TACTIC_analysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# TACTIC Analyses\n",
    "This notebook contains the code for all analyses conducted in the TACTIC manuscript."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Restore point (runtime: ~1 minute)\n",
    "import pickle\n",
    "from urllib.request import urlopen\n",
    "link = 'https://www.dropbox.com/scl/fi/38ewb000ihqcg6wev41g3/TACTIC_data_master.pkl?rlkey=zvxnt71ejj5ucexgpr1plmyvg&dl=1'\n",
    "ecoli, mtb, master, out = pickle.load(urlopen(link))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 0. Set Up Environment\n",
    "Import dependencies and load datasets."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "### 0.1 Import dependencies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import dependencies\n",
    "import cv2\n",
    "import math\n",
    "import pickle\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from tqdm import tqdm\n",
    "from scipy import sparse\n",
    "from scipy.stats import spearmanr, ttest_ind, hypergeom, chi2_contingency\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from itertools import compress, combinations, chain\n",
    "from statsmodels.stats.multitest import multipletests\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Import relevant function from INDIGOpy\n",
    "from indigopy.core import featurize, classify"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 0.2 Load datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load datasets\n",
    "omics_data  = {\n",
    "    'ecoli': pd.read_excel('./data/omics_data.xlsx', sheet_name='ecoli_chemogenomics', engine='openpyxl'), \n",
    "    'mtb': pd.read_excel('./data/omics_data.xlsx', sheet_name='mtb_transcriptomics', engine='openpyxl')\n",
    "}\n",
    "ixn_data    = pd.read_excel('./data/ixn_data.xlsx', sheet_name=None, engine='openpyxl')\n",
    "orthology   = {\n",
    "    'ecoli': pd.read_csv('./data/OrtholugeDB/Ec-MG1655_pathogens.txt', sep='\\t'), \n",
    "    'mtb': pd.read_csv('./data/OrtholugeDB/Mtb-H37Rv_pathogens.txt', sep='\\t')\n",
    "}\n",
    "\n",
    "# Define strain-specific orthology data\n",
    "strains         = sorted(set(ixn_data['data']['Strain']))\n",
    "strains         = strains + ['M. abscessus', 'M. smegmatis']\n",
    "jx              = [1, 3] + [x for x in range(4, orthology['ecoli'].shape[1], 3)]\n",
    "df1, df2        = orthology['ecoli'].iloc[:, jx], orthology['mtb'].iloc[:, jx]\n",
    "df1.columns     = [s.split(' - ')[0] if 'comparison' in s else s for s in df1.columns]\n",
    "df2.columns     = [s.split(' - ')[0] if 'comparison' in s else s for s in df2.columns]\n",
    "df2.iloc[:, 0]  = [s.replace('RVBD_', 'Rv') for s in df2.iloc[:, 0]]\n",
    "emap, mmap      = dict(), dict()\n",
    "for strain in strains: \n",
    "    # E. coli strains\n",
    "    if strain.endswith('iAi1'): \n",
    "        s1 = [s for s in df1.columns if s.endswith('IAI1')][0]\n",
    "        s2 = [s for s in df2.columns if s.endswith('IAI1')][0]\n",
    "    elif strain.startswith('E. coli'): \n",
    "        s1 = [s for s in df1.columns if 'MG1655' in s][0]\n",
    "        s2 = [s for s in df2.columns if 'MG1655' in s][0]\n",
    "    # M. tb strains\n",
    "    elif 'tuberculosis' in strain: \n",
    "        s1 = [s for s in df1.columns if 'H37Rv' in s][0]\n",
    "        s2 = [s for s in df2.columns if 'H37Rv' in s][0]\n",
    "    # TACTIC strains\n",
    "    elif 'aureus' in strain: \n",
    "        s1 = [s for s in df1.columns if 'aureus' in s][0]\n",
    "        s2 = [s for s in df2.columns if 'aureus' in s][0]\n",
    "    elif 'baumannii' in strain: \n",
    "        s1 = [s for s in df1.columns if 'baumannii' in s][0]\n",
    "        s2 = [s for s in df2.columns if 'baumannii' in s][0]\n",
    "    elif strain.endswith('PAO1'): \n",
    "        s1 = [s for s in df1.columns if s.endswith('PAO1')][0]\n",
    "        s2 = [s for s in df2.columns if s.endswith('PAO1')][0]\n",
    "    elif strain.endswith('PA14'): \n",
    "        s1 = [s for s in df1.columns if s.endswith('PA14')][0]\n",
    "        s2 = [s for s in df2.columns if s.endswith('PA14')][0]\n",
    "    elif strain.endswith('14028s'): \n",
    "        s1 = [s for s in df1.columns if s.endswith('14028S')][0]\n",
    "        s2 = [s for s in df2.columns if s.endswith('14028S')][0]\n",
    "    elif strain.endswith('LT2'): \n",
    "        s1 = [s for s in df1.columns if s.endswith('LT2')][0]\n",
    "        s2 = [s for s in df2.columns if s.endswith('LT2')][0]\n",
    "    # NTM strains\n",
    "    elif 'abscessus' in strain: \n",
    "        s1 = [s for s in df1.columns if 'abscessus' in s][0]\n",
    "        s2 = [s for s in df2.columns if 'abscessus' in s][0]\n",
    "    elif 'smegmatis' in strain: \n",
    "        s1 = [s for s in df1.columns if 'smegmatis' in s][0]\n",
    "        s2 = [s for s in df2.columns if 'smegmatis' in s][0]\n",
    "    emap[strain] = list(compress(df1.iloc[:, 0], df1[s1] != '0'))\n",
    "    mmap[strain] = list(compress(df2.iloc[:, 0], df2[s2] != '0'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Checkpoint\n",
    "with open('./TACTIC_data.pkl', 'wb') as f: \n",
    "    pickle.dump([ecoli, mtb, master, out], f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Define Inputs and Outputs\n",
    "Define main input (*ecoli*, *mtb*, *master*) and output (*out*) variables."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.1 Instantiate variables (runtime: ~2 minutes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 413 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 2965/2965 [00:11<00:00, 264.48it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 12/12 [00:58<00:00,  4.91s/it]\n",
      "Defining INDIGO features: 100%|██████████| 2965/2965 [00:15<00:00, 192.02it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 12/12 [00:33<00:00,  2.79s/it]\n"
     ]
    }
   ],
   "source": [
    "# I/O variables\n",
    "ecoli, mtb, out = dict(), dict(), dict()\n",
    "\n",
    "# Store orthology info\n",
    "ecoli['map'], mtb['map'] = emap, mmap\n",
    "\n",
    "# E. coli chemogenomic-based data\n",
    "ecoli['key']    = list(zip(ixn_data['drug_key']['Code'], ixn_data['drug_key']['Chemogenomic_label']))\n",
    "ecoli['key']    = [key for key in ecoli['key'] if not pd.isna(key[1])]\n",
    "ecoli['omics']  = omics_data['ecoli']\n",
    "profiles        = ecoli['omics'].iloc[:, 1:].to_dict('list')\n",
    "feature_names   = ecoli['omics']['Gene'].tolist()\n",
    "interactions    = [list(s.split(', ')) for s in ixn_data['data']['Interaction']]\n",
    "feature_data    = featurize(interactions, profiles, feature_names=feature_names, key=ecoli['key'], \n",
    "                            strains=ixn_data['data']['Strain'].tolist(), orthology_map=ecoli['map'])\n",
    "ecoli['idx']    = feature_data['idx']\n",
    "ecoli['df']     = feature_data['feature_df']\n",
    "\n",
    "# M. tb transcriptomic-based data\n",
    "mtb['key']      = list(zip(ixn_data['drug_key']['Code'], ixn_data['drug_key']['Transcriptomic_label']))\n",
    "mtb['omics']    = omics_data['mtb']\n",
    "profiles        = mtb['omics'].iloc[:, 1:].to_dict('list')\n",
    "feature_names   = mtb['omics']['Gene'].tolist()\n",
    "interactions    = [list(s.split(', ')) for s in ixn_data['data']['Interaction']]\n",
    "feature_data    = featurize(interactions, profiles, feature_names=feature_names, key=mtb['key'], \n",
    "                            strains=ixn_data['data']['Strain'].tolist(), orthology_map=mtb['map'])\n",
    "mtb['idx']      = feature_data['idx']\n",
    "mtb['df']       = feature_data['feature_df']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.2 Define feature data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define master data\n",
    "ecoli_genes     = [g for g in ecoli['omics']['Gene'].tolist() if g not in ecoli['map']['M. tuberculosis H37Rv']]\n",
    "idx             = [any(i.split('_')[0].endswith(gene) for gene in ecoli_genes) for i in list(ecoli['df'].index)]\n",
    "master          = {'data': ixn_data['data'], 'df': pd.concat([ecoli['df'].loc[idx], mtb['df']])}\n",
    "\n",
    "# Define ML inputs\n",
    "ecoli['X']      = sparse.csr_matrix(ecoli['df'].to_numpy().transpose())\n",
    "mtb['X']        = sparse.csr_matrix(mtb['df'].to_numpy().transpose())\n",
    "master['X']     = sparse.csr_matrix(master['df'].to_numpy().transpose())\n",
    "\n",
    "# Define output data fields\n",
    "out['strains']  = sorted(list(set(master['data']['Strain'])))\n",
    "out['species']  = sorted(list(set([' '.join(s.split(' ')[0:2]) for s in master['data']['Strain'].tolist()])))\n",
    "out['drugs']    = sorted(list(set(chain.from_iterable([list(combo.split(', ')) for combo in master['data']['Interaction']]))))\n",
