tutorials/example_sequences.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from spectrae import Spectra, SpectraDataset\n",
    "from torchdrug import datasets\n",
    "import numpy as np\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "from tqdm import tqdm\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SPECTRA Sequence Tutorial\n",
    "\n",
    "### We take an example dataset from PEER, the YeastPPI dataset, taking the first 50 samples.\n",
    "### We define a continuous spectral property based on a sequence alignment. When comparing two samples, we perform a pairwise alignment between all proteins in each sample, and take the mean sequence similarity. \n",
    "### We define cross-split overlap as the average mean sequence similarity (as described for the spectral property) between all test and train samples."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "class PEER_Dataset(SpectraDataset):\n",
    "    \n",
    "    def parse(self, input_file):\n",
    "        if input_file == 'YeastPPI':\n",
    "            return datasets.YeastPPI(\"~/protein-datasets/\", atom_feature=None, bond_feature=None, residue_feature=\"default\")\n",
    "\n",
    "    def __len__(self):\n",
    "        #Using a truncated form of the dataset for sake of demonstration \n",
    "        return 50\n",
    "\n",
    "    def convert_graph_sequence(self, graph):\n",
    "        res = graph.residue_feature.argmax(axis=1)\n",
    "        id2residue = {v: k for k, v in graph.residue2id.items()}\n",
    "        sequence = ''.join([id2residue[res_id.item()] for res_id in res])\n",
    "        return sequence\n",
    "    \n",
    "    def sample_to_index(self, sample):\n",
    "        if not hasattr(self, 'index_to_sequence'):\n",
    "            print(\"Generating index to sequence\")\n",
    "            self.index_to_sequence = {}\n",
    "            for i in tqdm(range(len(self))):\n",
    "                x = self.__getitem__(i)\n",
    "                self.index_to_sequence[f'{x[0]}-{x[1]}'] = i\n",
    "\n",
    "        return self.index_to_sequence[sample]            \n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        data = self.samples[idx]\n",
    "        return [self.convert_graph_sequence(data['graph1']),self.convert_graph_sequence(data['graph2']) ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from Bio import Align\n",
    "\n",
    "class SPECTRA_sequence(Spectra):\n",
    "\n",
    "    def spectra_properties(self, sample_one, sample_two):\n",
    "\n",
    "        aligner = Align.PairwiseAligner()\n",
    "\n",
    "        aligner.match_score = 1.0\n",
    "        aligner.mismatch_score = -2.0\n",
    "        aligner.gap_score = -2.5\n",
    "\n",
    "        def sequence_similarity(seq1, seq2):\n",
    "\n",
    "            input = aligner.align(seq1, seq2)[0]\n",
    "            align_data = str(input).split('\\n')\n",
    "            i, j = align_data[0], align_data[2]\n",
    "\n",
    "            count = 0\n",
    "            diff = 0\n",
    "\n",
    "            for orig, n in zip(i, j):\n",
    "                if orig != n:\n",
    "                    diff += 1\n",
    "                count += 1\n",
    "\n",
    "            return 1 - diff/count\n",
    "\n",
    "        similarities = []\n",
    "        for i in sample_one:\n",
    "            for j in sample_two:\n",
    "                similarities.append(sequence_similarity(i, j))\n",
    "\n",
    "        return np.mean(similarities)\n",
    "\n",
    "    def cross_split_overlap(self, train, test):\n",
    "        average_similarity = []\n",
    "\n",
    "        for i in test:\n",
    "            for j in train:\n",
    "                num_i = self.dataset.sample_to_index(f'{i[0]}-{i[1]}')\n",
    "                num_j = self.dataset.sample_to_index(f'{j[0]}-{j[1]}')\n",
    "                average_similarity.append(self.spectra_properties(num_i, num_j))\n",
    "\n",
    "        return np.mean(average_similarity)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "15:35:48   Extracting /Users/yashaektefaie/protein-datasets/yeast_ppi.zip to /Users/yashaektefaie/protein-datasets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Constructing proteins from sequences: 100%|██████████| 38154/38154 [04:11<00:00, 152.00it/s]\n"
     ]
    }
   ],
   "source": [
    "PG_Dataset = PEER_Dataset('YeastPPI', 'YeastPPI')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "pg_spectra = SPECTRA_sequence(PG_Dataset, binary = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "File YEASTPPI_50_precalculated_spectra_properties already exists, set force_recalculate to True to recalculate\n"
     ]
    }
