean acc = 0.6619214537056171, best acc = 0.66799520674314\n",
      "max depth = 7, min child weight = 1, mean acc = 0.6619214537056171, best acc = 0.66799520674314\n",
      "max depth = 7, min child weight = 2, mean acc = 0.6575046050853031, best acc = 0.66799520674314\n",
      "max depth = 7, min child weight = 3, mean acc = 0.6673114306825271, best acc = 0.66799520674314\n",
      "max depth = 7, min child weight = 4, mean acc = 0.6565445953994564, best acc = 0.66799520674314\n",
      "max depth = 7, min child weight = 5, mean acc = 0.6559926744315261, best acc = 0.66799520674314\n",
      "max depth = 7, min child weight = 6, mean acc = 0.6591561477401886, best acc = 0.66799520674314\n",
      "max depth = 7, min child weight = 7, mean acc = 0.6575236301829837, best acc = 0.66799520674314\n",
      "10-fold CV complete \n",
      " best max depth = None, best min child weight = 4\n",
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.53782   0.58076   0.55846       551\n",
      "        mito    0.72653   0.67424   0.69941       264\n",
      "     nuclear    0.65571   0.64363   0.64961       651\n",
      "    secreted    0.82530   0.79420   0.80945       345\n",
      "\n",
      "    accuracy                        0.65765      1811\n",
      "   macro avg    0.68634   0.67321   0.67924      1811\n",
      "weighted avg    0.66247   0.65765   0.65959      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    "### XGBoost ###\n",
    "best_acc = 0\n",
    "best_depth_idx = 0\n",
    "d_list = [None,1,2,3,4,5,6,7]\n",
    "c_list = [None,1,2,3,4,5,6,7]\n",
    "d_num = len(d_list)\n",
    "c_num = len(c_list)\n",
    "\n",
    "print(\"Running 10-fold CV\")\n",
    "for i in range(d_num):\n",
    "    for j in range(c_num):\n",
    "        clf_xg = OneVsRestClassifier(xgb.XGBClassifier(max_depth=d_list[i],min_child_weight=c_list[j],random_state=2020))\n",
    "        #clf_xg_acc = cross_val_score(clf_xg, X_reduced, y, cv=10)\n",
    "        clf_xg_acc = cross_val_score(clf_xg, X_train, y_train, cv=10)\n",
    "        print(\"max depth = {}, min child weight = {}, mean acc = {}, best acc = {}\".format(d_list[i],c_list[j],clf_xg_acc.mean(), best_acc))\n",
    "        if clf_xg_acc.mean() > best_acc:\n",
    "            best_acc = clf_xg_acc.mean()\n",
    "            best_d_idx = i\n",
    "            best_c_idx = j\n",
    "print(\"10-fold CV complete \\n best max depth = {}, best min child weight = {}\".format(d_list[best_d_idx], c_list[best_c_idx]))\n",
    "\n",
    "#best_d_idx=0\n",
    "#best_c_idx=4\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_xg = OneVsRestClassifier(xgb.XGBClassifier(max_depth=d_list[best_d_idx],min_child_weight=c_list[best_c_idx],random_state=0))\n",
    "#clf_xg_acc = cross_val_score(clf_xg, X_reduced, y, cv=10)\n",
    "clf_xg.fit(X_train,y_train)\n",
    "y_pred_xg = clf_xg.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_xg, target_names=target_names,digits=5))\n",
    "\n",
    "#XG_results = pd.DataFrame({'Model': 'XG Boost', 'Accuracy': clf_xg_acc.mean(), 'Acc_Std': clf_xg_acc.std()}, index=[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 344,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running 10-fold CV\n",
      "num estimators = 100, mean_acc = 0.6138891100768809, best_acc = 0\n",
      "num estimators = 200, mean_acc = 0.6227243325321684, best_acc = 0.6138891100768809\n",
      "num estimators = 300, mean_acc = 0.6249304864657438, best_acc = 0.6227243325321684\n",
      "num estimators = 400, mean_acc = 0.6203793165458722, best_acc = 0.6249304864657438\n",
      "num estimators = 500, mean_acc = 0.619553833343464, best_acc = 0.6249304864657438\n",
      "10-fold CV complete \n",
      " best num estimators = 300\n",
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.52261   0.56624   0.54355       551\n",
      "        mito    0.66284   0.65530   0.65905       264\n",
      "     nuclear    0.63016   0.60983   0.61983       651\n",
      "    secreted    0.82353   0.77101   0.79641       345\n",
      "\n",
      "    accuracy                        0.63390      1811\n",
      "   macro avg    0.65978   0.65060   0.65471      1811\n",
      "weighted avg    0.63904   0.63390   0.63598      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    "### AdaBoost ###\n",
    "best_acc = 0\n",
    "best_E_idx = 0\n",
    "E_list = [100,200,300,400,500]\n",
    "E_num = len(E_list)\n",
    "\n",
    "print(\"Running 10-fold CV\")\n",
    "for i in range(E_num):\n",
    "    clf_ada = OneVsRestClassifier(AdaBoostClassifier(n_estimators=E_list[i],random_state=2020))\n",
    "    #clf_ada_acc = cross_val_score(clf_ada, X_reduced, y, cv=10)\n",
    "    clf_ada_acc = cross_val_score(clf_ada, X_train, y_train, cv=10)\n",
    "    print(\"num estimators = {}, mean_acc = {}, best_acc = {}\".format(E_list[i], clf_ada_acc.mean(),best_acc))\n",
    "    if clf_ada_acc.mean() > best_acc:\n",
    "        best_acc = clf_ada_acc.mean()\n",
    "        best_E_idx = i\n",
    "print(\"10-fold CV complete \\n best num estimators = {}\".format(E_list[best_E_idx]))\n",
    "\n",
    "#best_E_idx = 2\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_ada = OneVsRestClassifier(AdaBoostClassifier(n_estimators=E_list[best_E_idx],random_state=2020))\n",
    "#clf_ada_acc = cross_val_score(clf_ada, X_reduced, y, cv=10)\n",
    "clf_ada.fit(X_train,y_train)\n",
    "y_pred_ada = clf_ada.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_ada, target_names=target_names,digits=5))\n",
    "\n",
    "#Ada_results = pd.DataFrame({'Model': 'AdaBoost', 'Accuracy': clf_ada_acc.mean(), 'Acc_Std': clf_ada_acc.std()}, index=[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 345,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running 10-fold CV\n",
      "num estimators = 100, max depth = None, mean_acc = 0.5779983628299961, best_acc = 0\n",
      "num estimators = 100, max depth = 1, mean_acc = 0.5966313136008239, best_acc = 0.5779983628299961\n",
      "num estimators = 100, max depth = 2, mean_acc = 0.6294891069124624, best_acc = 0.5966313136008239\n",
      "num estimators = 100, max depth = 3, mean_acc = 0.6479819957652705, best_acc = 0.6294891069124624\n",
      "num estimators = 100, max depth = 4, mean_acc = 0.657224921732233, best_acc = 0.6479819957652705\n",
      "num estimators = 100, max depth = 5, mean_acc = 0.6512883539512246, best_acc = 0.657224921732233\n",
      "num estimators = 200, max depth = None, mean_acc = 0.5779983628299961, best_acc = 0.657224921732233\n",
      "num estimators = 200, max depth = 1, mean_acc = 0.6134810433975548, best_acc = 0.657224921732233\n",
      "num estimators = 200, max depth = 2, mean_acc = 0.6445291496834994, best_acc = 0.657224921732233\n",
      "num estimators = 200, max depth = 3, mean_acc = 0.6525346892471243, best_acc = 0.657224921732233\n",
      "num estimators = 200, max depth = 4, mean_acc = 0.6663424910632133, best_acc = 0.657224921732233\n",
      "num estimators = 200, max depth = 5, mean_acc = 0.6635728211224359, best_acc = 0.6663424910632133\n",
      "num estimators = 300, max depth = None, mean_acc = 0.5779983628299961, best_acc = 0.6663424910632133\n",
      "num estimators = 300, max depth = 1, mean_acc = 0.6259021085276842, best_acc = 0.6663424910632133\n",
      "num estimators = 300, max depth = 2, mean_acc = 0.6507425336508217, best_acc = 0.6663424910632133\n",
      "num estimators = 300, max depth = 3, mean_acc = 0.6552965565061835, best_acc = 0.6663424910632133\n",
      "num estimators = 300, max depth = 4, mean_acc = 0.6662066572361706, best_acc = 0.6663424910632133\n",
      "num estimators = 300, max depth = 5, mean_acc = 0.6628837353838033, best_acc = 0.6663424910632133\n",
      "10-fold CV complete \n",
      " best num estimators = 200, best max depth = 4\n",
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.55106   0.56806   0.55943       551\n",
      "        mito    0.69721   0.66288   0.67961       264\n",
      "     nuclear    0.65574   0.67588   0.66566       651\n",
      "    secreted    0.82866   0.77101   0.79880       345\n",
      "\n",
      "    accuracy                        0.65930      1811\n",
      "   macro avg    0.68317   0.66946   0.67587      1811\n",
      "weighted avg    0.66288   0.65930   0.66074      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    "### GradientBoost ###\n",
    "best_acc = 0\n",
    "best_E_idx = 0\n",
    "E_list = [100,200,300]\n",
    "E_num = len(E_list)\n",
    "d_list=[None,1,2,3,4,5]\n",
    "d_num=len(d_list)\n",
    "\n",
    "\n",
    "print(\"Running 10-fold CV\")\n",
    "for i in range(E_num):\n",
    "    for j in range(d_num):\n",
    "        clf_grad = OneVsRestClassifier(GradientBoostingClassifier(n_estimators=E_list[i],max_depth=d_list[j],random_state=2020))\n",
    "        #clf_grad_acc = cross_val_score(clf_grad, X_reduced, y, cv=10)\n",
    "        clf_grad_acc = cross_val_score(clf_grad, X_train, y_train, cv=10)\n",
    "        print(\"num estimators = {}, max depth = {}, mean_acc = {}, best_acc = {}\".format(E_list[i], d_list[j], clf_grad_acc.mean(),best_acc))\n",
    "        if clf_grad_acc.mean() > best_acc:\n",
    "            best_acc = clf_grad_acc.mean()\n",
    "            best_E_idx = i\n",
    "            best_d_idx = j\n",
    "print(\"10-fold CV complete \\n best num estimators = {}, best max depth = {}\".format(E_list[best_E_idx],d_list[best_d_idx]))\n",
    "\n",
    "#best_E_idx = 1\n",
    "#best_d_idx = 4\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_grad = OneVsRestClassifier(GradientBoostingClassifier(n_estimators=E_list[best_E_idx],max_depth=d_list[best_d_idx],random_state=2020))\n",
    "#clf_grad_acc = cross_val_score(clf_grad, X_reduced, y, cv=10)\n",
    "clf_grad.fit(X_train,y_train)\n",
    "y_pred_grad = clf_grad.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_grad, target_names=target_names,digits=5))\n",
    "\n",
    "#Grad_results = pd.DataFrame({'Model': 'GradBoost', 'Accuracy': clf_grad_acc.mean(), 'Acc_Std': clf_grad_acc.std()}, index=[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 346,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.45158   0.70236   0.54972       551\n",
      "        mito    0.72826   0.50758   0.59821       264\n",
      "     nuclear    0.67807   0.51767   0.58711       651\n",
      "    secreted    0.87179   0.68986   0.77023       345\n",
      "\n",
      "    accuracy                        0.60519      1811\n",
      "   macro avg    0.68242   0.60436   0.62632      1811\n",
      "weighted avg    0.65338   0.60519   0.61223      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    ",# weight ensemble in proportion to the reported accuracy achieved above \n",
    "#weights = [clf_svmr_acc.mean(),clf_svmp_acc.mean(),clf_lr_acc.mean(),clf_knn_acc.mean(),clf_mlp_acc.mean(),clf_dt_acc.mean(),clf_rf_acc.mean(),clf_xg_acc.mean(),clf_ada_acc.mean(),clf_grad_acc.mean()]\n",
    "\n",
    "weights = [66.3,61.9,63.4,62.0,63.2,46.8,62.1,66.0,63.6,66.1]\n",
    "clf_ensemble1 = VotingClassifier(estimators=[('SVM rbf', clf_svmr), ('SVM poly', clf_svmp), ('Log Reg', clf_lr), ('KNN', clf_knn),\n",
    "                                    ('MLP', clf_mlp), ('DT', clf_dt), ('RF', clf_rf), ('XG', clf_xg), ('Ada', clf_ada),\n",
    "                                    ('Grad', clf_grad)], voting='soft', weights=weights)\n",
    "#print(weights)\n",
    "#clf_ensemble_acc = cross_val_score(clf_ensemble, X_reduced, y, cv=10)\n",
    "#Ensemble_results = pd.DataFrame({'Model': 'Ensemble', 'Accuracy': clf_ensemble_acc.mean(), 'Acc_Std': clf_ensemble_acc.std()}, index=[0])\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_ensemble1.fit(X_train,y_train)\n",
    "y_pred_ensemble1 = clf_ensemble1.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_ensemble1, target_names=target_names,digits=5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 347,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.56217   0.58258   0.57219       551\n",
      "        mito    0.70000   0.68939   0.69466       264\n",
      "     nuclear    0.68105   0.67896   0.68000       651\n",
      "    secreted    0.84592   0.81159   0.82840       345\n",
      "\n",
      "    accuracy                        0.67642      1811\n",
      "   macro avg    0.69729   0.69063   0.69381      1811\n",
      "weighted avg    0.67905   0.67642   0.67761      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#weight ensemble in proportion to the reported accuracy achieved above \n",
    "#weights2 = [clf_svmr_acc.mean(),clf_lr_acc.mean(),clf_xg_acc.mean()]\n",
    "\n",
    "weights2 = [66.3,63.4,66.0]\n",
    "clf_ensemble2 = VotingClassifier(estimators=[('SVM rbf', clf_svmr), ('LogReg', clf_lr), ('XG', clf_xg)], voting='soft', weights=weights2)\n",
    "#print(weights2)\n",
    "#clf_ensemble2_acc = cross_val_score(clf_ensemble2, X_reduced, y, cv=10)\n",
    "#Ensemble2_results = pd.DataFrame({'Model': 'Ensemble2', 'Accuracy': clf_ensemble2_acc.mean(), 'Acc_Std': clf_ensemble2_acc.std()}, index=[0])\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_ensemble2.fit(X_train,y_train)\n",
    "y_pred_ensemble2 = clf_ensemble2.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_ensemble2, target_names=target_names,digits=5))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 348,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.55376   0.56080   0.55726       551\n",
      "        mito    0.70543   0.68939   0.69732       264\n",