    "out['model']    = dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Checkpoint\n",
    "with open('./TACTIC_data.pkl', 'wb') as f: \n",
    "    pickle.dump([ecoli, mtb, master, out], f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. TACTIC Benchmarking\n",
    "\n",
    "TACTIC benchmarking against [INDIGO](https://doi.org/10.15252/msb.20156777) and [INDIGO-Mtb](https://doi.org/10.1128/mbio.02627-19) in predicting drug interaction outcomes for *A. baumannii*, *E. coli*, *P. aeruginosa*, *S.* Typhimurium, *S. aureus*, and *M. tuberculosis* (Mtb).  \n",
    "\n",
    "- Used **holdout validation**: for each species-specific data subset, the data for a strain of interest is held out for testing  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.1 I/O"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define I/O variables\n",
    "idx             = master['data']['Set'] == 'TACTIC'\n",
    "data            = master['data'].loc[idx]\n",
    "strains         = sorted(list(set(data['Strain'])))\n",
    "X1, X2, X       = ecoli['X'][idx], mtb['X'][idx], master['X'][idx]\n",
    "out['bench']    = dict()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2 Model training (runtime: ~1 minute)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define INDIGO (E. coli) model\n",
    "idx1    = master['data']['Set'] == 'ecoli'\n",
    "data1   = master['data'].loc[idx1]\n",
    "X1tr    = ecoli['X'][idx1]\n",
    "m1      = RandomForestRegressor()\n",
    "m1.fit(X1tr, data1['Score'].tolist())\n",
    "out['model']['EC'] = m1\n",
    "\n",
    "# Define INDIGO-Mtb (M. tb) model\n",
    "idx2    = master['data']['Set'] == 'mtb'\n",
    "data2   = master['data'].loc[idx2]\n",
    "X2tr    = mtb['X'][idx2]\n",
    "m2      = RandomForestRegressor()\n",
    "m2.fit(X2tr, data2['Score'].tolist())\n",
    "out['model']['TB'] = m2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.3 Model evaluation (runtime: ~3.5 hours)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Determine predictions for INDIGO (E. coli) model\n",
    "out['bench']['EC'] = out['model']['EC'].predict(X1)\n",
    "\n",
    "# Determine predictions for INDIGO-Mtb (M. tb) model\n",
    "out['bench']['TB'] = out['model']['TB'].predict(X2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO::   0%|          | 0/12 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: A. baumannii ATCC 17978, N = 45 (1.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO::   8%|▊         | 1/12 [18:50<3:27:11, 1130.13s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.28, p = 0.0611\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = -0.01, p = 0.952\n",
      "\t\tHoldout results (TACTIC):\tR = 0.56, p = 7.51e-05\n",
      "\tStrain: E. coli BW25113, N = 473 (18.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO::  17%|█▋        | 2/12 [34:45<2:51:13, 1027.40s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.24, p = 1.53e-07\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = -0.01, p = 0.88\n",
      "\t\tHoldout results (TACTIC):\tR = 0.51, p = 5.19e-32\n",
      "\tStrain: E. coli MC4100, N = 98 (3.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO::  25%|██▌       | 3/12 [54:18<2:44:03, 1093.67s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.28, p = 0.00508\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = -0.03, p = 0.79\n",
      "\t\tHoldout results (TACTIC):\tR = 0.28, p = 0.00449\n",
      "\tStrain: E. coli MG1655, N = 452 (17.6%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO::  33%|███▎      | 4/12 [1:08:11<2:12:07, 990.89s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.26, p = 2.15e-08\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = -0.09, p = 0.047\n",
      "\t\tHoldout results (TACTIC):\tR = 0.38, p = 8.67e-17\n",
      "\tStrain: E. coli iAi1, N = 316 (12.3%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO::  42%|████▏     | 5/12 [1:25:37<1:57:56, 1010.93s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.17, p = 0.00181\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.15, p = 0.00791\n",
      "\t\tHoldout results (TACTIC):\tR = 0.63, p = 6.14e-36\n",
      "\tStrain: M. tuberculosis ATCC 35801, N = 154 (6.0%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO::  50%|█████     | 6/12 [1:44:10<1:44:33, 1045.63s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.04, p = 0.662\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.35, p = 9.88e-06\n",
      "\t\tHoldout results (TACTIC):\tR = 0.16, p = 0.0519\n",
      "\tStrain: M. tuberculosis H37Rv, N = 159 (6.2%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO::  58%|█████▊    | 7/12 [2:00:39<1:25:34, 1026.98s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.25, p = 0.00155\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.3, p = 9.27e-05\n",
      "\t\tHoldout results (TACTIC):\tR = 0.17, p = 0.0275\n",
      "\tStrain: P. aeruginosa PA14, N = 163 (6.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO::  67%|██████▋   | 8/12 [2:18:58<1:09:59, 1049.98s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.06, p = 0.421\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.17, p = 0.0342\n",
      "\t\tHoldout results (TACTIC):\tR = 0.52, p = 1.82e-12\n",
      "\tStrain: P. aeruginosa PAO1, N = 163 (6.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO::  75%|███████▌  | 9/12 [2:37:39<53:36, 1072.07s/it]  "
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = -0.13, p = 0.111\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = -0.07, p = 0.358\n",
      "\t\tHoldout results (TACTIC):\tR = 0.41, p = 8.01e-08\n",
      "\tStrain: S. aureus ATCC 29213, N = 45 (1.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO::  83%|████████▎ | 10/12 [2:57:01<36:39, 1099.79s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.23, p = 0.135\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.16, p = 0.294\n",
      "\t\tHoldout results (TACTIC):\tR = 0.2, p = 0.188\n",
      "\tStrain: S. typhimurium 14028s, N = 248 (9.7%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO::  92%|█████████▏| 11/12 [3:14:42<18:08, 1088.01s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.05, p = 0.398\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.0, p = 0.987\n",
      "\t\tHoldout results (TACTIC):\tR = 0.63, p = 4.03e-29\n",
      "\tStrain: S. typhimurium LT2, N = 248 (9.7%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO:: 100%|██████████| 12/12 [3:32:19<00:00, 1061.60s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.08, p = 0.189\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.07, p = 0.307\n",
      "\t\tHoldout results (TACTIC):\tR = 0.67, p = 1.07e-33\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# Evaluate TACTIC vs. INDIGO\n",
    "N = data.shape[0]\n",
    "for strain in tqdm(strains, desc='Evaluating TACTIC vs. INDIGO:'): \n",
    "    # Test set (strain of interest)\n",
    "    test_ix         = [s == strain for s in data['Strain'].tolist()]\n",
    "    test_ixns       = list(compress(data['Interaction'], test_ix))\n",
    "    test_scores     = list(compress(data['Score'], test_ix))\n",
    "    Xte, n          = X[test_ix], test_ix.count(True)\n",
    "    print('\\tStrain: {}, N = {} ({}%)'.format(strain, n, round(100*n/N, 1)))\n",
    "    # Train set for TACTIC model\n",
    "    train_ix        = [not i for i in test_ix]\n",
    "    if strain==data1['Strain'].unique()[0]: \n",
    "        extra_X     = master['X'][idx2]\n",
    "        extra_s     = master['data']['Score'][idx2].tolist()\n",
    "    elif strain==data2['Strain'].unique()[0]: \n",
    "        extra_X     = master['X'][idx1]\n",
    "        extra_s     = master['data']['Score'][idx1].tolist()\n",
    "    else: \n",
    "        extra_X     = master['X'][idx1 | idx2]\n",
    "        extra_s     = master['data']['Score'][idx1 | idx2].tolist()\n",
    "    Xtr             = sparse.vstack((extra_X, X[train_ix]))\n",
    "    train_scores    = extra_s + list(compress(data['Score'], train_ix))\n",
    "    # Instantiate results\n",
    "    df = pd.DataFrame({'Ixn': test_ixns, 'Score': test_scores, \n",
    "                       'EC': list(compress(out['bench']['EC'], test_ix)), \n",
    "                       'TB': list(compress(out['bench']['TB'], test_ix))})\n",
    "    # Run TACTIC\n",
    "    model           = RandomForestRegressor()\n",
    "    model.fit(Xtr, train_scores)\n",
    "    df['TACTIC'] = model.predict(Xte)\n",
    "    # Print results\n",
    "    r, p = spearmanr(df['Score'], df['EC'])\n",
    "    print('\\t\\tHoldout results (INDIGO):\\tR = {}, p = {:.3g}'.format(round(r, 2), p))\n",
    "    r, p = spearmanr(df['Score'], df['TB'])\n",