   ],
   "source": [
    "pg_spectra.pre_calculate_spectra_properties('YEASTPPI_50')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1225/1225 [00:00<00:00, 283258.31it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Stats for SPECTRA property graph (SPG)\n",
      "Number of nodes: 50\n",
      "Number of edges: 1225\n",
      "Number of connected components: 1\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.0 and dataset YeastPPI\n",
      "Number of samples in independent set: 50\n",
      "Train size: 40\tTest size: 10\n",
      "Generating index to sequence\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 50/50 [00:00<00:00, 561.01it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cross split overlap: 0.31203000651611407\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.0 and dataset YeastPPI\n",
      "Number of samples in independent set: 50\n",
      "Train size: 40\tTest size: 10\n",
      "Cross split overlap: 0.3198363151824079\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.0 and dataset YeastPPI\n",
      "Number of samples in independent set: 50\n",
      "Train size: 40\tTest size: 10\n",
      "Cross split overlap: 0.31663113475988003\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.05 and dataset YeastPPI\n",
      "Number of samples in independent set: 48\n",
      "Train size: 38\tTest size: 10\n",
      "Cross split overlap: 0.3243081729550318\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.05 and dataset YeastPPI\n",
      "Number of samples in independent set: 48\n",
      "Train size: 38\tTest size: 10\n",
      "Cross split overlap: 0.32297593363787014\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.05 and dataset YeastPPI\n",
      "Number of samples in independent set: 50\n",
      "Train size: 40\tTest size: 10\n",
      "Cross split overlap: 0.31663113475988003\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.1 and dataset YeastPPI\n",
      "Number of samples in independent set: 45\n",
      "Train size: 36\tTest size: 9\n",
      "Cross split overlap: 0.3156767081627318\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.1 and dataset YeastPPI\n",
      "Number of samples in independent set: 43\n",
      "Train size: 34\tTest size: 9\n",
      "Cross split overlap: 0.3141867561225539\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.1 and dataset YeastPPI\n",
      "Number of samples in independent set: 41\n",
      "Train size: 32\tTest size: 9\n",
      "Cross split overlap: 0.31278834014602847\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.15 and dataset YeastPPI\n",
      "Number of samples in independent set: 37\n",
      "Train size: 29\tTest size: 8\n",
      "Cross split overlap: 0.3159083386737628\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.15 and dataset YeastPPI\n",
      "Number of samples in independent set: 34\n",
      "Train size: 27\tTest size: 7\n",
      "Cross split overlap: 0.3131607766376352\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.15 and dataset YeastPPI\n",
      "Number of samples in independent set: 39\n",
      "Train size: 31\tTest size: 8\n",
      "Cross split overlap: 0.3156654118171175\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.2 and dataset YeastPPI\n",
      "Number of samples in independent set: 30\n",
      "Train size: 24\tTest size: 6\n",
      "Cross split overlap: 0.3095185480659898\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.2 and dataset YeastPPI\n",
      "Number of samples in independent set: 32\n",
      "Train size: 25\tTest size: 7\n",
      "Cross split overlap: 0.31028505867221573\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.2 and dataset YeastPPI\n",
      "Number of samples in independent set: 32\n",
      "Train size: 25\tTest size: 7\n",
      "Cross split overlap: 0.30480017666670206\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.25 and dataset YeastPPI\n",
      "Number of samples in independent set: 29\n",
      "Train size: 23\tTest size: 6\n",
      "Cross split overlap: 0.3082023215059947\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.25 and dataset YeastPPI\n",
      "Number of samples in independent set: 26\n",
      "Train size: 20\tTest size: 6\n",
      "Cross split overlap: 0.3049011208260666\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.25 and dataset YeastPPI\n",
      "Number of samples in independent set: 26\n",
      "Train size: 20\tTest size: 6\n",
      "Cross split overlap: 0.2955613720123994\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.3 and dataset YeastPPI\n",
      "Number of samples in independent set: 24\n",
      "Train size: 19\tTest size: 5\n",