      "     nuclear    0.66069   0.67896   0.66970       651\n",
      "    secreted    0.84049   0.79420   0.81669       345\n",
      "\n",
      "    accuracy                        0.66648      1811\n",
      "   macro avg    0.69009   0.68084   0.68524      1811\n",
      "weighted avg    0.66893   0.66648   0.66752      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# weight ensemble in proportion to the reported accuracy achieved above \n",
    "#weights3 = [clf_svmr_acc.mean(),clf_svmp_acc.mean(),clf_xg_acc.mean()]\n",
    "\n",
    "weights3 = [66.3,61.9,66.0]\n",
    "clf_ensemble3 = VotingClassifier(estimators=[('SVM rbf', clf_svmr), ('SVM poly', clf_svmp), ('XG', clf_xg)], voting='soft', weights=weights3)\n",
    "#print(weights3)\n",
    "#clf_ensemble3_acc = cross_val_score(clf_ensemble3, X_reduced, y, cv=10)\n",
    "#Ensemble3_results = pd.DataFrame({'Model': 'Ensemble3', 'Accuracy': clf_ensemble3_acc.mean(), 'Acc_Std': clf_ensemble3_acc.std()}, index=[0])\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_ensemble3.fit(X_train,y_train)\n",
    "y_pred_ensemble3 = clf_ensemble3.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_ensemble3, target_names=target_names,digits=5))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 349,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      " ---------- Predicting on test data ---------- \n",
      "\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        cyto    0.56703   0.56806   0.56754       551\n",
      "        mito    0.70115   0.69318   0.69714       264\n",
      "     nuclear    0.67062   0.69432   0.68226       651\n",
      "    secreted    0.84568   0.79420   0.81913       345\n",
      "\n",
      "    accuracy                        0.67477      1811\n",
      "   macro avg    0.69612   0.68744   0.69152      1811\n",
      "weighted avg    0.67690   0.67477   0.67560      1811\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# weight ensemble in proportion to the reported accuracy achieved above \n",
    "#weights4 = [clf_svmr_acc.mean(),clf_svmp_acc.mean(),clf_lr_acc.mean(),clf_xg_acc.mean()]\n",
    "\n",
    "weights4 = [66.3,61.9,63.4,66.0]\n",
    "clf_ensemble4 = VotingClassifier(estimators=[('SVM rbf', clf_svmr), ('SVM poly', clf_svmp), ('LogReg', clf_lr), ('XG', clf_xg)], voting='soft', weights=weights4)\n",
    "#print(weights4)\n",
    "#clf_ensemble4_acc = cross_val_score(clf_ensemble4, X_reduced, y, cv=10)\n",
    "#Ensemble4_results = pd.DataFrame({'Model': 'Ensemble4', 'Accuracy': clf_ensemble4_acc.mean(), 'Acc_Std': clf_ensemble4_acc.std()}, index=[0])\n",
    "\n",
    "print(\"\\n\\n ---------- Predicting on test data ---------- \\n\\n\")\n",
    "clf_ensemble4.fit(X_train,y_train)\n",
    "y_pred_ensemble4 = clf_ensemble4.predict(X_test)\n",
    "target_names = ['cyto', 'mito', 'nuclear', 'secreted']\n",
    "print(classification_report(y_test, y_pred_ensemble4, target_names=target_names,digits=5))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "### Concatenate results ###\n",
    "#df_results = pd.concat([SVMrbf_results,\n",
    "#                        SVMpoly_results,\n",
    "#                        LogReg_results,\n",
    "#                        KNN_results,\n",
    "#                        MLP_results,\n",
    "#                        DT_results,\n",
    "#                        RF_results,\n",
    "#                        XG_results,\n",
    "#                        Ada_results,\n",
    "#                        Grad_results,\n",
    "#                        Ensemble_results,\n",
    "#                        Ensemble2_results,\n",
    "#                        Ensemble3_results,\n",
    "#                        Ensemble4_results])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "#df_results_perc = df_results.copy()\n",
    "#df_results_perc['Accuracy'] = df_results_perc['Accuracy']*100\n",
    "#df_results_perc['Acc_Std'] = df_results_perc['Acc_Std']*100\n",
    "#df_results_perc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Mean = df_results['Accuracy'].values.tolist()\n",
    "#Std = df_results['Acc_Std'].values.tolist()\n",
    "#N = len(Mean)\n",
    "#ind = np.arange(N)    # the x locations for the groups\n",
    "#width = 0.35       # the width of the bars: can also be len(x) sequence\n",
    "\n",
    "#p1 = plt.bar(ind, Mean, width, yerr=Std)\n",
    "\n",
    "\n",
    "#plt.ylabel('Accuracy')\n",
    "#plt.xticks(ind, ('SVM (rbf)', 'SVM (poly)', 'Logistic Regression', 'kNN', 'MLP', 'Decision Tree', 'Random Forest',\n",
    "#                'XGBoost','AdaBoost','GradientBoost','Ensemble1','Ensemble2','Ensemble3','Ensemble4'),rotation='vertical')\n",
    "#plt.yticks(np.arange(0,0.8,step=0.1))\n",
    "\n",
    "#plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ROC-AUC Analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\nloc_dict = {}\\nloc_dict[0] = \\'cytosolic\\'\\nloc_dict[1] = \\'mitochondrial\\'\\nloc_dict[2] = \\'nuclear\\'\\nloc_dict[3] = \\'secreted\\'\\ndef roc_fn(X_train, X_test, y_train, y_test,location):\\n    y_test_bin = label_binarize(y_test, classes=[0, 1, 2, 3])\\n    rocaucs_list = []\\n    auc_list = []\\n    \\n    # list of classifiers\\n    clf_list = [clf_svmr, clf_svmp, clf_lr, clf_knn, clf_mlp, clf_dt, clf_rf, clf_xg, clf_ada, \\n                clf_grad,clf_ensemble,clf_ensemble2,clf_ensemble3,clf_ensemble4]\\n    \\n    # ROC AUC scores for given location\\n    for clf in clf_list:\\n        print(\"Running {}/{}  Current AUC List = {}\".format(idx,len(clf_list),auc_list))\\n        probas_ = clf.fit(X_train, y_train).predict_proba(X_test)\\n        fpr, tpr, thresholds = roc_curve(y_test_bin[:, 2], probas_[:, 2])\\n        roc_auc = auc(fpr, tpr)\\n        auc_list.append(roc_auc)\\n        rocaucs_list.append([fpr,tpr,roc_auc])\\n\\n    f, axes = plt.subplots(1, figsize=(10, 5),dpi=400)\\n    for [fpr, tpr, roc_auc],label in zip(rocaucs_list,[\\'SVM (rbf)\\', \\'SVM (poly)\\', \\'Logistic Regression\\', \\'kNN\\', \\'MLP\\', \\'Decision Tree\\', \\'Random Forest\\',\\n                \\'XGBoost\\',\\'AdaBoost\\',\\'GradientBoost\\',\\'Ensemble1\\',\\'Ensemble2\\',\\'Ensemble3\\',\\'Ensemble4\\']):\\n        lab = \\'%s AUC=%.4f\\' % (label,roc_auc)\\n        axes.step(fpr, tpr, lw=3,label=lab)\\n    axes.legend(loc=\\'lower right\\', fontsize=\\'xx-small\\')\\n    plt.title(\\'ROC AUC curve for {} class using different classifiers\\'.format(loc_dict[location]))\\n    plt.savefig(\\'rocNuc.png\\')\\n    return auc_list\\n'"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#from sklearn.metrics import roc_curve, auc\n",
    "#from sklearn.preprocessing import label_binarize\n",
    "\n",
    "\n",
    "# location as follows\n",
    "# 'cystolic': 0, 'mitochondrial': 1, 'nuclear': 2, 'secreted': 3\n",
    "\"\"\"\n",
    "loc_dict = {}\n",
    "loc_dict[0] = 'cytosolic'\n",
    "loc_dict[1] = 'mitochondrial'\n",
    "loc_dict[2] = 'nuclear'\n",
    "loc_dict[3] = 'secreted'\n",
    "def roc_fn(X_train, X_test, y_train, y_test,location):\n",
    "    y_test_bin = label_binarize(y_test, classes=[0, 1, 2, 3])\n",
    "    rocaucs_list = []\n",
    "    auc_list = []\n",
    "    \n",
    "    # list of classifiers\n",
    "    clf_list = [clf_svmr, clf_svmp, clf_lr, clf_knn, clf_mlp, clf_dt, clf_rf, clf_xg, clf_ada, \n",
    "                clf_grad,clf_ensemble,clf_ensemble2,clf_ensemble3,clf_ensemble4]\n",
    "    \n",
    "    # ROC AUC scores for given location\n",
    "    for clf in clf_list:\n",
    "        print(\"Running {}/{}  Current AUC List = {}\".format(idx,len(clf_list),auc_list))\n",
    "        probas_ = clf.fit(X_train, y_train).predict_proba(X_test)\n",
    "        fpr, tpr, thresholds = roc_curve(y_test_bin[:, 2], probas_[:, 2])\n",
    "        roc_auc = auc(fpr, tpr)\n",
    "        auc_list.append(roc_auc)\n",
    "        rocaucs_list.append([fpr,tpr,roc_auc])\n",
    "\n",
    "    f, axes = plt.subplots(1, figsize=(10, 5),dpi=400)\n",
    "    for [fpr, tpr, roc_auc],label in zip(rocaucs_list,['SVM (rbf)', 'SVM (poly)', 'Logistic Regression', 'kNN', 'MLP', 'Decision Tree', 'Random Forest',\n",
    "                'XGBoost','AdaBoost','GradientBoost','Ensemble1','Ensemble2','Ensemble3','Ensemble4']):\n",
    "        lab = '%s AUC=%.4f' % (label,roc_auc)\n",
    "        axes.step(fpr, tpr, lw=3,label=lab)\n",
    "    axes.legend(loc='lower right', fontsize='xx-small')\n",
    "    plt.title('ROC AUC curve for {} class using different classifiers'.format(loc_dict[location]))\n",
    "    plt.savefig('rocNuc.png')\n",
    "    return auc_list\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\nfrom sklearn.linear_model import RidgeClassifier\\nfrom sklearn.model_selection import train_test_split\\nfrom sklearn.preprocessing import OrdinalEncoder, LabelEncoder\\n\\nfrom yellowbrick.classifier import ROCAUC\\nfrom yellowbrick.datasets import load_game\\n\\n# Encode the non-numeric columns\\nX_yb=X_reduced\\ny_yb = y\\n\\n# Create the train and test data\\nX_ybtrain, X_ybtest, y_ybtrain, y_ybtest = train_test_split(X_yb, y_yb, random_state=2020)\\n\\n# Instaniate the classification model and visualizer\\nmodel = clf_ensemble3\\nvisualizer = ROCAUC(model, classes=[\"cyto\",\"mito\",\"nuc\",\"sec\"])\\n\\nvisualizer.fit(X_ybtrain, y_ybtrain)        # Fit the training data to the visualizer\\nvisualizer.score(X_ybtest, y_ybtest)        # Evaluate the model on the test data\\nvisualizer.show()                       # Finalize and render the figure\\n'"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"\"\"\n",
    "from sklearn.linear_model import RidgeClassifier\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import OrdinalEncoder, LabelEncoder\n",
    "\n",
    "from yellowbrick.classifier import ROCAUC\n",
    "from yellowbrick.datasets import load_game\n",
    "\n",
    "# Encode the non-numeric columns\n",
    "X_yb=X_reduced\n",
    "y_yb = y\n",
    "\n",
    "# Create the train and test data\n",
    "X_ybtrain, X_ybtest, y_ybtrain, y_ybtest = train_test_split(X_yb, y_yb, random_state=2020)\n",
    "\n",
    "# Instaniate the classification model and visualizer\n",
    "model = clf_ensemble3\n",
    "visualizer = ROCAUC(model, classes=[\"cyto\",\"mito\",\"nuc\",\"sec\"])\n",
    "\n",
    "visualizer.fit(X_ybtrain, y_ybtrain)        # Fit the training data to the visualizer\n",
    "visualizer.score(X_ybtest, y_ybtest)        # Evaluate the model on the test data\n",
    "visualizer.show()                       # Finalize and render the figure\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "# cytosolic\n",
    "#auc0 = roc_fn(X_train, X_test, y_train, y_test,0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "# mitochondrial\n",
    "#auc1 = roc_fn(X_train, X_test, y_train, y_test,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "# nuclear\n",
    "#auc2 = roc_fn(X_train, X_test, y_train, y_test,2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "# secreted\n",
    "#auc3 = roc_fn(X_train, X_test, y_train, y_test,3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "#auc_array = np.zeros((14,4))\n",
    "#auc_array[:,0] = auc0\n",
    "#auc_array[:,1] = auc1\n",
    "#auc_array[:,2] = auc2\n",
    "#auc_array[:,3] = auc3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "#auc_array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"\\nclf_names = ['SVM (rbf)', 'SVM (poly)', 'Logistic Regression', 'kNN', 'MLP', 'Decision Tree', 'Random Forest',\\n                'XGBoost','AdaBoost','GradientBoost','Ensemble1','Ensemble2','Ensemble3','Ensemble4']\\nauc_mean = np.mean(auc_array,1)\\nd = {}\\nfor i in range(len(auc_mean)):\\n    d[clf_names[i]] = auc_mean[i]\\n\\ndf_auc = pd.DataFrame.from_dict(d, orient='index')\\ndf_auc.columns = ['Mean AUC']\\ndf_auc\\n\""
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"\"\"\n",
    "clf_names = ['SVM (rbf)', 'SVM (poly)', 'Logistic Regression', 'kNN', 'MLP', 'Decision Tree', 'Random Forest',\n",
    "                'XGBoost','AdaBoost','GradientBoost','Ensemble1','Ensemble2','Ensemble3','Ensemble4']\n",
    "auc_mean = np.mean(auc_array,1)\n",
    "d = {}\n",
    "for i in range(len(auc_mean)):\n",
    "    d[clf_names[i]] = auc_mean[i]\n",
    "\n",
    "df_auc = pd.DataFrame.from_dict(d, orient='index')\n",
    "df_auc.columns = ['Mean AUC']\n",
    "df_auc\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Confusion Matrix on 80/20 train/test split "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 350,
   "metadata": {},
   "outputs": [],
   "source": [
    "### Plot Confusion Matrix ###\n",
    "def plot_confusion_matrix(cm, classes,\n",
    "                          normalize=False,\n",
    "                          title='Confusion matrix',\n",
    "                          cmap=plt.cm.Blues):\n",
    "    \"\"\"\n",
    "    This function prints and plots the confusion matrix.\n",
    "    Normalization can be applied by setting `normalize=True`.\n",
    "    \"\"\"\n",
    "    if normalize:\n",
    "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "        print(\"Normalized confusion matrix\")\n",
    "    else:\n",
    "        print('Confusion matrix, without normalization')\n",
    "\n",
    "    print(cm)\n",
    "\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation=45)\n",
    "    plt.yticks(tick_marks, classes)\n",
    "\n",
    "    fmt = '.2f' if normalize else 'd'\n",