    "    print('\\t\\tHoldout results (INDIGO-Mtb):\\tR = {}, p = {:.3g}'.format(round(r, 2), p))\n",
    "    r, p = spearmanr(df['Score'], df['TACTIC'])\n",
    "    print('\\t\\tHoldout results (TACTIC):\\tR = {}, p = {:.3g}'.format(round(r, 2), p))\n",
    "    # Store results\n",
    "    out['bench'][strain] = df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.4 Model comparison (~15 minutes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating alternative ML algorithms:   0%|          | 0/5 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: A. baumannii ATCC 17978, N = 45 (1.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli BW25113, N = 473 (18.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli MC4100, N = 98 (3.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli MG1655, N = 452 (17.6%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli iAi1, N = 316 (12.3%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: M. tuberculosis ATCC 35801, N = 154 (6.0%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: M. tuberculosis H37Rv, N = 159 (6.2%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: P. aeruginosa PA14, N = 163 (6.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: P. aeruginosa PAO1, N = 163 (6.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. aureus ATCC 29213, N = 45 (1.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. typhimurium 14028s, N = 248 (9.7%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. typhimurium LT2, N = 248 (9.7%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating alternative ML algorithms:  20%|██        | 1/5 [03:20<13:20, 200.19s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: A. baumannii ATCC 17978, N = 45 (1.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli BW25113, N = 473 (18.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli MC4100, N = 98 (3.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli MG1655, N = 452 (17.6%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli iAi1, N = 316 (12.3%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: M. tuberculosis ATCC 35801, N = 154 (6.0%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: M. tuberculosis H37Rv, N = 159 (6.2%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: P. aeruginosa PA14, N = 163 (6.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: P. aeruginosa PAO1, N = 163 (6.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. aureus ATCC 29213, N = 45 (1.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. typhimurium 14028s, N = 248 (9.7%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. typhimurium LT2, N = 248 (9.7%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating alternative ML algorithms:  40%|████      | 2/5 [04:29<06:09, 123.19s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: A. baumannii ATCC 17978, N = 45 (1.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli BW25113, N = 473 (18.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli MC4100, N = 98 (3.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli MG1655, N = 452 (17.6%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli iAi1, N = 316 (12.3%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: M. tuberculosis ATCC 35801, N = 154 (6.0%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: M. tuberculosis H37Rv, N = 159 (6.2%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: P. aeruginosa PA14, N = 163 (6.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: P. aeruginosa PAO1, N = 163 (6.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. aureus ATCC 29213, N = 45 (1.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. typhimurium 14028s, N = 248 (9.7%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. typhimurium LT2, N = 248 (9.7%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating alternative ML algorithms:  60%|██████    | 3/5 [10:28<07:41, 230.75s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: A. baumannii ATCC 17978, N = 45 (1.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli BW25113, N = 473 (18.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli MC4100, N = 98 (3.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli MG1655, N = 452 (17.6%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli iAi1, N = 316 (12.3%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: M. tuberculosis ATCC 35801, N = 154 (6.0%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: M. tuberculosis H37Rv, N = 159 (6.2%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: P. aeruginosa PA14, N = 163 (6.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: P. aeruginosa PAO1, N = 163 (6.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. aureus ATCC 29213, N = 45 (1.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. typhimurium 14028s, N = 248 (9.7%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. typhimurium LT2, N = 248 (9.7%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating alternative ML algorithms:  80%|████████  | 4/5 [12:55<03:17, 197.96s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: A. baumannii ATCC 17978, N = 45 (1.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli BW25113, N = 473 (18.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli MC4100, N = 98 (3.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli MG1655, N = 452 (17.6%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: E. coli iAi1, N = 316 (12.3%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: M. tuberculosis ATCC 35801, N = 154 (6.0%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: M. tuberculosis H37Rv, N = 159 (6.2%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: P. aeruginosa PA14, N = 163 (6.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: P. aeruginosa PAO1, N = 163 (6.4%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. aureus ATCC 29213, N = 45 (1.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. typhimurium 14028s, N = 248 (9.7%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": []
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: S. typhimurium LT2, N = 248 (9.7%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating alternative ML algorithms: 100%|██████████| 5/5 [18:01<00:00, 216.38s/it]\n"
     ]
    }
   ],
   "source": [
    "# Check other ML algorithms\n",
    "from sklearn.linear_model import LinearRegression, Ridge\n",
    "from sklearn.svm import SVR \n",
    "from sklearn.tree import DecisionTreeRegressor\n",
    "from sklearn.ensemble import GradientBoostingRegressor\n",
    "\n",
    "# Loop through ML algorithms\n",
    "MLcomp = dict()\n",
    "algs = [LinearRegression(), Ridge(), SVR(), DecisionTreeRegressor(), GradientBoostingRegressor()]\n",
    "for alg in tqdm(algs, desc='Evaluating alternative ML algorithms'): \n",
    "    # Define ML label\n",
    "    label = str(alg)[:-2]\n",
    "    MLcomp[label] = dict()\n",
    "    # INDIGO\n",
    "    m1 = alg.fit(X1tr, data1['Score'].tolist())\n",
    "    MLcomp[label]['EC'] = m1.predict(X1)\n",
    "    # INDIGO-Mtb\n",
    "    m2 = alg.fit(X2tr, data2['Score'].tolist())\n",
    "    MLcomp[label]['TB'] = m2.predict(X2)\n",
    "    # TACTIC\n",
    "    MLcomp[label]['TACTIC'] = np.empty(data.shape[0], dtype=float)\n",
    "    for strain in tqdm(strains, desc='Building TACTIC models', leave=False): \n",
    "        # Test set (strain of interest)\n",
    "        test_ix         = [s == strain for s in data['Strain'].tolist()]\n",
    "        test_ixns       = list(compress(data['Interaction'], test_ix))\n",
    "        test_scores     = list(compress(data['Score'], test_ix))\n",
    "        Xte, n          = X[test_ix], test_ix.count(True)\n",
    "        print('\\tStrain: {}, N = {} ({}%)'.format(strain, n, round(100*n/N, 1)))\n",
    "        # Train set for TACTIC model\n",
    "        train_ix        = [not i for i in test_ix]\n",
    "        if strain==data1['Strain'].unique()[0]: \n",
    "            extra_X     = master['X'][idx2]\n",
    "            extra_s     = master['data']['Score'][idx2].tolist()\n",
    "        elif strain==data2['Strain'].unique()[0]: \n",
    "            extra_X     = master['X'][idx1]\n",
    "            extra_s     = master['data']['Score'][idx1].tolist()\n",
    "        else: \n",
    "            extra_X     = master['X'][idx1 | idx2]\n",
    "            extra_s     = master['data']['Score'][idx1 | idx2].tolist()\n",
    "        Xtr             = sparse.vstack((extra_X, X[train_ix]))\n",
    "        train_scores    = extra_s + list(compress(data['Score'], train_ix))\n",
    "        # Instantiate results\n",
    "        df = pd.DataFrame({'Ixn': test_ixns, 'Score': test_scores, \n",
    "                            'EC': list(compress(MLcomp[label]['EC'], test_ix)), \n",