      "Cross split overlap: 0.29933601786285363\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.3 and dataset YeastPPI\n",
      "Number of samples in independent set: 20\n",
      "Train size: 16\tTest size: 4\n",
      "Cross split overlap: 0.29012809950129065\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.3 and dataset YeastPPI\n",
      "Number of samples in independent set: 20\n",
      "Train size: 16\tTest size: 4\n",
      "Cross split overlap: 0.30021177556166867\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.35 and dataset YeastPPI\n",
      "Number of samples in independent set: 22\n",
      "Train size: 17\tTest size: 5\n",
      "Cross split overlap: 0.30053986929426335\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.35 and dataset YeastPPI\n",
      "Number of samples in independent set: 18\n",
      "Train size: 14\tTest size: 4\n",
      "Cross split overlap: 0.29317464448358915\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.35 and dataset YeastPPI\n",
      "Number of samples in independent set: 19\n",
      "Train size: 15\tTest size: 4\n",
      "Cross split overlap: 0.29716694296386137\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.4 and dataset YeastPPI\n",
      "Number of samples in independent set: 19\n",
      "Train size: 15\tTest size: 4\n",
      "Cross split overlap: 0.2907933276433827\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.4 and dataset YeastPPI\n",
      "Number of samples in independent set: 15\n",
      "Train size: 12\tTest size: 3\n",
      "Cross split overlap: 0.30794093112661\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.4 and dataset YeastPPI\n",
      "Number of samples in independent set: 13\n",
      "Train size: 10\tTest size: 3\n",
      "Cross split overlap: 0.2895169327745385\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.45 and dataset YeastPPI\n",
      "Number of samples in independent set: 17\n",
      "Train size: 13\tTest size: 4\n",
      "Cross split overlap: 0.2899094477246267\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.45 and dataset YeastPPI\n",
      "Number of samples in independent set: 13\n",
      "Train size: 10\tTest size: 3\n",
      "Cross split overlap: 0.30843452788155973\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.45 and dataset YeastPPI\n",
      "Number of samples in independent set: 15\n",
      "Train size: 12\tTest size: 3\n",
      "Cross split overlap: 0.3108147863983189\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.5 and dataset YeastPPI\n",
      "Number of samples in independent set: 13\n",
      "Train size: 10\tTest size: 3\n",
      "Cross split overlap: 0.29864789861538443\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.5 and dataset YeastPPI\n",
      "Number of samples in independent set: 13\n",
      "Train size: 10\tTest size: 3\n",
      "Cross split overlap: 0.28856667984295575\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.5 and dataset YeastPPI\n",
      "Number of samples in independent set: 13\n",
      "Train size: 10\tTest size: 3\n",
      "Cross split overlap: 0.29986320368533526\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.55 and dataset YeastPPI\n",
      "Number of samples in independent set: 10\n",
      "Train size: 8\tTest size: 2\n",
      "Cross split overlap: 0.27290131535031703\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.55 and dataset YeastPPI\n",
      "Number of samples in independent set: 11\n",
      "Train size: 8\tTest size: 3\n",
      "Cross split overlap: 0.29098669375926306\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.55 and dataset YeastPPI\n",
      "Number of samples in independent set: 12\n",
      "Train size: 9\tTest size: 3\n",
      "Cross split overlap: 0.28843486709696403\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.6 and dataset YeastPPI\n",
      "Number of samples in independent set: 11\n",
      "Train size: 8\tTest size: 3\n",
      "Cross split overlap: 0.28871298286343333\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.6 and dataset YeastPPI\n",
      "Number of samples in independent set: 9\n",
      "Train size: 7\tTest size: 2\n",
      "Cross split overlap: 0.28231093143282\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.6 and dataset YeastPPI\n",
      "Number of samples in independent set: 10\n",
      "Train size: 8\tTest size: 2\n",
      "Cross split overlap: 0.2920481887668393\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.65 and dataset YeastPPI\n",
      "Number of samples in independent set: 9\n",
      "Train size: 7\tTest size: 2\n",
      "Cross split overlap: 0.2932272870650309\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.65 and dataset YeastPPI\n",