    "    thresh = cm.max() / 2.\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, format(cm[i, j], fmt),\n",
    "                 horizontalalignment=\"center\",\n",
    "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "    \n",
    "    b, t = plt.ylim() # discover the values for bottom and top\n",
    "    b += 0.5 # Add 0.5 to the bottom\n",
    "    t -= 0.5 # Subtract 0.5 from the top\n",
    "    plt.ylim(b, t) # update the ylim(bottom, top) values\n",
    "\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "    plt.tight_layout()\n",
    "    \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CV Confusion Matrix Of Best Model on Training Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 351,
   "metadata": {},
   "outputs": [],
   "source": [
    "# insert relevatn y_pred into second argument below\n",
    "\n",
    "cm = np.asarray(confusion_matrix(y_test, y_pred_ensemble2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 352,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x396 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### Confusion Matrix ###\n",
    "num_samples = len(y_test)\n",
    "names = ['cyto', 'mito', 'nuc', 'sec']\n",
    "df_cm = pd.DataFrame(cm/num_samples, names, names)\n",
    "# plt.figure(figsize=(10,7))\n",
    "sns.set(font_scale=1.4) # for label size\n",
    "sns.heatmap(df_cm, annot=True, annot_kws={\"size\": 16},cmap='Blues') # font size\n",
    "\n",
    "# fix for mpl bug that cuts off top/bottom of seaborn viz\n",
    "b, t = plt.ylim() # discover the values for bottom and top\n",
    "b += 0.5 # Add 0.5 to the bottom\n",
    "t -= 0.5 # Subtract 0.5 from the top\n",
    "plt.ylim(b, t) # update the ylim(bottom, top) values\n",
    "\n",
    "plt.xlabel('Predicted Label')\n",
    "plt.ylabel('True Label')\n",
    "plt.savefig('cm.png')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 353,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Normalized confusion matrix\n",
      "[[0.58 0.06 0.31 0.05]\n",
      " [0.18 0.69 0.08 0.05]\n",
      " [0.25 0.05 0.68 0.01]\n",
      " [0.11 0.03 0.05 0.81]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix =  confusion_matrix(y_test, y_pred_ensemble2)\n",
    "np.set_printoptions(precision=2)\n",
    "plt.figure(figsize=(7,7))\n",
    "plot_confusion_matrix(cnf_matrix, classes=['cyto','mito','nuc','sec'], normalize=True, title='Row Normalized Confusion matrix')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 354,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix, without normalization\n",
      "[[321  32 169  29]\n",
      " [ 47 182  22  13]\n",
      " [165  35 442   9]\n",
      " [ 38  11  16 280]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix =  confusion_matrix(y_test, y_pred_ensemble2)\n",
    "np.set_printoptions(precision=2)\n",
    "plt.figure(figsize=(7,7))\n",
    "plot_confusion_matrix(cnf_matrix, classes=['cyto','mito','nuc','sec'], normalize=False, title='Confusion matrix')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train Best Model From Above on Full Training Set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 355,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "VotingClassifier(estimators=[('SVM rbf',\n",
       "                              SVC(C=1.0, cache_size=200,\n",
       "                                  class_weight='balanced', coef0=0.0,\n",
       "                                  decision_function_shape='ovr', degree=3,\n",
       "                                  gamma='auto_deprecated', kernel='rbf',\n",
       "                                  max_iter=-1, probability=True,\n",
       "                                  random_state=None, shrinking=True, tol=0.001,\n",
       "                                  verbose=False)),\n",
       "                             ('LogReg',\n",
       "                              LogisticRegression(C=1.0, class_weight=None,\n",
       "                                                 dual=False, fit_intercept=True,\n",
       "                                                 inter...\n",
       "                                                                          min_child_weight=4,\n",
       "                                                                          missing=nan,\n",
       "                                                                          monotone_constraints=None,\n",
       "                                                                          n_estimators=100,\n",
       "                                                                          n_jobs=None,\n",
       "                                                                          num_parallel_tree=None,\n",
       "                                                                          objective='binary:logistic',\n",
       "                                                                          random_state=0,\n",
       "                                                                          reg_alpha=None,\n",
       "                                                                          reg_lambda=None,\n",
       "                                                                          scale_pos_weight=None,\n",
       "                                                                          subsample=None,\n",
       "                                                                          tree_method=None,\n",
       "                                                                          validate_parameters=False,\n",
       "                                                                          verbosity=None),\n",
       "                                                  n_jobs=None))],\n",
       "                 flatten_transform=True, n_jobs=None, voting='soft',\n",
       "                 weights=[66.3, 63.4, 66.0])"
      ]
     },
     "execution_count": 355,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "### Train best model from above ###\n",
    "\n",
    "clf = clf_ensemble2\n",
    "\n",
    "clf.fit(X_reduced, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Predict on Blind Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 356,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_blind_data(text_file):\n",
    "    df = pd.DataFrame(columns=('ID',  \n",
    "                               'Sequence_Length', \n",
    "                               'Molar_Extinction_Coefficient',\n",
    "                               'Isoelectric_Point', \n",
    "                               'Aromaticity',\n",
    "                               'Secondary_Structure_Helix',\n",
    "                               'Secondary_Structure_Turn',\n",
    "                               'Secondary_Structure_Sheet',\n",
    "                               'Mean_Flexibility',\n",
    "                               'Gravy',\n",
    "                               'Molecular_Weight',\n",
    "                               'Instability_Index',\n",
    "                               'Ag', 'Cg', 'Dg', 'Eg', 'Fg',\n",
    "                               'Gg', 'Hg', 'Ig', 'Kg', 'Lg',\n",
    "                               'Mg', 'Ng', 'Pg', 'Qg', 'Rg',\n",
    "                               'Sg', 'Tg', 'Vg', 'Wg', 'Yg',\n",
    "                               'As', 'Cs', 'Ds', 'Es', 'Fs',\n",
    "                               'Gs', 'Hs', 'Is', 'Ks', 'Ls',\n",
    "                               'Ms', 'Ns', 'Ps', 'Qs', 'Rs',\n",
    "                               'Ss', 'Ts', 'Vs', 'Ws', 'Ys',\n",
    "                               'Ae', 'Ce', 'De', 'Ee', 'Fe',\n",
    "                               'Ge', 'He', 'Ie', 'Ke', 'Le',\n",
    "                               'Me', 'Ne', 'Pe', 'Qe', 'Re',\n",
    "                               'Se', 'Te', 'Ve', 'We', 'Ye'))\n",
    " \n",
    "    i=0 \n",
    "    ID_list = []\n",
    "    for seq_record in SeqIO.parse(text_file, \"fasta\"):\n",
    "        # get sequence id\n",
    "        seq_id = seq_record.id\n",
    "        ID_list.append(seq_id)\n",
    "        # get species\n",
    "        #_, species = seq_id.split(\"_\", 1)\n",
    "        # get sequence length\n",
    "        seq_len = len(seq_record)\n",
    "        \n",
    "        \n",
    "        seq_rep = str(seq_record.seq)\n",
    "        \n",
    "        # locate offending characters\n",
    "        B_list = [pos for pos, char in enumerate(seq_rep) if char == 'B']\n",
    "        Z_list = [pos for pos, char in enumerate(seq_rep) if char == 'Z']\n",
    "        X_list = [pos for pos, char in enumerate(seq_rep) if char == 'X']\n",
    "        \n",
    "        # replace \"B\" by asparagine (\"N\")/aspartic acid (\"D\")\n",
    "        if len(B_list) != 0:\n",
    "            print(\"B found at index {} in example {} of {}\".format(B_list,i,text_file))\n",
    "            for j in range(len(B_list)):\n",
    "                flip = random.randint(0, 1)\n",
    "                if flip == 1:\n",
    "                    seq_rep = seq_rep[:B_list[j]] + 'N' + seq_rep[B_list[j]+1:]\n",
    "                else:\n",
    "                    seq_rep = seq_rep[:B_list[j]] + 'D' + seq_rep[B_list[j]+1:]\n",
    "        # replace \"Z\" by glutamine (\"Q\")/glutamic acid (\"E\")\n",
    "        if len(Z_list) != 0:\n",
    "            print(\"Z found at index {} in example {} of {}\".format(Z_list,i,text_file))\n",
    "            for j in range(len(Z_list)):\n",
    "                flip = random.randint(0, 1)\n",
    "                if flip == 1:\n",
    "                    seq_rep = seq_rep[:Z_list[j]] + 'Q' + seq_rep[Z_list[j]+1:]\n",
    "                else:\n",
    "                    seq_rep = seq_rep[:Z_list[j]] + 'E' + seq_rep[Z_list[j]+1:]\n",
    "        \n",
    "        # replace \"U\" by cysteine (\"C\")\n",
    "        if len(B_list) != 0:\n",
    "            print(\"U found at index {} in example {} of {}\".format(U_list,i,text_file))\n",
    "            for j in range(len(U_list)):\n",
    "                seq_rep = seq_rep[:U_list[j]] + 'C' + seq_rep[U_list[j]+1:]\n",
    "                \n",
    "        # replace \"X\" by a random aminoa acid\n",
    "        AA_alphabet = ['A','C','D','E','F','G','H','I','K','L','M','N','P','Q','R','S','T','V','W','Y']\n",
    "        if len(X_list) != 0:\n",
    "            print(\"X found at index {} in example {} of {}\".format(X_list,i,text_file))\n",
    "            for j in range(len(Z_list)):\n",
    "                flip = random.randint(0, 19)\n",
    "                seq_rep = seq_rep[:X_list[j]] + AA_alphabet[flip] + seq_rep[X_list[j]+1:]\n",
    "        \n",
    "        \n",
    "        \n",
    "        # get analysed sequence and global count       \n",
    "        analysed_seq = ProteinAnalysis(seq_rep)\n",
    "        global_count = analysed_seq.count_amino_acids()\n",
    "        \n",
    "        # get N and C terminal amino acid sequences and counts\n",
    "        analysed_start = ProteinAnalysis(seq_rep[:50])\n",
    "        analysed_end = ProteinAnalysis(seq_rep[-50:])\n",
    "        aa_start = analysed_start.count_amino_acids()\n",
    "        aa_end = analysed_end.count_amino_acids()  \n",
    "        \n",
    "        \n",
    "        # molar extinction coefficient\n",
    "        molar = np.mean(analysed_seq.molar_extinction_coefficient())\n",
    "\n",
    "        \n",
    "        # isoelectric point\n",
    "        isoelectric = analysed_seq.isoelectric_point() \n",
    "        # aromaticity\n",
    "        aromaticity = analysed_seq.aromaticity()\n",
    "        # secondary structure fraction\n",
    "        helix, turn, sheet = analysed_seq.secondary_structure_fraction()    \n",
    "        \n",
    "        \"\"\"Calculations below only possible if sequence doesn't have X so use try-except to set to Nan if there is an X\"\"\"   \n",
    "                # gravy (grand average hydropathy)\n",
    "        try:\n",
    "            mean_flexibility = np.mean(analysed_seq.flexibility())\n",
    "        except:\n",
    "            mean_flexibility = float('NaN')\n",
    "        # gravy (grand average hydropathy)\n",
    "        try:\n",
    "            gravy = analysed_seq.gravy()\n",
    "        except:\n",
    "            gravy = float('NaN')\n",
    "        # molecular weight\n",
    "        try:\n",
    "            molecular_weight = analysed_seq.molecular_weight()\n",
    "        except:\n",
    "            molecular_weight = float('NaN')\n",
    "        # instability index\n",
    "        try:\n",
    "            instability_idx = analysed_seq.instability_index()\n",
    "        except:\n",
    "            instability_idx = float('NaN')\n",
    "            \n",
    "        \n",
    "\n",
    "        iso_s = analysed_start.isoelectric_point()         \n",
    "        df.loc[i] = [seq_id,\n",
    "                  seq_len,\n",
    "                  molar,\n",
    "                  isoelectric,\n",
    "                  aromaticity,\n",
    "                  helix,\n",
    "                  turn,\n",
    "                  sheet,\n",
    "                  mean_flexibility,\n",
    "                  gravy,\n",
    "                  molecular_weight,\n",
    "                  instability_idx,\n",
    "                  global_count['A'],global_count['C'],global_count['D'],global_count['E'],global_count['F'],\n",
    "                  global_count['G'],global_count['H'],global_count['I'],global_count['K'],global_count['L'],\n",