    "                            'TB': list(compress(MLcomp[label]['TB'], test_ix))})\n",
    "        # Run TACTIC\n",
    "        model            = alg.fit(Xtr, train_scores)\n",
    "        df['TACTIC']     = model.predict(Xte)\n",
    "        # Store results\n",
    "        MLcomp[label][strain] = df\n",
    "        MLcomp[label]['TACTIC'][test_ix] = df['TACTIC']\n",
    "\n",
    "# Save output\n",
    "out['bench']['ML'] = MLcomp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Checkpoint\n",
    "with open('./TACTIC_data.pkl', 'wb') as f: \n",
    "    pickle.dump([ecoli, mtb, master, out], f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Model Interpretation\n",
    "\n",
    "Inspection of feature importance and evaluation of pathways enriched by top features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define full TACTIC model (runtime: ~20 minutes)\n",
    "model = RandomForestRegressor()\n",
    "model.fit(master['X'], master['data']['Score'].tolist())\n",
    "out['model']['TACTIC'] = model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.1 Extract top important features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import ecoli-to-mtb gene map\n",
    "gmap = pd.read_csv('./data/OrtholugeDB/Ec-MG1655_Mtb-H37Rv.txt', sep='\\t')\n",
    "gmap = gmap[['Locus Tag (Strain 1)', 'Locus Tag (Strain 2)']].rename(\n",
    "\tcolumns={'Locus Tag (Strain 1)': 'ecoli', 'Locus Tag (Strain 2)': 'mtb'}\n",
    ")\n",
    "gmap['mtb'] = [g.replace('RVBD_', 'Rv') for g in gmap['mtb'].tolist()]\n",
    "\n",
    "# Feature importance\n",
    "mdl = out['model']['TACTIC']\n",
    "fns = master['df'].index\n",
    "gns = [s.split('-')[-1] for s in fns]\n",
    "gns = [s.replace('RVBD_', 'Rv') for s in gns]\n",
    "gns = [s.split('_')[0] for s in gns]\n",
    "imp = mdl.feature_importances_\n",
    "std = np.std([tree.feature_importances_ for tree in mdl.estimators_], axis=0)\n",
    "df  = pd.DataFrame({'Gene': gns, 'Score': imp, 'SD': std}, index=fns)\n",
    "\n",
    "# Calculate cumulative sum of importance scores\n",
    "df  = df.sort_values(by=['Score'], ascending=False)\n",
    "df['CS'] = np.cumsum(df['Score']) / np.sum(df['Score'])\n",
    "\n",
    "# Rename gene column\n",
    "idx = [g in df['Gene'].tolist() for g in gmap['mtb'].tolist()]\n",
    "oegenes = list(compress(gmap['ecoli'].tolist(), idx))\n",
    "omgenes = list(compress(gmap['mtb'].tolist(), idx))\n",
    "labels = ['{} / {}'.format(eg, mg) for eg, mg in zip(oegenes, omgenes)]\n",
    "df['Gene'] = df['Gene'].replace(dict(zip(omgenes, labels)))\n",
    "\n",
    "# Save main output\n",
    "out['importance'] = df\n",
    "\n",
    "# Extract features that explain 95% \n",
    "k = np.argmax(df['CS'] > 0.95) + 1\n",
    "genes = sorted(set(df['Gene'][:k]), key=lambda v: v.upper())\n",
    "\n",
    "# Separate overlapping vs. unique genes\n",
    "ogenes = [g for g in genes if '/' in g]\n",
    "ugenes = [g for g in genes if g not in ogenes]\n",
    "\n",
    "# Curate gene list for E. coli\n",
    "egenes = sorted(set([g for g in ugenes if g.startswith('b')] + [g.split(' / ')[0] for g in ogenes]))\n",
    "df = pd.read_csv('./data/OrtholugeDB/Ec-MG1655_all.txt', sep='\\t')\n",
    "df = df.iloc[:, 1:4]\n",
    "df.columns = ['Locus', 'RefSeq', 'Description']\n",
    "edf = df[df['Locus'].isin(egenes)]\n",
    "\n",
    "# Curate gene list for M. tb\n",
    "mgenes = sorted(set([g for g in ugenes if g.startswith('Rv')] + [g.split(' / ')[1] for g in ogenes]))\n",
    "df = pd.read_csv('./data/OrtholugeDB/Mtb-H37Rv_all.txt', sep='\\t')\n",
    "df = df.iloc[:, 1:4]\n",
    "df.columns = ['Locus', 'RefSeq', 'Description']\n",
    "df['Locus'] = [s.replace('RVBD_', 'Rv') for s in df['Locus'].tolist()]\n",
    "mdf = df[df['Locus'].isin(mgenes)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2 GSEA based on KEGG database"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Instantialize output variable\n",
    "out['gsea'] = dict()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.2.1 E. coli"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import gene info\n",
    "df = pd.read_csv('https://rest.kegg.jp/list/eco', sep='\\t', header=None)\n",
    "df.index = [s.split(':')[1] for s in df.iloc[:, 0].tolist()]\n",
    "df['Gene'] = [s.split(';')[0] for s in df.iloc[:, -1]]\n",
    "df['Description'] = [s.split('; ')[1] for s in df.iloc[:, -2]]\n",
    "eco_genes = df[['Gene', 'Description']]\n",
    "\n",
    "# Import pathway info\n",
    "df = pd.read_csv('https://rest.kegg.jp/list/pathway/eco', sep='\\t', header=None)\n",
    "df = df.set_index(0).sort_index().rename(columns={1: 'Pathway'})\n",
    "df['Pathway'] = [p.split(' - ')[0] for p in df['Pathway'].tolist()]\n",
    "eco_path = df[['Pathway']]\n",
    "\n",
    "# Import pathways + associated genes\n",
    "df = pd.read_csv('https://rest.kegg.jp/link/eco/pathway', sep='\\t', header=None)\n",
    "df['pathway'] = [s.split(':')[1] for s in df.iloc[:, 0]]\n",
    "df['gene'] = [s.split(':')[1] for s in df.iloc[:, 1]]\n",
    "pathways = sorted(set(df['pathway'].values))\n",
    "genes = sorted(set(df['gene'].values))\n",
    "eco = dict()\n",
    "for path in pathways: \n",
    "    eco[path] = df.loc[df['pathway']==path, 'gene'].tolist()\n",
    "eco_path['N'] = [len(eco[path]) for path in eco_path.index]\n",
    "\n",
    "# Add pathway to gene info\n",
    "gene_dict = dict()\n",
    "for g in genes: \n",
    "    pids = df['pathway'].values[df['gene']==g]\n",
    "    pathway = [eco_path.loc[p, 'Pathway'] for p in pids]\n",
    "    if len(pathway) > 1: \n",
    "        gene_dict[g] = ', '.join(pathway)\n",
    "    else: \n",
    "        gene_dict[g] = pathway[0]\n",
    "eco_genes['Pathway'] = eco_genes.index.map(gene_dict)\n",
    "\n",
    "# Load gene list\n",
    "with open('data/GSEA/ecoli_genes.txt') as file:\n",
    "    egenes = [line.rstrip() for line in file]\n",
    "eco_path['R'] = np.nan\n",
    "for idx in eco_path.index: \n",
    "    eco_path.loc[idx, 'R'] = [g in eco[idx] for g in egenes].count(True)\n",
    "\n",
    "# Hypergeometric test\n",
    "M = len(genes)\n",
    "N = [g in genes for g in egenes].count(True)\n",
    "eco_path = eco_path.assign(p=lambda x: (hypergeom(M=M, n=x['N'], N=N).sf(x['R']-1)))\n",
    "eco_path['padj'] = multipletests(eco_path['p'], method='fdr_bh')[1]\n",
    "out['gsea']['ecoli'] = eco_path"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.2.2 M. tb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import gene info\n",
    "df = pd.read_csv('https://rest.kegg.jp/list/mtu', sep='\\t', header=None)\n",
    "df.index = [s.split(':')[1] for s in df.iloc[:, 0].tolist()]\n",
    "df['Gene'] = [s.split(';')[0] for s in df.iloc[:, -1]]\n",
    "def dscr(s): \n",
    "    try: \n",
    "        return s.split('; ')[1]\n",
    "    except: \n",
    "        return s\n",
    "df['Description'] = [dscr(s) for s in df.iloc[:, -2]]\n",
    "mtu_genes = df[['Gene', 'Description']]\n",
    "\n",
    "# Import pathway info\n",
    "df = pd.read_csv('https://rest.kegg.jp/list/pathway/mtu', sep='\\t', header=None)\n",
    "df = df.set_index(0).sort_index().rename(columns={1: 'Pathway'})\n",
    "df['Pathway'] = [p.split(' - ')[0] for p in df['Pathway'].tolist()]\n",
    "mtu_path = df[['Pathway']]\n",
    "\n",
    "# Import pathways + associated genes\n",
    "df = pd.read_csv('https://rest.kegg.jp/link/mtu/pathway', sep='\\t', header=None)\n",
    "df['pathway'] = [s.split(':')[1] for s in df.iloc[:, 0]]\n",
    "df['gene'] = [s.split(':')[1] for s in df.iloc[:, 1]]\n",
    "pathways = sorted(set(df['pathway'].values))\n",
    "genes = sorted(set(df['gene'].values))\n",
    "mtu = dict()\n",
    "for path in pathways: \n",
    "    mtu[path] = df.loc[df['pathway']==path, 'gene'].tolist()\n",
    "mtu_path['N'] = [len(mtu[path]) for path in mtu_path.index]\n",
    "\n",
    "# Add pathway to gene info\n",
    "gene_dict = dict()\n",
    "for g in genes: \n",
    "    pids = df['pathway'].values[df['gene']==g]\n",
    "    pathway = [mtu_path.loc[p, 'Pathway'] for p in pids]\n",
    "    if len(pathway) > 1: \n",
    "        gene_dict[g] = ', '.join(pathway)\n",
    "    else: \n",
    "        gene_dict[g] = pathway[0]\n",
    "mtu_genes['Pathway'] = mtu_genes.index.map(gene_dict)\n",
    "\n",
    "# Load gene list\n",
    "with open('data/GSEA/mtb_genes.txt') as file:\n",
    "    mgenes = [line.rstrip() for line in file]\n",
    "mtu_path['R'] = np.nan\n",
    "for idx in mtu_path.index: \n",
    "    mtu_path.loc[idx, 'R'] = [g in mtu[idx] for g in mgenes].count(True)\n",
    "\n",
    "# Hypergeometric test\n",
    "M = len(genes)\n",
    "N = [g in genes for g in mgenes].count(True)\n",
