ipynb/Figure3-rerun.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/jmorton/miniconda3/envs/qiime2-2019.10/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:526: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint8 = np.dtype([(\"qint8\", np.int8, 1)])\n",
      "/Users/jmorton/miniconda3/envs/qiime2-2019.10/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:527: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint8 = np.dtype([(\"quint8\", np.uint8, 1)])\n",
      "/Users/jmorton/miniconda3/envs/qiime2-2019.10/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:528: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint16 = np.dtype([(\"qint16\", np.int16, 1)])\n",
      "/Users/jmorton/miniconda3/envs/qiime2-2019.10/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:529: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint16 = np.dtype([(\"quint16\", np.uint16, 1)])\n",
      "/Users/jmorton/miniconda3/envs/qiime2-2019.10/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:530: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint32 = np.dtype([(\"qint32\", np.int32, 1)])\n",
      "/Users/jmorton/miniconda3/envs/qiime2-2019.10/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:535: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  np_resource = np.dtype([(\"resource\", np.ubyte, 1)])\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import tensorflow as tf\n",
    "from biom import load_table\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "#from gneiss.util import match\n",
    "from scipy.spatial import KDTree\n",
    "from scipy.stats import rankdata\n",
    "from skbio.stats.composition import clr, clr_inv, centralize\n",
    "from sklearn.neighbors import NearestNeighbors, KDTree\n",
    "from skbio.stats.ordination import OrdinationResults\n",
    "from scipy.spatial.distance import euclidean\n",
    "from scipy.stats import spearmanr\n",
    "from src.evaluate import _edge_roc_curve\n",
    "\n",
    "\n",
    "from matplotlib import rc\n",
    "import matplotlib\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prop_matrix_soil_PHI.csv      prop_matrix_soil_pearson.csv\r\n",
      "prop_matrix_soil_RHO.csv      prop_matrix_soil_spearman.csv\r\n"
     ]
    }
   ],
   "source": [
    "!ls ../results/soil_output/rerun"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Read in the clr transformed correlation metrics generated using `figure3-rerun.R` originally developed from https://www.biorxiv.org/content/10.1101/847475v1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "tools = ['mmvec', 'spearman', 'pearson', 'phi', 'rho']\n",
    "fnames = [\n",
    "    'mmvec_ranks.csv',\n",
    "    'rerun/prop_matrix_soil_pearson.csv',\n",
    "    'rerun/prop_matrix_soil_spearman.csv',\n",
    "    'rerun/prop_matrix_soil_PHI.csv',\n",
    "    'rerun/prop_matrix_soil_RHO.csv'\n",
    "]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Focus on Microcoleus and their respective metabolites"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "select_microbes = ['rplo 1 (Cyanobacteria)', 'rplo 2 (Firmicutes)', \n",
    "                   'rplo 60 (Firmicutes)', 'rplo 7 (Actinobacteria)']\n",
    "microcoleus_metabolites = ['(3-methyladenine)', '7-methyladenine', '4-guanidinobutanoate', 'uracil',\n",
    "                           'xanthine', 'hypoxanthine', '(N6-acetyl-lysine)', 'cytosine',\n",
    "                           'N-acetylornithine', 'N-acetylornithine', 'succinate', \n",
    "                           'adenosine', 'guanine', 'adenine']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Read in the mmvec conditional probabilities (called ranks here)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "ranks = pd.read_csv(os.path.join('../results/soil_output', fnames[0]), index_col=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Functions for calculating sensitivity, specificity, precision, recall and F1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "sensitivity = lambda x: x['TP'] / (x['TP'] + x['FN'])\n",
    "specificity = lambda x: x['TN'] / (x['TN'] + x['FP'])\n",
    "precision = lambda x: x['TP'] / (x['TP'] + x['FP'])\n",
    "recall = lambda x: x['TP'] / (x['TP'] + x['FN'])\n",