    "                  global_count['M'],global_count['N'],global_count['P'],global_count['Q'],global_count['R'], \n",
    "                  global_count['S'],global_count['T'],global_count['V'],global_count['W'],global_count['Y'], \n",
    "                  aa_start['A'],aa_start['C'],aa_start['D'],aa_start['E'],aa_start['F'],\n",
    "                  aa_start['G'],aa_start['H'],aa_start['I'],aa_start['K'],aa_start['L'],\n",
    "                  aa_start['M'],aa_start['N'],aa_start['P'],aa_start['Q'],aa_start['R'], \n",
    "                  aa_start['S'],aa_start['T'],aa_start['V'],aa_start['W'],aa_start['Y'], \n",
    "                  aa_end['A'],aa_end['C'],aa_end['D'],aa_end['E'],aa_end['F'],\n",
    "                  aa_end['G'],aa_end['H'],aa_end['I'],aa_end['K'],aa_end['L'],\n",
    "                  aa_end['M'],aa_end['N'],aa_end['P'],aa_end['Q'],aa_end['R'], \n",
    "                  aa_end['S'],aa_end['T'],aa_end['V'],aa_end['W'],aa_end['Y']]\n",
    "                     \n",
    "\n",
    "        i+=1\n",
    "        \n",
    "        # ensure the sequence length column is of dtype int\n",
    "        df.Sequence_Length = df.Sequence_Length.astype(int)\n",
    "    print(i)\n",
    "    return df, ID_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 357,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20\n"
     ]
    }
   ],
   "source": [
    "# Load blind data\n",
    "df_blind, ID_list = load_blind_data('blind.txt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 358,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-----Missing Features Before Imputation------\n",
      "ID                              0\n",
      "Sequence_Length                 0\n",
      "Molar_Extinction_Coefficient    0\n",
      "Isoelectric_Point               0\n",
      "Aromaticity                     0\n",
      "                               ..\n",
      "Se                              0\n",
      "Te                              0\n",
      "Ve                              0\n",
      "We                              0\n",
      "Ye                              0\n",
      "Length: 72, dtype: int64\n",
      "-----Missing Features After Imputation------\n",
      "ID                              0\n",
      "Sequence_Length                 0\n",
      "Molar_Extinction_Coefficient    0\n",
      "Isoelectric_Point               0\n",
      "Aromaticity                     0\n",
      "                               ..\n",
      "Se                              0\n",
      "Te                              0\n",
      "Ve                              0\n",
      "We                              0\n",
      "Ye                              0\n",
      "Length: 72, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "categorical_cols = ['ID']\n",
    "df_blind = preprocess_data(df_blind,categorical_cols,seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 359,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>Sequence_Length</th>\n",
       "      <th>Molar_Extinction_Coefficient</th>\n",
       "      <th>Isoelectric_Point</th>\n",
       "      <th>Aromaticity</th>\n",
       "      <th>Secondary_Structure_Helix</th>\n",
       "      <th>Secondary_Structure_Turn</th>\n",
       "      <th>Secondary_Structure_Sheet</th>\n",
       "      <th>Mean_Flexibility</th>\n",
       "      <th>Gravy</th>\n",
       "      <th>...</th>\n",
       "      <th>Me</th>\n",
       "      <th>Ne</th>\n",
       "      <th>Pe</th>\n",
       "      <th>Qe</th>\n",
       "      <th>Re</th>\n",
       "      <th>Se</th>\n",
       "      <th>Te</th>\n",
       "      <th>Ve</th>\n",
       "      <th>We</th>\n",
       "      <th>Ye</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>-0.433239</td>\n",
       "      <td>-0.515963</td>\n",
       "      <td>1.151919</td>\n",
       "      <td>-0.868188</td>\n",
       "      <td>-0.410363</td>\n",
       "      <td>-0.268194</td>\n",
       "      <td>0.781389</td>\n",
       "      <td>0.224094</td>\n",
       "      <td>0.259319</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>0.048280</td>\n",
       "      <td>-0.330175</td>\n",
       "      <td>0.475002</td>\n",
       "      <td>-0.591843</td>\n",
       "      <td>0.271484</td>\n",
       "      <td>-0.902828</td>\n",
       "      <td>-0.816994</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>1.128152</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>-0.838955</td>\n",
       "      <td>-0.530375</td>\n",
       "      <td>-1.182085</td>\n",
       "      <td>0.560559</td>\n",
       "      <td>-0.100613</td>\n",
       "      <td>-0.706219</td>\n",
       "      <td>1.798115</td>\n",
       "      <td>-0.708992</td>\n",
       "      <td>1.006882</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>-0.917329</td>\n",
       "      <td>-0.718616</td>\n",
       "      <td>-1.108338</td>\n",
       "      <td>1.099137</td>\n",
       "      <td>-1.538407</td>\n",
       "      <td>0.844581</td>\n",
       "      <td>0.949480</td>\n",
       "      <td>1.069045</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>10</td>\n",
       "      <td>-0.729815</td>\n",
       "      <td>-0.207026</td>\n",
       "      <td>1.225676</td>\n",
       "      <td>0.683511</td>\n",
       "      <td>-0.356537</td>\n",
       "      <td>1.115924</td>\n",
       "      <td>-1.284450</td>\n",
       "      <td>-0.520519</td>\n",
       "      <td>-0.100817</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>1.496694</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>-0.052778</td>\n",
       "      <td>-0.591843</td>\n",
       "      <td>0.271484</td>\n",
       "      <td>0.262111</td>\n",
       "      <td>-1.258612</td>\n",
       "      <td>0.178174</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.390532</td>\n",
       "      <td>0.584023</td>\n",
       "      <td>-0.783093</td>\n",
       "      <td>2.682023</td>\n",
       "      <td>1.348737</td>\n",
       "      <td>-1.528675</td>\n",
       "      <td>-0.038853</td>\n",
       "      <td>-1.605862</td>\n",
       "      <td>-0.037508</td>\n",
       "      <td>...</td>\n",
       "      <td>0.164895</td>\n",
       "      <td>-0.434524</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>1.530561</td>\n",
       "      <td>0.535477</td>\n",
       "      <td>-0.935110</td>\n",
       "      <td>-0.320358</td>\n",
       "      <td>-0.816994</td>\n",
       "      <td>1.069045</td>\n",
       "      <td>1.128152</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>13</td>\n",
       "      <td>0.413783</td>\n",
       "      <td>-0.425000</td>\n",
       "      <td>-1.115693</td>\n",
       "      <td>-1.055700</td>\n",
       "      <td>-0.904704</td>\n",
       "      <td>0.911975</td>\n",
       "      <td>-0.171175</td>\n",
       "      <td>1.234026</td>\n",
       "      <td>-0.803807</td>\n",
       "      <td>...</td>\n",
       "      <td>1.264194</td>\n",
       "      <td>1.496694</td>\n",
       "      <td>0.835148</td>\n",
       "      <td>1.530561</td>\n",
       "      <td>-0.591843</td>\n",
       "      <td>-0.331813</td>\n",
       "      <td>-0.902828</td>\n",
       "      <td>-0.375376</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-1.289317</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>-0.516280</td>\n",
       "      <td>-0.765299</td>\n",
       "      <td>-0.359976</td>\n",
       "      <td>-1.306744</td>\n",
       "      <td>-1.880121</td>\n",
       "      <td>1.480231</td>\n",
       "      <td>-0.969447</td>\n",
       "      <td>1.361296</td>\n",
       "      <td>-2.034525</td>\n",
       "      <td>...</td>\n",
       "      <td>0.164895</td>\n",
       "      <td>0.048280</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>2.058341</td>\n",
       "      <td>0.535477</td>\n",
       "      <td>-0.331813</td>\n",
       "      <td>-1.485297</td>\n",
       "      <td>0.507861</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>4</td>\n",
       "      <td>-0.141408</td>\n",
       "      <td>0.350531</td>\n",
       "      <td>-0.018720</td>\n",
       "      <td>0.136607</td>\n",
       "      <td>-0.132222</td>\n",
       "      <td>1.233746</td>\n",
       "      <td>-0.862837</td>\n",
       "      <td>-0.348423</td>\n",
       "      <td>-0.013456</td>\n",
       "      <td>...</td>\n",
       "      <td>0.164895</td>\n",
       "      <td>0.048280</td>\n",
       "      <td>-0.718616</td>\n",
       "      <td>-1.108338</td>\n",
       "      <td>-1.155503</td>\n",
       "      <td>-0.935110</td>\n",
       "      <td>-0.320358</td>\n",
       "      <td>2.274335</td>\n",
       "      <td>2.850787</td>\n",
       "      <td>1.933975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>16</td>\n",
       "      <td>-0.962331</td>\n",
       "      <td>-0.855449</td>\n",
       "      <td>0.792799</td>\n",
       "      <td>-0.558551</td>\n",
       "      <td>-0.823612</td>\n",
       "      <td>0.148896</td>\n",
       "      <td>0.579369</td>\n",
       "      <td>0.024470</td>\n",
       "      <td>-0.076215</td>\n",
       "      <td>...</td>\n",
       "      <td>2.363494</td>\n",
       "      <td>-0.434524</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>-0.052778</td>\n",
       "      <td>1.662797</td>\n",
       "      <td>-0.331813</td>\n",
       "      <td>-0.320358</td>\n",
       "      <td>-1.258612</td>\n",
       "      <td>0.178174</td>\n",
       "      <td>0.322329</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>19</td>\n",
       "      <td>-0.952841</td>\n",
       "      <td>-0.854636</td>\n",
       "      <td>-1.709162</td>\n",
       "      <td>-1.227487</td>\n",
       "      <td>-1.908166</td>\n",
       "      <td>0.314833</td>\n",
       "      <td>-2.034345</td>\n",
       "      <td>2.333899</td>\n",
       "      <td>-2.493635</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>-0.917329</td>\n",
       "      <td>-0.718616</td>\n",
       "      <td>0.475002</td>\n",
       "      <td>0.535477</td>\n",
       "      <td>0.874781</td>\n",
       "      <td>-0.902828</td>\n",
       "      <td>0.066243</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>11</td>\n",
       "      <td>0.537159</td>\n",
       "      <td>0.966453</td>\n",
       "      <td>1.560026</td>\n",
       "      <td>0.671040</td>\n",
       "      <td>0.630263</td>\n",
       "      <td>-0.712784</td>\n",
       "      <td>0.097624</td>\n",
       "      <td>-0.687514</td>\n",
       "      <td>0.308330</td>\n",
       "      <td>...</td>\n",
       "      <td>0.164895</td>\n",
       "      <td>0.048280</td>\n",
       "      <td>-0.330175</td>\n",
       "      <td>-1.108338</td>\n",
       "      <td>-0.028183</td>\n",
       "      <td>1.478078</td>\n",
       "      <td>0.262111</td>\n",
       "      <td>-0.816994</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>3</td>\n",
       "      <td>-0.872172</td>\n",
       "      <td>-0.607740</td>\n",
       "      <td>-0.311357</td>\n",
       "      <td>1.637039</td>\n",
       "      <td>2.096570</td>\n",
       "      <td>-1.243563</td>\n",
       "      <td>-0.597761</td>\n",
       "      <td>-1.545297</td>\n",
       "      <td>1.222385</td>\n",
       "      <td>...</td>\n",
       "      <td>1.264194</td>\n",
       "      <td>0.531085</td>\n",
       "      <td>-0.718616</td>\n",
       "      <td>-0.052778</td>\n",
       "      <td>-0.028183</td>\n",
       "      <td>0.874781</td>\n",
       "      <td>-1.485297</td>\n",
       "      <td>-0.375376</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>0.322329</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>15</td>\n",
       "      <td>3.154149</td>\n",
       "      <td>3.614214</td>\n",
       "      <td>0.046290</td>\n",
       "      <td>1.298641</td>\n",
       "      <td>1.182965</td>\n",
       "      <td>-0.751155</td>\n",
       "      <td>-0.021664</td>\n",
       "      <td>-1.067450</td>\n",
       "      <td>0.786429</td>\n",
       "      <td>...</td>\n",
       "      <td>0.164895</td>\n",
       "      <td>0.531085</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>0.475002</td>\n",
       "      <td>-1.155503</td>\n",
       "      <td>0.874781</td>\n",
       "      <td>3.174458</td>\n",
       "      <td>-0.375376</td>\n",
       "      <td>0.178174</td>\n",
       "      <td>1.128152</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>17</td>\n",
       "      <td>0.100599</td>\n",
       "      <td>-0.409156</td>\n",
       "      <td>-1.319103</td>\n",
       "      <td>-0.998524</td>\n",
       "      <td>0.873123</td>\n",
       "      <td>-1.522516</td>\n",
       "      <td>1.936684</td>\n",
       "      <td>-0.461861</td>\n",
       "      <td>1.379813</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>-0.917329</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>1.530561</td>\n",
       "      <td>-1.155503</td>\n",
       "      <td>-0.331813</td>\n",
       "      <td>0.262111</td>\n",
       "      <td>1.391098</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>18</td>\n",
       "      <td>0.167032</td>\n",
       "      <td>-0.277493</td>\n",
       "      <td>1.643821</td>\n",
       "      <td>-0.020386</td>\n",
       "      <td>-0.477818</td>\n",
       "      <td>1.653008</td>\n",
       "      <td>-1.024235</td>\n",
       "      <td>0.599292</td>\n",
       "      <td>-0.631749</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>2.462303</td>\n",