    "mtu_path = mtu_path.assign(p=lambda x: (hypergeom(M=M, n=x['N'], N=N).sf(x['R']-1)))\n",
    "mtu_path['padj'] = multipletests(mtu_path['p'], method='fdr_bh')[1]\n",
    "out['gsea']['mtb'] = mtu_path"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.2.3 Evaluate top predictors in enriched pathways"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define genes\n",
    "edf = out['gsea']['ecoli'][out['gsea']['ecoli']['padj'] < 0.05]\n",
    "mdf = out['gsea']['mtb'][out['gsea']['mtb']['padj'] < 0.05]\n",
    "genes = []\n",
    "for i in edf.index: \n",
    "    genes = genes + eco[i]\n",
    "for i in mdf.index: \n",
    "    genes = genes + mtu[i]\n",
    "\n",
    "# Isolate input data\n",
    "df = out['importance'][out['importance']['CS'] <= 0.95]\n",
    "x = [f for f in df.index if any(g in f for g in genes)]\n",
    "fdf = master['df'].loc[x]\n",
    "\n",
    "# Conduct statistical test\n",
    "t, p = [], []\n",
    "sjx = master['data']['Label'] == 'Synergy'\n",
    "ajx = master['data']['Label'] == 'Antagonism'\n",
    "for i in fdf.index: \n",
    "    if i.startswith('sigma'): \n",
    "        x = fdf.loc[i].values[sjx]\n",
    "        y = fdf.loc[i].values[ajx]\n",
    "        tstat, pval = ttest_ind(a=x, b=y, equal_var=True)\n",
    "        \"\"\"\n",
    "        if np.var(x)/np.var(y) > 4 or np.var(y)/np.var(x) > 4: \n",
    "            tstat, pval = ttest_ind(a=x, b=y, equal_var=False)\n",
    "        else: \n",
    "            tstat, pval = ttest_ind(a=x, b=y, equal_var=True)\n",
    "        \"\"\"\n",
    "    elif i.startswith('delta'): \n",
    "        cdf = pd.crosstab(master['data']['Label'][sjx | ajx], fdf.loc[i].values[sjx | ajx])\n",
    "        tstat, pval, _, _ = chi2_contingency(cdf, correction=True)\n",
    "    # Save results\n",
    "    t.append(tstat)\n",
    "    p.append(pval)\n",
    "tdf = pd.DataFrame({'t': t, 'p': p}, index=fdf.index)\n",
    "tdf['padj'] = multipletests(tdf['p'], method='fdr_bh')[1]\n",
    "\n",
    "# Save results\n",
    "out['gsea']['features'] = tdf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Checkpoint\n",
    "with open('./TACTIC_data.pkl', 'wb') as f: \n",
    "    pickle.dump([ecoli, mtb, master, out], f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Prediction Landscape\n",
    "\n",
    "Prediction of all possible 2-way interactions between 86 drugs across 18 bacterial strains and 9 media conditions."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.1 Define input"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Instantiate output variable\n",
    "out['landscape'] = dict()\n",
    "\n",
    "# Determine all possible 2-way interactions\n",
    "drugs           = [d for d in out['drugs'] if d not in ['GLUCOSE', 'GLYCEROL']]\n",
    "interactions    = []\n",
    "for k in range(2, 3):\n",
    "    interactions.extend(combinations(drugs, k))\n",
    "interactions    = [list(i) for i in interactions]\n",
    "\n",
    "# Add media-specific interactions\n",
    "media           = ['GLUCOSE', 'GLYCEROL', 'ACETATE', 'GLUCOSAMINE', 'MALTOSE', 'SUCCINATE', 'N-ACETYLGLUCOSAMINE', 'NH4CL']\n",
    "ixn_media       = []\n",
    "for m in media: \n",
    "    ixn_media   = ixn_media + [list(i) + [m] for i in interactions]\n",
    "\n",
    "# Define TACTIC strains\n",
    "tactic_strains  = sorted(out['strains'])\n",
    "\n",
    "# Define NTM strains\n",
    "ntm_strains     = ['M. abscessus', 'M. smegmatis']\n",
    "\n",
    "# Define gut microbiome strains\n",
    "orth            = {'ecoli': pd.read_csv('./data/OrtholugeDB/Ec-MG1655_gut.txt', sep='\\t'), \n",
    "                    'mtb': pd.read_csv('./data/OrtholugeDB/Mtb-H37Rv_gut.txt', sep='\\t')}\n",
    "jx              = [1, 3] + [x for x in range(4, orth['ecoli'].shape[1], 3)]\n",
    "df1, df2        = orth['ecoli'].iloc[:, jx], orth['mtb'].iloc[:, jx]\n",
    "df1.columns     = [s.split(' - ')[0] if 'comparison' in s else s for s in df1.columns]\n",
    "df2.columns     = [s.split(' - ')[0] if 'comparison' in s else s for s in df2.columns]\n",
    "gut_strains     = sorted(df1.columns[2:])\n",
    "emap, mmap      = dict(), dict()\n",
    "for strain in gut_strains: \n",
    "    emap[strain] = list(compress(df1.iloc[:, 0], df1[strain] != '0'))\n",
    "    mmap[strain] = list(compress(df2.iloc[:, 0], df2[strain] != '0'))\n",
    "\n",
    "# Define strains list + ML inputs\n",
    "strains         = tactic_strains + ntm_strains + gut_strains\n",
    "all_ixns        = interactions + ixn_media\n",
    "n1, n2, n       = len(all_ixns), len(strains), 50000\n",
    "ixn_list        = n2 * all_ixns\n",
    "str_list        = sorted(n1 * strains)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.2 Determine predictions (runtime: ~4 hours)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Request size: 592110 predictions. Partitioning into 12 batches...\n",
      "\tProcessing batch 1...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 24876 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [06:16<00:00, 132.83it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:24<00:00, 12.37s/it]\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 3655 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [08:43<00:00, 95.50it/s] \n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:24<00:00, 12.37s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tProcessing batch 2...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 24915 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [06:36<00:00, 126.22it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 3/3 [00:36<00:00, 12.22s/it]\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 7310 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [08:32<00:00, 97.56it/s] \n",
      "Mapping orthologous genes: 100%|██████████| 3/3 [00:28<00:00,  9.62s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tProcessing batch 3...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 25076 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [06:56<00:00, 119.95it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:37<00:00, 18.86s/it]\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 3655 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [09:02<00:00, 92.24it/s] \n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:23<00:00, 11.77s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tProcessing batch 4...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 24940 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [06:58<00:00, 119.41it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 3/3 [00:39<00:00, 13.24s/it]\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 7310 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [08:43<00:00, 95.44it/s] \n",
      "Mapping orthologous genes: 100%|██████████| 3/3 [00:33<00:00, 11.10s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tProcessing batch 5...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 24859 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [06:55<00:00, 120.32it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:22<00:00, 11.40s/it]\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 3655 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [09:10<00:00, 90.86it/s] \n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:24<00:00, 12.06s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tProcessing batch 6...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 25070 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [06:53<00:00, 120.89it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 3/3 [00:30<00:00, 10.04s/it]\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 7310 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [08:56<00:00, 93.19it/s] \n",
      "Mapping orthologous genes: 100%|██████████| 3/3 [00:27<00:00,  9.29s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tProcessing batch 7...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 24990 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [07:04<00:00, 117.79it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:27<00:00, 13.84s/it]\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 3655 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [09:15<00:00, 90.05it/s] \n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:30<00:00, 15.21s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tProcessing batch 8...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 24857 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [07:06<00:00, 117.23it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 3/3 [00:27<00:00,  9.24s/it]\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 7310 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [08:49<00:00, 94.50it/s] \n",