    "f1 = lambda x: 2 * x['precision'] * x['recall'] / (x['precision'] + x['recall'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Generate ground truth edge list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/jmorton/miniconda3/envs/qiime2-2019.10/lib/python3.6/site-packages/ipykernel_launcher.py:13: FutureWarning: Sorting because non-concatenation axis is not aligned. A future version\n",
      "of pandas will change to not sort by default.\n",
      "\n",
      "To accept the future behavior, pass 'sort=False'.\n",
      "\n",
      "To retain the current behavior and silence the warning, pass 'sort=True'.\n",
      "\n",
      "  del sys.path[0]\n"
     ]
    }
   ],
   "source": [
    "edges = pd.DataFrame(list(zip(['rplo 1 (Cyanobacteria)'] * len(microcoleus_metabolites), \n",
    "                     microcoleus_metabolites)), columns=['microbe', 'metabolite'])\n",
    "edges['direction'] = 'R'\n",
    "\n",
    "microbes = list(ranks.columns)\n",
    "metabolites = list(ranks.columns)\n",
    "other_metabolites = list(set(metabolites) - set(microcoleus_metabolites))\n",
    "\n",
    "other_edges = pd.DataFrame(other_metabolites, columns=['metabolite'])\n",
    "other_edges['microbe'] = 'rplo 1 (Cyanobacteria)'\n",
    "other_edges['direction'] = 0\n",
    "\n",
    "edges = pd.concat((edges, other_edges), axis=0)\n",
    "edges = edges[['microbe', 'metabolite', 'direction']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Read in all of the correlation results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def read_df(fname):\n",
    "    if '.txt' in fname:\n",
    "        return pd.read_table(os.path.join('../results/soil_output', fname), \n",
    "                                          index_col=0)\n",
    "    else:\n",
    "        return pd.read_csv(os.path.join('../results/soil_output', fname), \n",
    "                                        index_col=0)\n",
    "    \n",
    "rankz = list(map(read_df, fnames))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compute ROC curves form Microcoleus for all correlation methods."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def roc_f(ranks, tool):\n",
    "    r = ranks.T[['rplo 1 (Cyanobacteria)']].loc[metabolites].T\n",
    "    if tool == 'phi':\n",
    "        r = 1 / r\n",
    "        \n",
    "    res = _edge_roc_curve(r, edges, k_max=40)\n",
    "    return res\n",
    "rocs = list(map(lambda x: roc_f(x[0], x[1]), zip(rankz, tools)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/jmorton/miniconda3/envs/qiime2-2019.10/lib/python3.6/site-packages/ipykernel_launcher.py:1: RuntimeWarning: invalid value encountered in long_scalars\n",
      "  \"\"\"Entry point for launching an IPython kernel.\n",
      "/Users/jmorton/miniconda3/envs/qiime2-2019.10/lib/python3.6/site-packages/ipykernel_launcher.py:4: RuntimeWarning: invalid value encountered in double_scalars\n",
      "  after removing the cwd from sys.path.\n",
      "/Users/jmorton/miniconda3/envs/qiime2-2019.10/lib/python3.6/site-packages/ipykernel_launcher.py:5: RuntimeWarning: invalid value encountered in double_scalars\n",
      "  \"\"\"\n"
     ]
    }
   ],
   "source": [
    "for res in rocs:\n",
    "    for df in res:\n",
    "        df['TPR'] = df.apply(sensitivity, axis=1)\n",
    "        df['TNR'] = df.apply(specificity, axis=1)\n",
    "        df['precision'] = df.apply(precision, axis=1)\n",
    "        df['recall'] = df.apply(recall, axis=1)\n",
    "        df['f1_score'] = df.apply(f1, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "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": [
    "fig, ax = plt.subplots(figsize=(6, 4))\n",
    "#plt.subplots_adjust(wspace=0.4, hspace=0.5)\n",
    "\n",
    "for tool, roc in zip(tools, rocs):\n",
    "    ax.plot(roc[0]['TP'], label=tool)            \n",
    "    ax.set_ylabel('True positives', fontsize=18)\n",
    "    ax.set_xlabel('top K hits', fontsize=18)\n",
    "    ax.tick_params(axis=\"x\", labelsize=14)\n",
    "    ax.tick_params(axis=\"y\", labelsize=14)\n",
    "    ax.xaxis.set_major_locator(plt.MaxNLocator(4))\n",
    "    ax.yaxis.set_major_locator(plt.MaxNLocator(4))      \n",
    "\n",
    "ax.set_title('Microcoleus molecular detection rate', fontsize=18)\n",
    "ax.legend(loc='center left', bbox_to_anchor=(1, 0.5), fontsize=14)\n",
    "fig.savefig('Figure3b.svg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}