       "      <td>-0.330175</td>\n",
       "      <td>-1.108338</td>\n",
       "      <td>1.662797</td>\n",
       "      <td>1.478078</td>\n",
       "      <td>0.262111</td>\n",
       "      <td>-1.258612</td>\n",
       "      <td>1.959916</td>\n",
       "      <td>0.322329</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>2.200359</td>\n",
       "      <td>0.916286</td>\n",
       "      <td>-0.566792</td>\n",
       "      <td>-0.068358</td>\n",
       "      <td>0.539400</td>\n",
       "      <td>-0.011350</td>\n",
       "      <td>-0.062460</td>\n",
       "      <td>-0.147406</td>\n",
       "      <td>0.516784</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>-0.434524</td>\n",
       "      <td>0.446707</td>\n",
       "      <td>-0.052778</td>\n",
       "      <td>-0.591843</td>\n",
       "      <td>-1.538407</td>\n",
       "      <td>0.262111</td>\n",
       "      <td>0.949480</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>1.933975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>9</td>\n",
       "      <td>-0.395277</td>\n",
       "      <td>-0.665909</td>\n",
       "      <td>1.521628</td>\n",
       "      <td>-0.787892</td>\n",
       "      <td>-0.529846</td>\n",
       "      <td>-0.814698</td>\n",
       "      <td>1.097886</td>\n",
       "      <td>0.977622</td>\n",
       "      <td>-0.523221</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>-1.400133</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>-0.052778</td>\n",
       "      <td>1.099137</td>\n",
       "      <td>0.874781</td>\n",
       "      <td>0.844581</td>\n",
       "      <td>-0.816994</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-1.289317</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>1</td>\n",
       "      <td>-0.212586</td>\n",
       "      <td>-0.457793</td>\n",
       "      <td>-0.261725</td>\n",
       "      <td>-0.261959</td>\n",
       "      <td>-0.670333</td>\n",
       "      <td>-0.867389</td>\n",
       "      <td>0.996081</td>\n",
       "      <td>1.172775</td>\n",
       "      <td>-0.877051</td>\n",
       "      <td>...</td>\n",
       "      <td>1.264194</td>\n",
       "      <td>0.048280</td>\n",
       "      <td>-0.330175</td>\n",
       "      <td>-0.580558</td>\n",
       "      <td>-1.155503</td>\n",
       "      <td>-0.935110</td>\n",
       "      <td>0.844581</td>\n",
       "      <td>0.066243</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>8</td>\n",
       "      <td>-0.001424</td>\n",
       "      <td>0.639069</td>\n",
       "      <td>0.362960</td>\n",
       "      <td>-0.110787</td>\n",
       "      <td>0.310333</td>\n",
       "      <td>1.438336</td>\n",
       "      <td>-0.354755</td>\n",
       "      <td>-0.638613</td>\n",
       "      <td>0.866905</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>-1.400133</td>\n",
       "      <td>3.942677</td>\n",
       "      <td>-0.580558</td>\n",
       "      <td>-0.591843</td>\n",
       "      <td>0.271484</td>\n",
       "      <td>0.262111</td>\n",
       "      <td>-0.375376</td>\n",
       "      <td>0.178174</td>\n",
       "      <td>-1.289317</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>7</td>\n",
       "      <td>-0.212586</td>\n",
       "      <td>-0.451189</td>\n",
       "      <td>-0.026823</td>\n",
       "      <td>-0.132687</td>\n",
       "      <td>0.879556</td>\n",
       "      <td>-0.030928</td>\n",
       "      <td>0.705208</td>\n",
       "      <td>-0.485064</td>\n",
       "      <td>1.147099</td>\n",
       "      <td>...</td>\n",
       "      <td>0.164895</td>\n",
       "      <td>-0.917329</td>\n",
       "      <td>-0.718616</td>\n",
       "      <td>-1.108338</td>\n",
       "      <td>1.662797</td>\n",
       "      <td>1.478078</td>\n",
       "      <td>-0.320358</td>\n",
       "      <td>1.391098</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-1.289317</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>12</td>\n",
       "      <td>0.086363</td>\n",
       "      <td>-0.047548</td>\n",
       "      <td>-0.650587</td>\n",
       "      <td>-0.272159</td>\n",
       "      <td>0.333387</td>\n",
       "      <td>0.160521</td>\n",
       "      <td>-0.570374</td>\n",
       "      <td>0.289527</td>\n",
       "      <td>0.098037</td>\n",
       "      <td>...</td>\n",
       "      <td>1.264194</td>\n",
       "      <td>1.013890</td>\n",
       "      <td>-0.718616</td>\n",
       "      <td>-1.108338</td>\n",
       "      <td>-1.155503</td>\n",
       "      <td>-1.538407</td>\n",
       "      <td>-0.320358</td>\n",
       "      <td>0.949480</td>\n",
       "      <td>0.178174</td>\n",
       "      <td>0.322329</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>20 rows × 72 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    ID  Sequence_Length  Molar_Extinction_Coefficient  Isoelectric_Point  \\\n",
       "0    6        -0.433239                     -0.515963           1.151919   \n",
       "1    2        -0.838955                     -0.530375          -1.182085   \n",
       "2   10        -0.729815                     -0.207026           1.225676   \n",
       "3    0        -0.390532                      0.584023          -0.783093   \n",
       "4   13         0.413783                     -0.425000          -1.115693   \n",
       "5    5        -0.516280                     -0.765299          -0.359976   \n",
       "6    4        -0.141408                      0.350531          -0.018720   \n",
       "7   16        -0.962331                     -0.855449           0.792799   \n",
       "8   19        -0.952841                     -0.854636          -1.709162   \n",
       "9   11         0.537159                      0.966453           1.560026   \n",
       "10   3        -0.872172                     -0.607740          -0.311357   \n",
       "11  15         3.154149                      3.614214           0.046290   \n",
       "12  17         0.100599                     -0.409156          -1.319103   \n",
       "13  18         0.167032                     -0.277493           1.643821   \n",
       "14  14         2.200359                      0.916286          -0.566792   \n",
       "15   9        -0.395277                     -0.665909           1.521628   \n",
       "16   1        -0.212586                     -0.457793          -0.261725   \n",
       "17   8        -0.001424                      0.639069           0.362960   \n",
       "18   7        -0.212586                     -0.451189          -0.026823   \n",
       "19  12         0.086363                     -0.047548          -0.650587   \n",
       "\n",
       "    Aromaticity  Secondary_Structure_Helix  Secondary_Structure_Turn  \\\n",
       "0     -0.868188                  -0.410363                 -0.268194   \n",
       "1      0.560559                  -0.100613                 -0.706219   \n",
       "2      0.683511                  -0.356537                  1.115924   \n",
       "3      2.682023                   1.348737                 -1.528675   \n",
       "4     -1.055700                  -0.904704                  0.911975   \n",
       "5     -1.306744                  -1.880121                  1.480231   \n",
       "6      0.136607                  -0.132222                  1.233746   \n",
       "7     -0.558551                  -0.823612                  0.148896   \n",
       "8     -1.227487                  -1.908166                  0.314833   \n",
       "9      0.671040                   0.630263                 -0.712784   \n",
       "10     1.637039                   2.096570                 -1.243563   \n",
       "11     1.298641                   1.182965                 -0.751155   \n",
       "12    -0.998524                   0.873123                 -1.522516   \n",
       "13    -0.020386                  -0.477818                  1.653008   \n",
       "14    -0.068358                   0.539400                 -0.011350   \n",
       "15    -0.787892                  -0.529846                 -0.814698   \n",
       "16    -0.261959                  -0.670333                 -0.867389   \n",
       "17    -0.110787                   0.310333                  1.438336   \n",
       "18    -0.132687                   0.879556                 -0.030928   \n",
       "19    -0.272159                   0.333387                  0.160521   \n",
       "\n",
       "    Secondary_Structure_Sheet  Mean_Flexibility     Gravy  ...        Me  \\\n",
       "0                    0.781389          0.224094  0.259319  ... -0.934405   \n",
       "1                    1.798115         -0.708992  1.006882  ... -0.934405   \n",
       "2                   -1.284450         -0.520519 -0.100817  ... -0.934405   \n",
       "3                   -0.038853         -1.605862 -0.037508  ...  0.164895   \n",
       "4                   -0.171175          1.234026 -0.803807  ...  1.264194   \n",
       "5                   -0.969447          1.361296 -2.034525  ...  0.164895   \n",
       "6                   -0.862837         -0.348423 -0.013456  ...  0.164895   \n",
       "7                    0.579369          0.024470 -0.076215  ...  2.363494   \n",
       "8                   -2.034345          2.333899 -2.493635  ... -0.934405   \n",
       "9                    0.097624         -0.687514  0.308330  ...  0.164895   \n",
       "10                  -0.597761         -1.545297  1.222385  ...  1.264194   \n",
       "11                  -0.021664         -1.067450  0.786429  ...  0.164895   \n",
       "12                   1.936684         -0.461861  1.379813  ... -0.934405   \n",
       "13                  -1.024235          0.599292 -0.631749  ... -0.934405   \n",
       "14                  -0.062460         -0.147406  0.516784  ... -0.934405   \n",
       "15                   1.097886          0.977622 -0.523221  ... -0.934405   \n",
       "16                   0.996081          1.172775 -0.877051  ...  1.264194   \n",
       "17                  -0.354755         -0.638613  0.866905  ... -0.934405   \n",
       "18                   0.705208         -0.485064  1.147099  ...  0.164895   \n",
       "19                  -0.570374          0.289527  0.098037  ...  1.264194   \n",
       "\n",
       "          Ne        Pe        Qe        Re        Se        Te        Ve  \\\n",
       "0   0.048280 -0.330175  0.475002 -0.591843  0.271484 -0.902828 -0.816994   \n",
       "1  -0.917329 -0.718616 -1.108338  1.099137 -1.538407  0.844581  0.949480   \n",
       "2   1.496694  0.058266 -0.052778 -0.591843  0.271484  0.262111 -1.258612   \n",
       "3  -0.434524  0.058266  1.530561  0.535477 -0.935110 -0.320358 -0.816994   \n",
       "4   1.496694  0.835148  1.530561 -0.591843 -0.331813 -0.902828 -0.375376   \n",
       "5   0.048280  0.058266  2.058341  0.535477 -0.331813 -1.485297  0.507861   \n",
       "6   0.048280 -0.718616 -1.108338 -1.155503 -0.935110 -0.320358  2.274335   \n",
       "7  -0.434524  0.058266 -0.052778  1.662797 -0.331813 -0.320358 -1.258612   \n",
       "8  -0.917329 -0.718616  0.475002  0.535477  0.874781 -0.902828  0.066243   \n",
       "9   0.048280 -0.330175 -1.108338 -0.028183  1.478078  0.262111 -0.816994   \n",
       "10  0.531085 -0.718616 -0.052778 -0.028183  0.874781 -1.485297 -0.375376   \n",
       "11  0.531085  0.058266  0.475002 -1.155503  0.874781  3.174458 -0.375376   \n",
       "12 -0.917329  0.058266  1.530561 -1.155503 -0.331813  0.262111  1.391098   \n",
       "13  2.462303 -0.330175 -1.108338  1.662797  1.478078  0.262111 -1.258612   \n",
       "14 -0.434524  0.446707 -0.052778 -0.591843 -1.538407  0.262111  0.949480   \n",
       "15 -1.400133  0.058266 -0.052778  1.099137  0.874781  0.844581 -0.816994   \n",
       "16  0.048280 -0.330175 -0.580558 -1.155503 -0.935110  0.844581  0.066243   \n",
       "17 -1.400133  3.942677 -0.580558 -0.591843  0.271484  0.262111 -0.375376   \n",
       "18 -0.917329 -0.718616 -1.108338  1.662797  1.478078 -0.320358  1.391098   \n",
       "19  1.013890 -0.718616 -1.108338 -1.155503 -1.538407 -0.320358  0.949480   \n",
       "\n",
       "          We        Ye  \n",
       "0  -0.712697  1.128152  \n",
       "1   1.069045 -0.483494  \n",
       "2   0.178174 -0.483494  \n",
       "3   1.069045  1.128152  \n",
       "4  -0.712697 -1.289317  \n",
       "5  -0.712697 -0.483494  \n",
       "6   2.850787  1.933975  \n",
       "7   0.178174  0.322329  \n",
       "8  -0.712697 -0.483494  \n",
       "9  -0.712697 -0.483494  \n",
       "10 -0.712697  0.322329  \n",
       "11  0.178174  1.128152  \n",
       "12 -0.712697 -0.483494  \n",
       "13  1.959916  0.322329  \n",
       "14 -0.712697  1.933975  \n",
       "15 -0.712697 -1.289317  \n",
       "16 -0.712697 -0.483494  \n",
       "17  0.178174 -1.289317  \n",
       "18 -0.712697 -1.289317  \n",
       "19  0.178174  0.322329  \n",
       "\n",
       "[20 rows x 72 columns]"
      ]
     },
     "execution_count": 359,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_blind"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 360,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Sequence_Length</th>\n",
       "      <th>Molar_Extinction_Coefficient</th>\n",
       "      <th>Isoelectric_Point</th>\n",
       "      <th>Aromaticity</th>\n",
       "      <th>Secondary_Structure_Helix</th>\n",
       "      <th>Secondary_Structure_Turn</th>\n",