      "Mapping orthologous genes: 100%|██████████| 3/3 [00:41<00:00, 13.89s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tProcessing batch 9...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 25021 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [07:04<00:00, 117.70it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:24<00:00, 12.47s/it]\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 3655 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [09:13<00:00, 90.36it/s] \n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:28<00:00, 14.33s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tProcessing batch 10...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 25007 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [07:11<00:00, 115.97it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 3/3 [00:34<00:00, 11.40s/it]\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 7310 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [08:53<00:00, 93.63it/s] \n",
      "Mapping orthologous genes: 100%|██████████| 3/3 [00:25<00:00,  8.59s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tProcessing batch 11...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 24888 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [08:04<00:00, 103.29it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:27<00:00, 13.96s/it]\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 3655 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 50000/50000 [09:16<00:00, 89.84it/s] \n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:24<00:00, 12.11s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tProcessing batch 12...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 21151 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 42110/42110 [05:09<00:00, 136.15it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:28<00:00, 14.15s/it]\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 7310 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 42110/42110 [06:43<00:00, 104.31it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 2/2 [00:21<00:00, 10.65s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "All predictions determined.\n"
     ]
    }
   ],
   "source": [
    "# Define phenotype data + feature names + orthology maps\n",
    "p1, p2  = ecoli['omics'].iloc[:, 1:].to_dict('list'), mtb['omics'].iloc[:, 1:].to_dict('list')\n",
    "g1, g2  = ecoli['omics']['Gene'].tolist(), mtb['omics']['Gene'].tolist()\n",
    "ecoli['map'].update(emap)\n",
    "mtb['map'].update(mmap)\n",
    "\n",
    "# Generate predictions (separate into batches)\n",
    "print('Request size: {} predictions. Partitioning into {} batches...'.format(n1 * n2, math.ceil(n1 * n2 / n)))\n",
    "start, stop, i, pred = 0, n, 1, []\n",
    "while start < n1 * n2: \n",
    "    # Print batch number\n",
    "    print('\\tProcessing batch {}...'.format(i))\n",
    "    # Define partitions\n",
    "    if stop > n1 * n2: \n",
    "        i_list, s_list = ixn_list[start:], str_list[start:]\n",
    "    else: \n",
    "        i_list, s_list = ixn_list[start:stop], str_list[start:stop]\n",
    "    # Determine features\n",
    "    f1      = featurize(i_list, p1, feature_names=g1, key=ecoli['key'], strains=s_list, orthology_map=ecoli['map'])\n",
    "    f2      = featurize(i_list, p2, feature_names=g2, key=mtb['key'], strains=s_list, orthology_map=mtb['map'])\n",
    "    d1, d2  = f1['feature_df'].to_numpy().transpose(), f2['feature_df'].to_numpy().transpose()\n",
    "    # Re-format input data\n",
    "    x1      = np.zeros([d2.shape[0], d1.shape[1]], dtype=d2.dtype)\n",
    "    x1[np.array(f1['idx'])] = d1[np.array(f1['idx'])]\n",
    "    idx1    = [f in master['df'].index for f in list(f1['feature_df'].index)]\n",
    "    idx2    = [f in master['df'].index for f in list(f2['feature_df'].index)]\n",
    "    s1, s2  = sparse.csr_matrix(x1[:, np.array(idx1)]), sparse.csr_matrix(d2[:, np.array(idx2)])\n",
    "    X       = sparse.hstack((s1, s2))\n",
    "    # Determine predictions\n",
    "    pred.append(out['model']['TACTIC'].predict(X))\n",
    "    # Re-define batch indices\n",
    "    start   = stop\n",
    "    stop    = start + n\n",
    "    i       += 1\n",
    "print('All predictions determined.')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.3 Define output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Convert prediction data into dataframe\n",
    "y       = np.concatenate(pred, axis=0)\n",
    "ixns    = [', '.join(ixn) for ixn in ixn_list]\n",
    "df      = pd.DataFrame({'Strain': str_list, 'Interaction': ixns, 'Prediction': y})\n",
    "df      = df.pivot_table(index='Interaction', columns='Strain', sort=False)['Prediction']\n",
    "\n",
    "# Extract LB predictions\n",
    "df_LB   = df.iloc[:len(interactions), :]\n",
    "df_LB.index = [i + ', LB' for i in list(df_LB.index)]\n",
    "out['landscape']['lb'] = df_LB\n",
    "\n",
    "# Extract media-specific predictions\n",
    "for m in media: \n",
    "    if m=='GLUCOSAMINE': \n",
    "        idx = [s.endswith(', GLUCOSAMINE') for s in list(df.index)]\n",
    "    else: \n",
    "        idx = [s.endswith(m) for s in list(df.index)]\n",
    "    out['landscape'][m.lower()] = df.loc[idx, ]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.4 Predict AMP-CLA-MEC (runtime: ~2 minutes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:337: UserWarning: Interaction entries do not match profile names, but key provided\n",
      "  warn('Interaction entries do not match profile names, but key provided')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "Defining INDIGO features: 100%|██████████| 18/18 [00:00<00:00, 178.54it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 18/18 [01:01<00:00,  3.39s/it]\n",
      "Defining INDIGO features: 100%|██████████| 18/18 [00:00<00:00, 224.69it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 18/18 [00:41<00:00,  2.33s/it]\n"
     ]
    }
   ],
   "source": [
    "# Define input data\n",
    "ixns    = len(strains)*[['AMP', 'CLA', 'MEC']]\n",
    "f1      = featurize(ixns, p1, feature_names=g1, key=ecoli['key'], strains=strains, orthology_map=ecoli['map'])\n",
    "f2      = featurize(ixns, p2, feature_names=g2, key=mtb['key'], strains=strains, orthology_map=mtb['map'])\n",
    "d1, d2  = f1['feature_df'].to_numpy().transpose(), f2['feature_df'].to_numpy().transpose()\n",
    "\n",
    "# Re-format input data\n",
    "x1      = np.zeros([d2.shape[0], d1.shape[1]], dtype=d2.dtype)\n",
    "x1[np.array(f1['idx'])] = d1[np.array(f1['idx'])]\n",
    "idx1    = [f in master['df'].index for f in list(f1['feature_df'].index)]\n",
    "idx2    = [f in master['df'].index for f in list(f2['feature_df'].index)]\n",
    "s1, s2  = sparse.csr_matrix(x1[:, np.array(idx1)]), sparse.csr_matrix(d2[:, np.array(idx2)])\n",
    "X       = sparse.hstack((s1, s2))\n",
    "\n",
    "# Determine predictions\n",
    "y       = out['model']['TACTIC'].predict(X)\n",
    "\n",
    "# Define output\n",
    "ixns    = [', '.join(ixn) for ixn in ixns]\n",
    "df      = pd.DataFrame({'Strain': strains, 'Interaction': ixns, 'Prediction': y})\n",
    "df      = df.pivot_table(index='Interaction', columns='Strain', sort=False)['Prediction']\n",
    "out['AMP-CLA-MEC'] = df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Checkpoint\n",
    "with open('./TACTIC_data.pkl', 'wb') as f: \n",
    "    pickle.dump([ecoli, mtb, master, out], f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Endophthalmitis application\n",
    "\n",
    "TACTIC application to predict narrow-spectrum synergies against causative agents of endophthalmitis. In collaboration with Dr. Karthik Srinivasan (U-M Kelogg Eye Center)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.1 Load and process data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load datasets\n",
    "data = pd.read_excel('./data/Endophthalmitis/endophthalmitis_data.xlsx', sheet_name=None, engine='openpyxl')\n",
    "orth = {'ecoli': pd.read_csv('./data/Endophthalmitis/Eco_orthologs.txt', sep='\\t'), \n",
    "        'mtb': pd.read_csv('./data/Endophthalmitis/Mtb_orthologs.txt', sep='\\t')}\n",
    "\n",
    "# Define strain-specific orthology data\n",
    "df          = data['strains'].dropna()\n",
    "strains     = sorted(list(df['Organism'] + ' ' + df['Strain']))\n",
    "jx          = [1, 3] + [x for x in range(4, orth['ecoli'].shape[1], 3)]\n",
    "df1, df2    = orth['ecoli'].iloc[:, jx], orth['mtb'].iloc[:, jx]\n",
    "df1.columns = [s.split(' - ')[0] if 'comparison' in s else s for s in df1.columns]\n",
    "df2.columns = [s.split(' - ')[0] if 'comparison' in s else s for s in df2.columns]\n",
    "df2.iloc[:, 0] = [s.replace('RVBD_', 'Rv') for s in list(df2.iloc[:, 0])]\n",
    "emap, mmap  = dict(), dict()\n",