      "Number of samples in independent set: 8\n",
      "Train size: 6\tTest size: 2\n",
      "Cross split overlap: 0.2817129257790943\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.65 and dataset YeastPPI\n",
      "Number of samples in independent set: 9\n",
      "Train size: 7\tTest size: 2\n",
      "Cross split overlap: 0.26354445184823994\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.7 and dataset YeastPPI\n",
      "Number of samples in independent set: 8\n",
      "Train size: 6\tTest size: 2\n",
      "Cross split overlap: 0.277677673615444\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.7 and dataset YeastPPI\n",
      "Number of samples in independent set: 8\n",
      "Train size: 6\tTest size: 2\n",
      "Cross split overlap: 0.2914835856229439\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.7 and dataset YeastPPI\n",
      "Number of samples in independent set: 8\n",
      "Train size: 6\tTest size: 2\n",
      "Cross split overlap: 0.28820872747967236\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.75 and dataset YeastPPI\n",
      "Number of samples in independent set: 6\n",
      "Train size: 4\tTest size: 2\n",
      "Cross split overlap: 0.2906530368694926\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.75 and dataset YeastPPI\n",
      "Number of samples in independent set: 7\n",
      "Train size: 5\tTest size: 2\n",
      "Cross split overlap: 0.28201052198981585\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.75 and dataset YeastPPI\n",
      "Number of samples in independent set: 6\n",
      "Train size: 4\tTest size: 2\n",
      "Cross split overlap: 0.2863110411650706\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.8 and dataset YeastPPI\n",
      "Split for SP_0.80_0 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 0.8 and dataset YeastPPI\n",
      "Split for SP_0.80_1 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 0.8 and dataset YeastPPI\n",
      "Number of samples in independent set: 6\n",
      "Train size: 4\tTest size: 2\n",
      "Cross split overlap: 0.2450612832228558\n",
      "\n",
      "\n",
      "\n",
      "Generating SPECTRA split for spectral parameter 0.85 and dataset YeastPPI\n",
      "Split for SP_0.85_0 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 0.85 and dataset YeastPPI\n",
      "Split for SP_0.85_1 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 0.85 and dataset YeastPPI\n",
      "Split for SP_0.85_2 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 0.9 and dataset YeastPPI\n",
      "Split for SP_0.90_0 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 0.9 and dataset YeastPPI\n",
      "Split for SP_0.90_1 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 0.9 and dataset YeastPPI\n",
      "Split for SP_0.90_2 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 0.95 and dataset YeastPPI\n",
      "Split for SP_0.95_0 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 0.95 and dataset YeastPPI\n",
      "Split for SP_0.95_1 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 0.95 and dataset YeastPPI\n",
      "Split for SP_0.95_2 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 1.0 and dataset YeastPPI\n",
      "Split for SP_1.00_0 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 1.0 and dataset YeastPPI\n",
      "Split for SP_1.00_1 could not be generated since independent set only has one sample\n",
      "Generating SPECTRA split for spectral parameter 1.0 and dataset YeastPPI\n",
      "Split for SP_1.00_2 could not be generated since independent set only has one sample\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "spectra_parameters = {'number_repeats': 3, \n",
    "                      'random_seed': [42, 44, 46],\n",
    "                      'spectral_parameters': [\"{:.2f}\".format(i) for i in np.arange(0, 1.05, 0.05)],\n",
    "                      'force_reconstruct': True,\n",
    "                                              }\n",
    "\n",
    "pg_spectra.generate_spectra_splits(**spectra_parameters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x14c9a51c0>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "stats = pg_spectra.return_all_split_stats()\n",
    "plt.scatter(stats['SPECTRA_parameter'], stats['cross_split_overlap'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x14c8d40a0>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(stats['SPECTRA_parameter'], stats['train_size'])\n",
    "plt.scatter(stats['SPECTRA_parameter'], stats['test_size']) "
   ]
  }
 ],
 "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": "3.9.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}