       "      <th>Secondary_Structure_Sheet</th>\n",
       "      <th>Mean_Flexibility</th>\n",
       "      <th>Gravy</th>\n",
       "      <th>Molecular_Weight</th>\n",
       "      <th>...</th>\n",
       "      <th>Me</th>\n",
       "      <th>Ne</th>\n",
       "      <th>Pe</th>\n",
       "      <th>Qe</th>\n",
       "      <th>Re</th>\n",
       "      <th>Se</th>\n",
       "      <th>Te</th>\n",
       "      <th>Ve</th>\n",
       "      <th>We</th>\n",
       "      <th>Ye</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>-0.433239</td>\n",
       "      <td>-0.515963</td>\n",
       "      <td>1.151919</td>\n",
       "      <td>-0.868188</td>\n",
       "      <td>-0.410363</td>\n",
       "      <td>-0.268194</td>\n",
       "      <td>0.781389</td>\n",
       "      <td>0.224094</td>\n",
       "      <td>0.259319</td>\n",
       "      <td>-0.458396</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>0.048280</td>\n",
       "      <td>-0.330175</td>\n",
       "      <td>0.475002</td>\n",
       "      <td>-0.591843</td>\n",
       "      <td>0.271484</td>\n",
       "      <td>-0.902828</td>\n",
       "      <td>-0.816994</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>1.128152</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>-0.838955</td>\n",
       "      <td>-0.530375</td>\n",
       "      <td>-1.182085</td>\n",
       "      <td>0.560559</td>\n",
       "      <td>-0.100613</td>\n",
       "      <td>-0.706219</td>\n",
       "      <td>1.798115</td>\n",
       "      <td>-0.708992</td>\n",
       "      <td>1.006882</td>\n",
       "      <td>-0.847695</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>-0.917329</td>\n",
       "      <td>-0.718616</td>\n",
       "      <td>-1.108338</td>\n",
       "      <td>1.099137</td>\n",
       "      <td>-1.538407</td>\n",
       "      <td>0.844581</td>\n",
       "      <td>0.949480</td>\n",
       "      <td>1.069045</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>-0.729815</td>\n",
       "      <td>-0.207026</td>\n",
       "      <td>1.225676</td>\n",
       "      <td>0.683511</td>\n",
       "      <td>-0.356537</td>\n",
       "      <td>1.115924</td>\n",
       "      <td>-1.284450</td>\n",
       "      <td>-0.520519</td>\n",
       "      <td>-0.100817</td>\n",
       "      <td>-0.719995</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>1.496694</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>-0.052778</td>\n",
       "      <td>-0.591843</td>\n",
       "      <td>0.271484</td>\n",
       "      <td>0.262111</td>\n",
       "      <td>-1.258612</td>\n",
       "      <td>0.178174</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>-0.390532</td>\n",
       "      <td>0.584023</td>\n",
       "      <td>-0.783093</td>\n",
       "      <td>2.682023</td>\n",
       "      <td>1.348737</td>\n",
       "      <td>-1.528675</td>\n",
       "      <td>-0.038853</td>\n",
       "      <td>-1.605862</td>\n",
       "      <td>-0.037508</td>\n",
       "      <td>-0.320326</td>\n",
       "      <td>...</td>\n",
       "      <td>0.164895</td>\n",
       "      <td>-0.434524</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>1.530561</td>\n",
       "      <td>0.535477</td>\n",
       "      <td>-0.935110</td>\n",
       "      <td>-0.320358</td>\n",
       "      <td>-0.816994</td>\n",
       "      <td>1.069045</td>\n",
       "      <td>1.128152</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.413783</td>\n",
       "      <td>-0.425000</td>\n",
       "      <td>-1.115693</td>\n",
       "      <td>-1.055700</td>\n",
       "      <td>-0.904704</td>\n",
       "      <td>0.911975</td>\n",
       "      <td>-0.171175</td>\n",
       "      <td>1.234026</td>\n",
       "      <td>-0.803807</td>\n",
       "      <td>0.387257</td>\n",
       "      <td>...</td>\n",
       "      <td>1.264194</td>\n",
       "      <td>1.496694</td>\n",
       "      <td>0.835148</td>\n",
       "      <td>1.530561</td>\n",
       "      <td>-0.591843</td>\n",
       "      <td>-0.331813</td>\n",
       "      <td>-0.902828</td>\n",
       "      <td>-0.375376</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-1.289317</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>-0.516280</td>\n",
       "      <td>-0.765299</td>\n",
       "      <td>-0.359976</td>\n",
       "      <td>-1.306744</td>\n",
       "      <td>-1.880121</td>\n",
       "      <td>1.480231</td>\n",
       "      <td>-0.969447</td>\n",
       "      <td>1.361296</td>\n",
       "      <td>-2.034525</td>\n",
       "      <td>-0.511519</td>\n",
       "      <td>...</td>\n",
       "      <td>0.164895</td>\n",
       "      <td>0.048280</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>2.058341</td>\n",
       "      <td>0.535477</td>\n",
       "      <td>-0.331813</td>\n",
       "      <td>-1.485297</td>\n",
       "      <td>0.507861</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>-0.141408</td>\n",
       "      <td>0.350531</td>\n",
       "      <td>-0.018720</td>\n",
       "      <td>0.136607</td>\n",
       "      <td>-0.132222</td>\n",
       "      <td>1.233746</td>\n",
       "      <td>-0.862837</td>\n",
       "      <td>-0.348423</td>\n",
       "      <td>-0.013456</td>\n",
       "      <td>-0.162065</td>\n",
       "      <td>...</td>\n",
       "      <td>0.164895</td>\n",
       "      <td>0.048280</td>\n",
       "      <td>-0.718616</td>\n",
       "      <td>-1.108338</td>\n",
       "      <td>-1.155503</td>\n",
       "      <td>-0.935110</td>\n",
       "      <td>-0.320358</td>\n",
       "      <td>2.274335</td>\n",
       "      <td>2.850787</td>\n",
       "      <td>1.933975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>-0.962331</td>\n",
       "      <td>-0.855449</td>\n",
       "      <td>0.792799</td>\n",
       "      <td>-0.558551</td>\n",
       "      <td>-0.823612</td>\n",
       "      <td>0.148896</td>\n",
       "      <td>0.579369</td>\n",
       "      <td>0.024470</td>\n",
       "      <td>-0.076215</td>\n",
       "      <td>-0.954302</td>\n",
       "      <td>...</td>\n",
       "      <td>2.363494</td>\n",
       "      <td>-0.434524</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>-0.052778</td>\n",
       "      <td>1.662797</td>\n",
       "      <td>-0.331813</td>\n",
       "      <td>-0.320358</td>\n",
       "      <td>-1.258612</td>\n",
       "      <td>0.178174</td>\n",
       "      <td>0.322329</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>-0.952841</td>\n",
       "      <td>-0.854636</td>\n",
       "      <td>-1.709162</td>\n",
       "      <td>-1.227487</td>\n",
       "      <td>-1.908166</td>\n",
       "      <td>0.314833</td>\n",
       "      <td>-2.034345</td>\n",
       "      <td>2.333899</td>\n",
       "      <td>-2.493635</td>\n",
       "      <td>-0.943760</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>-0.917329</td>\n",
       "      <td>-0.718616</td>\n",
       "      <td>0.475002</td>\n",
       "      <td>0.535477</td>\n",
       "      <td>0.874781</td>\n",
       "      <td>-0.902828</td>\n",
       "      <td>0.066243</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.537159</td>\n",
       "      <td>0.966453</td>\n",
       "      <td>1.560026</td>\n",
       "      <td>0.671040</td>\n",
       "      <td>0.630263</td>\n",
       "      <td>-0.712784</td>\n",
       "      <td>0.097624</td>\n",
       "      <td>-0.687514</td>\n",
       "      <td>0.308330</td>\n",
       "      <td>0.568182</td>\n",
       "      <td>...</td>\n",
       "      <td>0.164895</td>\n",
       "      <td>0.048280</td>\n",
       "      <td>-0.330175</td>\n",
       "      <td>-1.108338</td>\n",
       "      <td>-0.028183</td>\n",
       "      <td>1.478078</td>\n",
       "      <td>0.262111</td>\n",
       "      <td>-0.816994</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>-0.872172</td>\n",
       "      <td>-0.607740</td>\n",
       "      <td>-0.311357</td>\n",
       "      <td>1.637039</td>\n",
       "      <td>2.096570</td>\n",
       "      <td>-1.243563</td>\n",
       "      <td>-0.597761</td>\n",
       "      <td>-1.545297</td>\n",
       "      <td>1.222385</td>\n",
       "      <td>-0.840806</td>\n",
       "      <td>...</td>\n",
       "      <td>1.264194</td>\n",
       "      <td>0.531085</td>\n",
       "      <td>-0.718616</td>\n",
       "      <td>-0.052778</td>\n",
       "      <td>-0.028183</td>\n",
       "      <td>0.874781</td>\n",
       "      <td>-1.485297</td>\n",
       "      <td>-0.375376</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>0.322329</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>3.154149</td>\n",
       "      <td>3.614214</td>\n",
       "      <td>0.046290</td>\n",
       "      <td>1.298641</td>\n",
       "      <td>1.182965</td>\n",
       "      <td>-0.751155</td>\n",
       "      <td>-0.021664</td>\n",
       "      <td>-1.067450</td>\n",
       "      <td>0.786429</td>\n",
       "      <td>3.215393</td>\n",
       "      <td>...</td>\n",
       "      <td>0.164895</td>\n",
       "      <td>0.531085</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>0.475002</td>\n",
       "      <td>-1.155503</td>\n",
       "      <td>0.874781</td>\n",
       "      <td>3.174458</td>\n",
       "      <td>-0.375376</td>\n",
       "      <td>0.178174</td>\n",
       "      <td>1.128152</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.100599</td>\n",
       "      <td>-0.409156</td>\n",
       "      <td>-1.319103</td>\n",
       "      <td>-0.998524</td>\n",
       "      <td>0.873123</td>\n",
       "      <td>-1.522516</td>\n",
       "      <td>1.936684</td>\n",
       "      <td>-0.461861</td>\n",
       "      <td>1.379813</td>\n",
       "      <td>0.084793</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>-0.917329</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>1.530561</td>\n",
       "      <td>-1.155503</td>\n",
       "      <td>-0.331813</td>\n",
       "      <td>0.262111</td>\n",
       "      <td>1.391098</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.167032</td>\n",
       "      <td>-0.277493</td>\n",
       "      <td>1.643821</td>\n",
       "      <td>-0.020386</td>\n",
       "      <td>-0.477818</td>\n",
       "      <td>1.653008</td>\n",
       "      <td>-1.024235</td>\n",
       "      <td>0.599292</td>\n",
       "      <td>-0.631749</td>\n",
       "      <td>0.139774</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>2.462303</td>\n",
       "      <td>-0.330175</td>\n",
       "      <td>-1.108338</td>\n",
       "      <td>1.662797</td>\n",
       "      <td>1.478078</td>\n",
       "      <td>0.262111</td>\n",
       "      <td>-1.258612</td>\n",
       "      <td>1.959916</td>\n",
       "      <td>0.322329</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>2.200359</td>\n",
       "      <td>0.916286</td>\n",
       "      <td>-0.566792</td>\n",
       "      <td>-0.068358</td>\n",
       "      <td>0.539400</td>\n",
       "      <td>-0.011350</td>\n",
       "      <td>-0.062460</td>\n",
       "      <td>-0.147406</td>\n",
       "      <td>0.516784</td>\n",
       "      <td>2.136331</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>-0.434524</td>\n",
       "      <td>0.446707</td>\n",
       "      <td>-0.052778</td>\n",
       "      <td>-0.591843</td>\n",
       "      <td>-1.538407</td>\n",
       "      <td>0.262111</td>\n",
       "      <td>0.949480</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>1.933975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>-0.395277</td>\n",
       "      <td>-0.665909</td>\n",
       "      <td>1.521628</td>\n",
       "      <td>-0.787892</td>\n",
       "      <td>-0.529846</td>\n",
       "      <td>-0.814698</td>\n",
       "      <td>1.097886</td>\n",
       "      <td>0.977622</td>\n",
       "      <td>-0.523221</td>\n",
       "      <td>-0.390358</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>-1.400133</td>\n",
       "      <td>0.058266</td>\n",
       "      <td>-0.052778</td>\n",
       "      <td>1.099137</td>\n",
       "      <td>0.874781</td>\n",
       "      <td>0.844581</td>\n",
       "      <td>-0.816994</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-1.289317</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>-0.212586</td>\n",
       "      <td>-0.457793</td>\n",
       "      <td>-0.261725</td>\n",
       "      <td>-0.261959</td>\n",
       "      <td>-0.670333</td>\n",
       "      <td>-0.867389</td>\n",
       "      <td>0.996081</td>\n",
       "      <td>1.172775</td>\n",
       "      <td>-0.877051</td>\n",
       "      <td>-0.210824</td>\n",
       "      <td>...</td>\n",
       "      <td>1.264194</td>\n",
       "      <td>0.048280</td>\n",
       "      <td>-0.330175</td>\n",
       "      <td>-0.580558</td>\n",
       "      <td>-1.155503</td>\n",
       "      <td>-0.935110</td>\n",
       "      <td>0.844581</td>\n",
       "      <td>0.066243</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-0.483494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>-0.001424</td>\n",
       "      <td>0.639069</td>\n",
       "      <td>0.362960</td>\n",
       "      <td>-0.110787</td>\n",
       "      <td>0.310333</td>\n",
       "      <td>1.438336</td>\n",
       "      <td>-0.354755</td>\n",
       "      <td>-0.638613</td>\n",
       "      <td>0.866905</td>\n",
       "      <td>-0.035169</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.934405</td>\n",