    "for strain in strains: \n",
    "    emap[strain] = list(compress(df1.iloc[:, 0], df1[strain] != '0'))\n",
    "    mmap[strain] = list(compress(df2.iloc[:, 0], df2[strain] != '0'))\n",
    "\n",
    "# Define drugs and key\n",
    "ekey = list(zip(data['key']['Code'], data['key']['ecoli_label']))\n",
    "ekey = [key for key in ekey if not pd.isna(key[1])]\n",
    "mkey = list(zip(data['key']['Code'], data['key']['mtb_label']))\n",
    "drugs = sorted([k[0] for k in mkey])\n",
    "abx   = data['drugs'].loc[data['drugs']['Route_1']==1, 'Code'].tolist()\n",
    "\n",
    "# Define model\n",
    "model = out['model']['TACTIC']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2 Determine predictions (runtime: ~20 minutes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Request size: 31536 predictions. Partitioning into 1 batches...\n",
      "\tProcessing batch 1...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:337: UserWarning: Interaction entries do not match profile names, but key provided\n",
      "  warn('Interaction entries do not match profile names, but key provided')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 6576 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 31536/31536 [06:12<00:00, 84.63it/s] \n",
      "Mapping orthologous genes: 100%|██████████| 12/12 [02:06<00:00, 10.51s/it]\n",
      "Defining INDIGO features: 100%|██████████| 31536/31536 [09:54<00:00, 53.03it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 12/12 [00:43<00:00,  3.61s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "All predictions determined.\n"
     ]
    }
   ],
   "source": [
    "# Determine all possible 2-way interactions\n",
    "interactions    = []\n",
    "for k in range(2, 3):\n",
    "    interactions.extend(combinations(drugs, k))\n",
    "interactions    = [list(i) for i in interactions]\n",
    "n1, n2, n       = len(interactions), len(strains), 50000\n",
    "ixn_list        = interactions * n2\n",
    "str_list        = sorted(strains * n1)\n",
    "\n",
    "# Determine phenotype data\n",
    "p1, p2  = ecoli['omics'].iloc[:, 1:].to_dict('list'), mtb['omics'].iloc[:, 1:].to_dict('list')\n",
    "g1, g2  = ecoli['omics']['Gene'].tolist(), mtb['omics']['Gene'].tolist()\n",
    "\n",
    "# Generate predictions\n",
    "print('Request size: {} predictions. Partitioning into {} batches...'.format(n1 * n2, math.ceil(n1 * n2 / n)))\n",
    "start, stop, i, pred = 0, n, 1, []\n",
    "while start < n1 * n2: \n",
    "    # Print batch number\n",
    "    print('\\tProcessing batch {}...'.format(i))\n",
    "    # Define partitioned input\n",
    "    if stop > n1 * n2: \n",
    "        ixns = ixn_list[start:]\n",
    "        strs = str_list[start:]\n",
    "    else: \n",
    "        ixns = ixn_list[start:stop]\n",
    "        strs = str_list[start:stop]\n",
    "    # Determine features\n",
    "    f1      = featurize(ixns, p1, feature_names=g1, key=ekey, strains=strs, orthology_map=emap)\n",
    "    f2      = featurize(ixns, p2, feature_names=g2, key=mkey, strains=strs, orthology_map=mmap)\n",
    "    d1, d2  = f1['feature_df'].to_numpy().transpose(), f2['feature_df'].to_numpy().transpose()\n",
    "    # Define TACTIC input\n",
    "    idx1    = [f in master['df'].index for f in list(f1['feature_df'].index)]\n",
    "    idx2    = [f in master['df'].index for f in list(f2['feature_df'].index)]\n",
    "    s1, s2  = sparse.csr_matrix(d1[:, np.array(idx1)]), sparse.csr_matrix(d2[:, np.array(idx2)])\n",
    "    X       = sparse.hstack((s1, s2))\n",
    "    # Determine predictions\n",
    "    pred.append(model.predict(X))\n",
    "    # Re-define batch indices\n",
    "    start   = stop\n",
    "    stop    = start + n\n",
    "    i       += 1\n",
    "print('All predictions determined.')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.3 Define output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Re-format + save data\n",
    "y       = np.concatenate(pred, axis=0)\n",
    "ixns    = [', '.join(ixn) for ixn in ixn_list]\n",
    "df      = pd.DataFrame({'Strain': str_list, 'Interaction': ixns, 'Prediction': y})\n",
    "out['endophthalmitis'] = df.pivot(index='Interaction', columns='Strain')['Prediction']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Checkpoint\n",
    "with open('./TACTIC_data.pkl', 'wb') as f: \n",
    "    pickle.dump([ecoli, mtb, master, out], f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6. Gram-positive benchmarking\n",
    "\n",
    "Additional analysis using data from [Cacace et al., 2023](https://pubmed.ncbi.nlm.nih.gov/37770760/)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "### 6.1 Load and process data (runtime: ~3 minutes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 510 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 4570/4570 [00:18<00:00, 246.91it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 15/15 [01:13<00:00,  4.93s/it]\n",
      "Defining INDIGO features: 100%|██████████| 4570/4570 [00:20<00:00, 218.94it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 15/15 [00:34<00:00,  2.30s/it]\n"
     ]
    }
   ],
   "source": [
    "# Load Gram-positive data\n",
    "pos_data = pd.read_excel('./data/ixn_data.xlsx', sheet_name='Cacace_2023', engine='openpyxl')\n",
    "pos_orthology = {\n",
    "    'ecoli': pd.read_csv('./data/OrtholugeDB/Ec-MG1655_Gpos.txt', sep='\\t'), \n",
    "    'mtb': pd.read_csv('./data/OrtholugeDB/Mtb-H37Rv_Gpos.txt', sep='\\t')\n",
    "}\n",
    "df_master = pd.concat([master['data'].copy(), pos_data.copy()])\n",
    "\n",
    "# Obtain orthology maps\n",
    "strains         = sorted(set(pos_data['Strain']))\n",
    "jx              = [1, 3] + [x for x in range(4, pos_orthology['ecoli'].shape[1], 3)]\n",
    "df1, df2        = pos_orthology['ecoli'].iloc[:, jx], pos_orthology['mtb'].iloc[:, jx]\n",
    "df1.columns     = [s.split(' - ')[0] if 'comparison' in s else s for s in df1.columns]\n",
    "df2.columns     = [s.split(' - ')[0] if 'comparison' in s else s for s in df2.columns]\n",
    "df2.iloc[:, 0]  = [s.replace('RVBD_', 'Rv') for s in df2.iloc[:, 0]]\n",
    "for strain in strains: \n",
    "    # G+ strains\n",
    "    if 'subtilis' in strain: \n",
    "        s1 = [s for s in df1.columns if 'subtilis' in s][0]\n",
    "        s2 = [s for s in df2.columns if 'subtilis' in s][0]\n",
    "    elif 'pneumoniae' in strain: \n",
    "        s1 = [s for s in df1.columns if 'pneumoniae' in s][0]\n",
    "        s2 = [s for s in df2.columns if 'pneumoniae' in s][0]\n",
    "    elif 'Newman' in strain: \n",
    "        s1 = [s for s in df1.columns if 'Newman' in s][0]\n",
    "        s2 = [s for s in df2.columns if 'Newman' in s][0]\n",
    "    elif strain.endswith('29213'): \n",
    "        s1 = [s for s in df1.columns if 'ECT-R 2' in s][0]\n",
    "        s2 = [s for s in df2.columns if 'ECT-R 2' in s][0]\n",
    "    ecoli['map'][strain] = list(compress(df1.iloc[:, 0], df1[s1] != '0'))\n",
    "    mtb['map'][strain] = list(compress(df2.iloc[:, 0], df2[s2] != '0'))\n",
    "\n",
    "# Determine ML features (based on E. coli chemogenomic-based data)\n",
    "profiles        = ecoli['omics'].iloc[:, 1:].to_dict('list')\n",
    "feature_names   = ecoli['omics']['Gene'].tolist()\n",
    "interactions    = [list(s.split(', ')) for s in df_master['Interaction']]\n",
    "ecoli_feat      = featurize(interactions, profiles, feature_names=feature_names, key=ecoli['key'], \n",
    "                            strains=df_master['Strain'].tolist(), orthology_map=ecoli['map'])\n",
    "ecoli_X         = sparse.csr_matrix(ecoli_feat['feature_df'].to_numpy().transpose())\n",
    "\n",
    "# Determine ML features (based on M. tb transcriptomic-based data)\n",
    "profiles        = mtb['omics'].iloc[:, 1:].to_dict('list')\n",
    "feature_names   = mtb['omics']['Gene'].tolist()\n",
    "interactions    = [list(s.split(', ')) for s in df_master['Interaction']]\n",
    "mtb_feat        = featurize(interactions, profiles, feature_names=feature_names, key=mtb['key'], \n",
    "                            strains=df_master['Strain'].tolist(), orthology_map=mtb['map'])\n",
    "mtb_X           = sparse.csr_matrix(mtb_feat['feature_df'].to_numpy().transpose())\n",
    "\n",
    "# Determine ML features for TACTIC\n",
    "ecoli_genes     = [g for g in ecoli['omics']['Gene'].tolist() if g not in ecoli['map']['M. tuberculosis H37Rv']]\n",
    "idx             = [any(i.split('_')[0].endswith(gene) for gene in ecoli_genes) for i in list(ecoli['df'].index)]\n",
    "feature_df      = pd.concat([ecoli_feat['feature_df'].loc[idx], mtb_feat['feature_df']])\n",
    "master_X        = sparse.csr_matrix(feature_df.to_numpy().transpose())\n",
    "\n",
    "# Sanity check\n",