       "      <td>-1.400133</td>\n",
       "      <td>3.942677</td>\n",
       "      <td>-0.580558</td>\n",
       "      <td>-0.591843</td>\n",
       "      <td>0.271484</td>\n",
       "      <td>0.262111</td>\n",
       "      <td>-0.375376</td>\n",
       "      <td>0.178174</td>\n",
       "      <td>-1.289317</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>-0.212586</td>\n",
       "      <td>-0.451189</td>\n",
       "      <td>-0.026823</td>\n",
       "      <td>-0.132687</td>\n",
       "      <td>0.879556</td>\n",
       "      <td>-0.030928</td>\n",
       "      <td>0.705208</td>\n",
       "      <td>-0.485064</td>\n",
       "      <td>1.147099</td>\n",
       "      <td>-0.217367</td>\n",
       "      <td>...</td>\n",
       "      <td>0.164895</td>\n",
       "      <td>-0.917329</td>\n",
       "      <td>-0.718616</td>\n",
       "      <td>-1.108338</td>\n",
       "      <td>1.662797</td>\n",
       "      <td>1.478078</td>\n",
       "      <td>-0.320358</td>\n",
       "      <td>1.391098</td>\n",
       "      <td>-0.712697</td>\n",
       "      <td>-1.289317</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.086363</td>\n",
       "      <td>-0.047548</td>\n",
       "      <td>-0.650587</td>\n",
       "      <td>-0.272159</td>\n",
       "      <td>0.333387</td>\n",
       "      <td>0.160521</td>\n",
       "      <td>-0.570374</td>\n",
       "      <td>0.289527</td>\n",
       "      <td>0.098037</td>\n",
       "      <td>0.080851</td>\n",
       "      <td>...</td>\n",
       "      <td>1.264194</td>\n",
       "      <td>1.013890</td>\n",
       "      <td>-0.718616</td>\n",
       "      <td>-1.108338</td>\n",
       "      <td>-1.155503</td>\n",
       "      <td>-1.538407</td>\n",
       "      <td>-0.320358</td>\n",
       "      <td>0.949480</td>\n",
       "      <td>0.178174</td>\n",
       "      <td>0.322329</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>20 rows × 65 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    Sequence_Length  Molar_Extinction_Coefficient  Isoelectric_Point  \\\n",
       "0         -0.433239                     -0.515963           1.151919   \n",
       "1         -0.838955                     -0.530375          -1.182085   \n",
       "2         -0.729815                     -0.207026           1.225676   \n",
       "3         -0.390532                      0.584023          -0.783093   \n",
       "4          0.413783                     -0.425000          -1.115693   \n",
       "5         -0.516280                     -0.765299          -0.359976   \n",
       "6         -0.141408                      0.350531          -0.018720   \n",
       "7         -0.962331                     -0.855449           0.792799   \n",
       "8         -0.952841                     -0.854636          -1.709162   \n",
       "9          0.537159                      0.966453           1.560026   \n",
       "10        -0.872172                     -0.607740          -0.311357   \n",
       "11         3.154149                      3.614214           0.046290   \n",
       "12         0.100599                     -0.409156          -1.319103   \n",
       "13         0.167032                     -0.277493           1.643821   \n",
       "14         2.200359                      0.916286          -0.566792   \n",
       "15        -0.395277                     -0.665909           1.521628   \n",
       "16        -0.212586                     -0.457793          -0.261725   \n",
       "17        -0.001424                      0.639069           0.362960   \n",
       "18        -0.212586                     -0.451189          -0.026823   \n",
       "19         0.086363                     -0.047548          -0.650587   \n",
       "\n",
       "    Aromaticity  Secondary_Structure_Helix  Secondary_Structure_Turn  \\\n",
       "0     -0.868188                  -0.410363                 -0.268194   \n",
       "1      0.560559                  -0.100613                 -0.706219   \n",
       "2      0.683511                  -0.356537                  1.115924   \n",
       "3      2.682023                   1.348737                 -1.528675   \n",
       "4     -1.055700                  -0.904704                  0.911975   \n",
       "5     -1.306744                  -1.880121                  1.480231   \n",
       "6      0.136607                  -0.132222                  1.233746   \n",
       "7     -0.558551                  -0.823612                  0.148896   \n",
       "8     -1.227487                  -1.908166                  0.314833   \n",
       "9      0.671040                   0.630263                 -0.712784   \n",
       "10     1.637039                   2.096570                 -1.243563   \n",
       "11     1.298641                   1.182965                 -0.751155   \n",
       "12    -0.998524                   0.873123                 -1.522516   \n",
       "13    -0.020386                  -0.477818                  1.653008   \n",
       "14    -0.068358                   0.539400                 -0.011350   \n",
       "15    -0.787892                  -0.529846                 -0.814698   \n",
       "16    -0.261959                  -0.670333                 -0.867389   \n",
       "17    -0.110787                   0.310333                  1.438336   \n",
       "18    -0.132687                   0.879556                 -0.030928   \n",
       "19    -0.272159                   0.333387                  0.160521   \n",
       "\n",
       "    Secondary_Structure_Sheet  Mean_Flexibility     Gravy  Molecular_Weight  \\\n",
       "0                    0.781389          0.224094  0.259319         -0.458396   \n",
       "1                    1.798115         -0.708992  1.006882         -0.847695   \n",
       "2                   -1.284450         -0.520519 -0.100817         -0.719995   \n",
       "3                   -0.038853         -1.605862 -0.037508         -0.320326   \n",
       "4                   -0.171175          1.234026 -0.803807          0.387257   \n",
       "5                   -0.969447          1.361296 -2.034525         -0.511519   \n",
       "6                   -0.862837         -0.348423 -0.013456         -0.162065   \n",
       "7                    0.579369          0.024470 -0.076215         -0.954302   \n",
       "8                   -2.034345          2.333899 -2.493635         -0.943760   \n",
       "9                    0.097624         -0.687514  0.308330          0.568182   \n",
       "10                  -0.597761         -1.545297  1.222385         -0.840806   \n",
       "11                  -0.021664         -1.067450  0.786429          3.215393   \n",
       "12                   1.936684         -0.461861  1.379813          0.084793   \n",
       "13                  -1.024235          0.599292 -0.631749          0.139774   \n",
       "14                  -0.062460         -0.147406  0.516784          2.136331   \n",
       "15                   1.097886          0.977622 -0.523221         -0.390358   \n",
       "16                   0.996081          1.172775 -0.877051         -0.210824   \n",
       "17                  -0.354755         -0.638613  0.866905         -0.035169   \n",
       "18                   0.705208         -0.485064  1.147099         -0.217367   \n",
       "19                  -0.570374          0.289527  0.098037          0.080851   \n",
       "\n",
       "    ...        Me        Ne        Pe        Qe        Re        Se        Te  \\\n",
       "0   ... -0.934405  0.048280 -0.330175  0.475002 -0.591843  0.271484 -0.902828   \n",
       "1   ... -0.934405 -0.917329 -0.718616 -1.108338  1.099137 -1.538407  0.844581   \n",
       "2   ... -0.934405  1.496694  0.058266 -0.052778 -0.591843  0.271484  0.262111   \n",
       "3   ...  0.164895 -0.434524  0.058266  1.530561  0.535477 -0.935110 -0.320358   \n",
       "4   ...  1.264194  1.496694  0.835148  1.530561 -0.591843 -0.331813 -0.902828   \n",
       "5   ...  0.164895  0.048280  0.058266  2.058341  0.535477 -0.331813 -1.485297   \n",
       "6   ...  0.164895  0.048280 -0.718616 -1.108338 -1.155503 -0.935110 -0.320358   \n",
       "7   ...  2.363494 -0.434524  0.058266 -0.052778  1.662797 -0.331813 -0.320358   \n",
       "8   ... -0.934405 -0.917329 -0.718616  0.475002  0.535477  0.874781 -0.902828   \n",
       "9   ...  0.164895  0.048280 -0.330175 -1.108338 -0.028183  1.478078  0.262111   \n",
       "10  ...  1.264194  0.531085 -0.718616 -0.052778 -0.028183  0.874781 -1.485297   \n",
       "11  ...  0.164895  0.531085  0.058266  0.475002 -1.155503  0.874781  3.174458   \n",
       "12  ... -0.934405 -0.917329  0.058266  1.530561 -1.155503 -0.331813  0.262111   \n",
       "13  ... -0.934405  2.462303 -0.330175 -1.108338  1.662797  1.478078  0.262111   \n",
       "14  ... -0.934405 -0.434524  0.446707 -0.052778 -0.591843 -1.538407  0.262111   \n",
       "15  ... -0.934405 -1.400133  0.058266 -0.052778  1.099137  0.874781  0.844581   \n",
       "16  ...  1.264194  0.048280 -0.330175 -0.580558 -1.155503 -0.935110  0.844581   \n",
       "17  ... -0.934405 -1.400133  3.942677 -0.580558 -0.591843  0.271484  0.262111   \n",
       "18  ...  0.164895 -0.917329 -0.718616 -1.108338  1.662797  1.478078 -0.320358   \n",
       "19  ...  1.264194  1.013890 -0.718616 -1.108338 -1.155503 -1.538407 -0.320358   \n",
       "\n",
       "          Ve        We        Ye  \n",
       "0  -0.816994 -0.712697  1.128152  \n",
       "1   0.949480  1.069045 -0.483494  \n",
       "2  -1.258612  0.178174 -0.483494  \n",
       "3  -0.816994  1.069045  1.128152  \n",
       "4  -0.375376 -0.712697 -1.289317  \n",
       "5   0.507861 -0.712697 -0.483494  \n",
       "6   2.274335  2.850787  1.933975  \n",
       "7  -1.258612  0.178174  0.322329  \n",
       "8   0.066243 -0.712697 -0.483494  \n",
       "9  -0.816994 -0.712697 -0.483494  \n",
       "10 -0.375376 -0.712697  0.322329  \n",
       "11 -0.375376  0.178174  1.128152  \n",
       "12  1.391098 -0.712697 -0.483494  \n",
       "13 -1.258612  1.959916  0.322329  \n",
       "14  0.949480 -0.712697  1.933975  \n",
       "15 -0.816994 -0.712697 -1.289317  \n",
       "16  0.066243 -0.712697 -0.483494  \n",
       "17 -0.375376  0.178174 -1.289317  \n",
       "18  1.391098 -0.712697 -1.289317  \n",
       "19  0.949480  0.178174  0.322329  \n",
       "\n",
       "[20 rows x 65 columns]"
      ]
     },
     "execution_count": 360,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# drop the ID feature and other redundantg features\n",
    "df_blind = df_blind.drop(columns=['ID'])\n",
    "X_blind = df_blind.drop(columns=deselect)\n",
    "X_blind "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 361,
   "metadata": {},
   "outputs": [],
   "source": [
    "# predict on blind data\n",
    "y_probs = clf.predict_proba(X_blind)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 362,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[3.42e-01, 9.69e-02, 5.45e-01, 1.60e-02],\n",
       "       [4.18e-01, 3.38e-01, 1.58e-02, 2.28e-01],\n",
       "       [3.14e-02, 4.65e-02, 1.67e-01, 7.55e-01],\n",
       "       [5.00e-01, 6.19e-02, 4.04e-01, 3.43e-02],\n",
       "       [4.12e-01, 3.85e-04, 5.87e-01, 1.06e-03],\n",
       "       [2.28e-01, 9.10e-03, 7.44e-01, 1.89e-02],\n",
       "       [1.10e-01, 3.21e-02, 1.46e-01, 7.11e-01],\n",
       "       [2.20e-02, 1.68e-01, 1.84e-01, 6.26e-01],\n",
       "       [1.79e-01, 3.38e-02, 7.49e-01, 3.77e-02],\n",
       "       [2.97e-01, 4.77e-01, 2.17e-01, 9.46e-03],\n",
       "       [5.20e-01, 8.46e-02, 2.86e-01, 1.10e-01],\n",
       "       [5.69e-01, 6.85e-02, 3.49e-01, 1.33e-02],\n",
       "       [7.46e-01, 1.59e-02, 2.36e-01, 1.78e-03],\n",
       "       [2.24e-01, 1.67e-02, 7.56e-01, 2.73e-03],\n",
       "       [6.43e-01, 9.27e-03, 8.37e-02, 2.64e-01],\n",
       "       [1.67e-01, 3.53e-01, 4.72e-01, 7.75e-03],\n",
       "       [5.13e-01, 1.16e-02, 4.75e-01, 5.66e-04],\n",
       "       [5.01e-01, 5.97e-03, 4.40e-01, 5.25e-02],\n",
       "       [5.13e-01, 1.49e-01, 3.36e-01, 2.02e-03],\n",
       "       [4.52e-01, 1.51e-01, 2.66e-01, 1.31e-01]])"
      ]
     },
     "execution_count": 362,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_probs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 363,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SEQ ID</th>\n",
       "      <th>cyto</th>\n",
       "      <th>mito</th>\n",
       "      <th>nuc</th>\n",
       "      <th>sec</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>SEQ677</td>\n",
       "      <td>0.342121</td>\n",
       "      <td>0.096901</td>\n",
       "      <td>0.544929</td>\n",
       "      <td>0.016049</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>SEQ231</td>\n",
       "      <td>0.417735</td>\n",
       "      <td>0.338048</td>\n",
       "      <td>0.015849</td>\n",
       "      <td>0.228367</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>SEQ871</td>\n",
       "      <td>0.031388</td>\n",
       "      <td>0.046542</td>\n",
       "      <td>0.167224</td>\n",
       "      <td>0.754847</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>SEQ388</td>\n",
       "      <td>0.499552</td>\n",
       "      <td>0.061892</td>\n",
       "      <td>0.404229</td>\n",
       "      <td>0.034328</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>SEQ122</td>\n",
       "      <td>0.411901</td>\n",
       "      <td>0.000385</td>\n",
       "      <td>0.586656</td>\n",
       "      <td>0.001057</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>SEQ758</td>\n",
       "      <td>0.227540</td>\n",
       "      <td>0.009099</td>\n",
       "      <td>0.744499</td>\n",
       "      <td>0.018863</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>SEQ333</td>\n",
       "      <td>0.110427</td>\n",
       "      <td>0.032051</td>\n",
       "      <td>0.146313</td>\n",
       "      <td>0.711209</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>SEQ937</td>\n",
       "      <td>0.022010</td>\n",
       "      <td>0.167663</td>\n",
       "      <td>0.184420</td>\n",
       "      <td>0.625908</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>SEQ351</td>\n",
       "      <td>0.179313</td>\n",
       "      <td>0.033805</td>\n",
       "      <td>0.749226</td>\n",
       "      <td>0.037655</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>SEQ202</td>\n",
       "      <td>0.296682</td>\n",
       "      <td>0.477004</td>\n",
       "      <td>0.216856</td>\n",
       "      <td>0.009458</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>SEQ608</td>\n",
       "      <td>0.520254</td>\n",
       "      <td>0.084634</td>\n",
       "      <td>0.285591</td>\n",
       "      <td>0.109521</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>SEQ402</td>\n",
       "      <td>0.568981</td>\n",
       "      <td>0.068475</td>\n",
       "      <td>0.349196</td>\n",
       "      <td>0.013348</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>SEQ433</td>\n",
       "      <td>0.745915</td>\n",
       "      <td>0.015914</td>\n",
       "      <td>0.236390</td>\n",
       "      <td>0.001781</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>SEQ821</td>\n",
       "      <td>0.224178</td>\n",
       "      <td>0.016668</td>\n",
       "      <td>0.756427</td>\n",
       "      <td>0.002727</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>SEQ322</td>\n",
       "      <td>0.643423</td>\n",
       "      <td>0.009268</td>\n",
       "      <td>0.083650</td>\n",
       "      <td>0.263660</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>SEQ982</td>\n",
       "      <td>0.166972</td>\n",
       "      <td>0.353282</td>\n",
       "      <td>0.472001</td>\n",
       "      <td>0.007745</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>SEQ951</td>\n",
       "      <td>0.512998</td>\n",
       "      <td>0.011602</td>\n",
       "      <td>0.474834</td>\n",
       "      <td>0.000566</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>SEQ173</td>\n",
       "      <td>0.501497</td>\n",
       "      <td>0.005972</td>\n",
       "      <td>0.439996</td>\n",
       "      <td>0.052536</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>SEQ862</td>\n",
       "      <td>0.513225</td>\n",
       "      <td>0.148538</td>\n",
       "      <td>0.336218</td>\n",
       "      <td>0.002018</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>SEQ224</td>\n",
       "      <td>0.451742</td>\n",
       "      <td>0.151130</td>\n",
       "      <td>0.266349</td>\n",
       "      <td>0.130778</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    SEQ ID      cyto      mito       nuc       sec\n",
       "0   SEQ677  0.342121  0.096901  0.544929  0.016049\n",
       "1   SEQ231  0.417735  0.338048  0.015849  0.228367\n",
       "2   SEQ871  0.031388  0.046542  0.167224  0.754847\n",
       "3   SEQ388  0.499552  0.061892  0.404229  0.034328\n",
       "4   SEQ122  0.411901  0.000385  0.586656  0.001057\n",
       "5   SEQ758  0.227540  0.009099  0.744499  0.018863\n",
       "6   SEQ333  0.110427  0.032051  0.146313  0.711209\n",
       "7   SEQ937  0.022010  0.167663  0.184420  0.625908\n",
       "8   SEQ351  0.179313  0.033805  0.749226  0.037655\n",
       "9   SEQ202  0.296682  0.477004  0.216856  0.009458\n",
       "10  SEQ608  0.520254  0.084634  0.285591  0.109521\n",
       "11  SEQ402  0.568981  0.068475  0.349196  0.013348\n",
       "12  SEQ433  0.745915  0.015914  0.236390  0.001781\n",
       "13  SEQ821  0.224178  0.016668  0.756427  0.002727\n",
       "14  SEQ322  0.643423  0.009268  0.083650  0.263660\n",
       "15  SEQ982  0.166972  0.353282  0.472001  0.007745\n",
       "16  SEQ951  0.512998  0.011602  0.474834  0.000566\n",
       "17  SEQ173  0.501497  0.005972  0.439996  0.052536\n",
       "18  SEQ862  0.513225  0.148538  0.336218  0.002018\n",
       "19  SEQ224  0.451742  0.151130  0.266349  0.130778"
      ]
     },
     "execution_count": 363,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Creating pandas dataframe from numpy array\n",
    "prob_data = pd.DataFrame({'SEQ ID':ID_list,'cyto': y_probs[:, 0], 'mito': y_probs[:, 1], 'nuc': y_probs[:, 2], 'sec': y_probs[:, 3]})\n",
    "prob_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 374,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\\begin{tabular}{lrrrr}\n",
      "\\toprule\n",
      " SEQ ID &      cyto &      mito &       nuc &       sec \\\\\n",
      "\\midrule\n",
      " SEQ677 &  0.342121 &  0.096901 &  0.544929 &  0.016049 \\\\\n",
      " SEQ231 &  0.417735 &  0.338048 &  0.015849 &  0.228367 \\\\\n",
      " SEQ871 &  0.031388 &  0.046542 &  0.167224 &  0.754847 \\\\\n",
      " SEQ388 &  0.499552 &  0.061892 &  0.404229 &  0.034328 \\\\\n",
      " SEQ122 &  0.411901 &  0.000385 &  0.586656 &  0.001057 \\\\\n",
      " SEQ758 &  0.227540 &  0.009099 &  0.744499 &  0.018863 \\\\\n",
      " SEQ333 &  0.110427 &  0.032051 &  0.146313 &  0.711209 \\\\\n",
      " SEQ937 &  0.022010 &  0.167663 &  0.184420 &  0.625908 \\\\\n",
      " SEQ351 &  0.179313 &  0.033805 &  0.749226 &  0.037655 \\\\\n",
      " SEQ202 &  0.296682 &  0.477004 &  0.216856 &  0.009458 \\\\\n",
      " SEQ608 &  0.520254 &  0.084634 &  0.285591 &  0.109521 \\\\\n",
      " SEQ402 &  0.568981 &  0.068475 &  0.349196 &  0.013348 \\\\\n",
      " SEQ433 &  0.745915 &  0.015914 &  0.236390 &  0.001781 \\\\\n",
      " SEQ821 &  0.224178 &  0.016668 &  0.756427 &  0.002727 \\\\\n",
      " SEQ322 &  0.643423 &  0.009268 &  0.083650 &  0.263660 \\\\\n",
      " SEQ982 &  0.166972 &  0.353282 &  0.472001 &  0.007745 \\\\\n",
      " SEQ951 &  0.512998 &  0.011602 &  0.474834 &  0.000566 \\\\\n",
      " SEQ173 &  0.501497 &  0.005972 &  0.439996 &  0.052536 \\\\\n",
      " SEQ862 &  0.513225 &  0.148538 &  0.336218 &  0.002018 \\\\\n",
      " SEQ224 &  0.451742 &  0.151130 &  0.266349 &  0.130778 \\\\\n",
      "\\bottomrule\n",
      "\\end{tabular}\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(prob_data.to_latex(index=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 364,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[2, 0, 3, 0, 2, 2, 3, 3, 2, 1, 0, 0, 0, 2, 0, 2, 0, 0, 0, 0]"
      ]
     },
     "execution_count": 364,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# get prediction and its confidence\n",
    "y_blind = []\n",
    "y_conf = []\n",
    "\n",
    "for idx in range(len(y_probs)):\n",
    "    # predicted class is argmax\n",
    "    a = np.argmax(y_probs[idx])\n",
    "    y_blind.append(a)\n",
    "    # confidence is corresponding y_probs value\n",
    "    c = y_probs[idx,a]\n",
    "    y_conf.append(c)\n",
    "\n",
    "y_blind"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 365,
   "metadata": {},
   "outputs": [],
   "source": [
    "names = ['cyto', 'mito', 'nuc', 'sec']\n",
    "y_blind_names = []\n",
    "for i in range(len(y_blind)):\n",
    "    y_blind_names.append(names[y_blind[i]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 366,
   "metadata": {},
   "outputs": [],
   "source": [
    "d = {'SEQ ID':ID_list,'Predicted Class':y_blind_names}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 367,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_predictions = pd.DataFrame(d)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 368,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_predictions['Confidence'] = y_conf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 369,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SEQ ID</th>\n",
       "      <th>Predicted Class</th>\n",
       "      <th>Confidence</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>SEQ677</td>\n",
       "      <td>nuc</td>\n",
       "      <td>0.544929</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>SEQ231</td>\n",
       "      <td>cyto</td>\n",
       "      <td>0.417735</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>SEQ871</td>\n",
       "      <td>sec</td>\n",
       "      <td>0.754847</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>SEQ388</td>\n",
       "      <td>cyto</td>\n",
       "      <td>0.499552</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>SEQ122</td>\n",
       "      <td>nuc</td>\n",
       "      <td>0.586656</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>SEQ758</td>\n",
       "      <td>nuc</td>\n",
       "      <td>0.744499</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>SEQ333</td>\n",
       "      <td>sec</td>\n",
       "      <td>0.711209</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>SEQ937</td>\n",
       "      <td>sec</td>\n",
       "      <td>0.625908</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>SEQ351</td>\n",
       "      <td>nuc</td>\n",
       "      <td>0.749226</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>SEQ202</td>\n",
       "      <td>mito</td>\n",
       "      <td>0.477004</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>SEQ608</td>\n",
       "      <td>cyto</td>\n",
       "      <td>0.520254</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>SEQ402</td>\n",
       "      <td>cyto</td>\n",
       "      <td>0.568981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>SEQ433</td>\n",
       "      <td>cyto</td>\n",
       "      <td>0.745915</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>SEQ821</td>\n",
       "      <td>nuc</td>\n",
       "      <td>0.756427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>SEQ322</td>\n",
       "      <td>cyto</td>\n",
       "      <td>0.643423</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>SEQ982</td>\n",
       "      <td>nuc</td>\n",
       "      <td>0.472001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>SEQ951</td>\n",
       "      <td>cyto</td>\n",
       "      <td>0.512998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>SEQ173</td>\n",
       "      <td>cyto</td>\n",
       "      <td>0.501497</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>SEQ862</td>\n",
       "      <td>cyto</td>\n",
       "      <td>0.513225</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>SEQ224</td>\n",
       "      <td>cyto</td>\n",
       "      <td>0.451742</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    SEQ ID Predicted Class  Confidence\n",
       "0   SEQ677             nuc    0.544929\n",
       "1   SEQ231            cyto    0.417735\n",
       "2   SEQ871             sec    0.754847\n",
       "3   SEQ388            cyto    0.499552\n",
       "4   SEQ122             nuc    0.586656\n",
       "5   SEQ758             nuc    0.744499\n",
       "6   SEQ333             sec    0.711209\n",
       "7   SEQ937             sec    0.625908\n",
       "8   SEQ351             nuc    0.749226\n",
       "9   SEQ202            mito    0.477004\n",
       "10  SEQ608            cyto    0.520254\n",
       "11  SEQ402            cyto    0.568981\n",
       "12  SEQ433            cyto    0.745915\n",
       "13  SEQ821             nuc    0.756427\n",
       "14  SEQ322            cyto    0.643423\n",
       "15  SEQ982             nuc    0.472001\n",
       "16  SEQ951            cyto    0.512998\n",
       "17  SEQ173            cyto    0.501497\n",
       "18  SEQ862            cyto    0.513225\n",
       "19  SEQ224            cyto    0.451742"
      ]
     },
     "execution_count": 369,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 375,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\\begin{tabular}{llr}\n",
      "\\toprule\n",
      " SEQ ID & Predicted Class &  Confidence \\\\\n",
      "\\midrule\n",
      " SEQ677 &             nuc &    0.544929 \\\\\n",
      " SEQ231 &            cyto &    0.417735 \\\\\n",
      " SEQ871 &             sec &    0.754847 \\\\\n",
      " SEQ388 &            cyto &    0.499552 \\\\\n",
      " SEQ122 &             nuc &    0.586656 \\\\\n",
      " SEQ758 &             nuc &    0.744499 \\\\\n",
      " SEQ333 &             sec &    0.711209 \\\\\n",
      " SEQ937 &             sec &    0.625908 \\\\\n",
      " SEQ351 &             nuc &    0.749226 \\\\\n",
      " SEQ202 &            mito &    0.477004 \\\\\n",
      " SEQ608 &            cyto &    0.520254 \\\\\n",
      " SEQ402 &            cyto &    0.568981 \\\\\n",
      " SEQ433 &            cyto &    0.745915 \\\\\n",
      " SEQ821 &             nuc &    0.756427 \\\\\n",
      " SEQ322 &            cyto &    0.643423 \\\\\n",
      " SEQ982 &             nuc &    0.472001 \\\\\n",
      " SEQ951 &            cyto &    0.512998 \\\\\n",
      " SEQ173 &            cyto &    0.501497 \\\\\n",
      " SEQ862 &            cyto &    0.513225 \\\\\n",
      " SEQ224 &            cyto &    0.451742 \\\\\n",
      "\\bottomrule\n",
      "\\end{tabular}\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(df_predictions.to_latex(index=False))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}