    "assert (ecoli_X.shape[0]==mtb_X.shape[0]) and (ecoli_X.shape[0]==master_X.shape[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.2 Model building and evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO (including Cacace data)::   7%|▋         | 1/15 [00:00<00:01,  8.31it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\tStrain: A. baumannii ATCC 17978, N = 45 (1.1%)\n",
      "\t\tHoldout results (INDIGO):\tR = 0.33, p = 0.028\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = -0.18, p = 0.233\n",
      "\t\tHoldout results (TACTIC):\tR = 0.98, p = 3.7e-30\n",
      "\tStrain: B. subtilis 168, N = 533 (12.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO (including Cacace data)::  20%|██        | 3/15 [00:00<00:01,  7.37it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.09, p = 0.0395\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = -0.01, p = 0.833\n",
      "\t\tHoldout results (TACTIC):\tR = 0.15, p = 0.000502\n",
      "\tStrain: E. coli BW25113, N = 473 (11.3%)\n",
      "\t\tHoldout results (INDIGO):\tR = 0.23, p = 2.35e-07\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = -0.1, p = 0.0307\n",
      "\t\tHoldout results (TACTIC):\tR = 0.83, p = 2.75e-123\n",
      "\tStrain: E. coli MC4100, N = 98 (2.4%)\n",
      "\t\tHoldout results (INDIGO):\tR = 0.3, p = 0.00238\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = -0.1, p = 0.331\n",
      "\t\tHoldout results (TACTIC):\tR = 0.72, p = 5.48e-17\n",
      "\tStrain: E. coli MG1655, N = 452 (10.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO (including Cacace data)::  47%|████▋     | 7/15 [00:00<00:00,  9.86it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.26, p = 3.79e-08\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = -0.13, p = 0.00455\n",
      "\t\tHoldout results (TACTIC):\tR = 0.86, p = 2.41e-132\n",
      "\tStrain: E. coli iAi1, N = 316 (7.6%)\n",
      "\t\tHoldout results (INDIGO):\tR = 0.16, p = 0.00379\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.06, p = 0.293\n",
      "\t\tHoldout results (TACTIC):\tR = 0.95, p = 1.57e-155\n",
      "\tStrain: M. tuberculosis ATCC 35801, N = 154 (3.7%)\n",
      "\t\tHoldout results (INDIGO):\tR = -0.04, p = 0.662\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.32, p = 4.32e-05\n",
      "\t\tHoldout results (TACTIC):\tR = 0.86, p = 5.86e-46\n",
      "\tStrain: M. tuberculosis H37Rv, N = 159 (3.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO (including Cacace data)::  73%|███████▎  | 11/15 [00:00<00:00, 13.62it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.29, p = 0.000176\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.29, p = 0.000176\n",
      "\t\tHoldout results (TACTIC):\tR = 0.83, p = 2.23e-41\n",
      "\tStrain: P. aeruginosa PA14, N = 163 (3.9%)\n",
      "\t\tHoldout results (INDIGO):\tR = 0.02, p = 0.81\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.09, p = 0.244\n",
      "\t\tHoldout results (TACTIC):\tR = 0.94, p = 4.18e-75\n",
      "\tStrain: P. aeruginosa PAO1, N = 163 (3.9%)\n",
      "\t\tHoldout results (INDIGO):\tR = -0.15, p = 0.0541\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = -0.06, p = 0.467\n",
      "\t\tHoldout results (TACTIC):\tR = 0.95, p = 3.14e-81\n",
      "\tStrain: S. aureus ATCC 29213, N = 45 (1.1%)\n",
      "\t\tHoldout results (INDIGO):\tR = 0.26, p = 0.0886\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.22, p = 0.141\n",
      "\t\tHoldout results (TACTIC):\tR = 0.98, p = 7.39e-30\n",
      "\tStrain: S. aureus Newman, N = 534 (12.8%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO (including Cacace data)::  87%|████████▋ | 13/15 [00:01<00:00, 11.25it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.06, p = 0.171\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.0, p = 0.956\n",
      "\t\tHoldout results (TACTIC):\tR = 0.04, p = 0.367\n",
      "\tStrain: S. pneumoniae D39V, N = 538 (12.9%)\n",
      "\t\tHoldout results (INDIGO):\tR = -0.02, p = 0.672\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.1, p = 0.0273\n",
      "\t\tHoldout results (TACTIC):\tR = 0.11, p = 0.0108\n",
      "\tStrain: S. typhimurium 14028s, N = 248 (5.9%)\n",
      "\t\tHoldout results (INDIGO):\tR = 0.08, p = 0.19\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = -0.07, p = 0.304\n",
      "\t\tHoldout results (TACTIC):\tR = 0.94, p = 1.27e-118\n",
      "\tStrain: S. typhimurium LT2, N = 248 (5.9%)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating TACTIC vs. INDIGO (including Cacace data):: 100%|██████████| 15/15 [00:01<00:00, 10.54it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t\tHoldout results (INDIGO):\tR = 0.12, p = 0.0689\n",
      "\t\tHoldout results (INDIGO-Mtb):\tR = 0.03, p = 0.626\n",
      "\t\tHoldout results (TACTIC):\tR = 0.97, p = 1.65e-148\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# Define I/O variables\n",
    "idx     = df_master['Set'] == 'TACTIC'\n",
    "data    = df_master.loc[idx]\n",
    "strains = sorted(list(set(data['Strain'])))\n",
    "X1, X2, X = ecoli_X[idx], mtb_X[idx], master_X[idx]\n",
    "out['Cacace'] = dict()\n",
    "\n",
    "# Define INDIGO data subsets\n",
    "idx1    = df_master['Set'] == 'ecoli'\n",
    "data1   = df_master.loc[idx1]\n",
    "idx2    = df_master['Set'] == 'mtb'\n",
    "data2   = df_master.loc[idx2]\n",
    "\n",
    "# Determine predictions for INDIGO (E. coli) model\n",
    "out['Cacace']['EC'] = out['model']['EC'].predict(X1)\n",
    "\n",
    "# Determine predictions for INDIGO-Mtb (M. tb) model\n",
    "out['Cacace']['TB'] = out['model']['TB'].predict(X2)\n",
    "\n",
    "# Evaluate TACTIC vs. INDIGO\n",
    "N = data.shape[0]\n",
    "for strain in tqdm(strains, desc='Evaluating TACTIC vs. INDIGO (including Cacace data):'): \n",
    "    # Test set (strain of interest)\n",
    "    test_ix         = [s == strain for s in data['Strain'].tolist()]\n",
    "    test_ixns       = list(compress(data['Interaction'], test_ix))\n",
    "    test_scores     = list(compress(data['Score'], test_ix))\n",
    "    Xte, n          = X[test_ix], test_ix.count(True)\n",
    "    print('\\tStrain: {}, N = {} ({}%)'.format(strain, n, round(100*n/N, 1)))\n",
    "    # Train set for TACTIC model\n",
    "    train_ix        = [not i for i in test_ix]\n",
    "    if strain==data1['Strain'].unique()[0]: \n",
    "        extra_X     = master_X[idx2]\n",
    "        extra_s     = df_master['Score'][idx2].tolist()\n",
    "    elif strain==data2['Strain'].unique()[0]: \n",
    "        extra_X     = master_X[idx1]\n",
    "        extra_s     = df_master['Score'][idx1].tolist()\n",
    "    else: \n",
    "        extra_X     = master_X[idx1 | idx2]\n",
    "        extra_s     = df_master['Score'][idx1 | idx2].tolist()\n",
    "    Xtr             = sparse.vstack((extra_X, X[train_ix]))\n",
    "    train_scores    = extra_s + list(compress(data['Score'], train_ix))\n",
    "    # Instantiate results\n",
    "    df = pd.DataFrame({'Ixn': test_ixns, 'Score': test_scores, \n",
    "                    'EC': list(compress(out['Cacace']['EC'], test_ix)), \n",
    "                    'TB': list(compress(out['Cacace']['TB'], test_ix))})\n",
    "    df['TACTIC'] = out['model']['TACTIC'].predict(Xte)\n",
    "    \"\"\"\n",
    "    # Run TACTIC\n",
    "    model           = RandomForestRegressor()\n",
    "    model.fit(Xtr, train_scores)\n",
    "    df['TACTIC'] = model.predict(Xte)\n",
    "    \"\"\"\n",
    "    # Print results\n",
    "    r, p = spearmanr(df['Score'], df['EC'])\n",
    "    print('\\t\\tHoldout results (INDIGO):\\tR = {}, p = {:.3g}'.format(round(r, 2), p))\n",
    "    r, p = spearmanr(df['Score'], df['TB'])\n",
    "    print('\\t\\tHoldout results (INDIGO-Mtb):\\tR = {}, p = {:.3g}'.format(round(r, 2), p))\n",
    "    r, p = spearmanr(df['Score'], df['TACTIC'])\n",
    "    print('\\t\\tHoldout results (TACTIC):\\tR = {}, p = {:.3g}'.format(round(r, 2), p))\n",
    "    # Store results\n",
    "    out['Cacace'][strain] = df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Checkpoint\n",
    "with open('./TACTIC_data.pkl', 'wb') as f: \n",
    "    pickle.dump([ecoli, mtb, master, out], f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7. M. abscessus validation\n",
    "\n",
    "Additional analysis using data from [Van et al., 2023](https://pubmed.ncbi.nlm.nih.gov/37278639/)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7.1 Digitize Figure 2A\n",
    "\n",
    "Infer log2FIC50 values via image processing."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Reduced heatmap (18 x 18)')"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x300 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Define output variable\n",
    "out['Van'] = dict()\n",
    "\n",
    "# Define drug dictionary\n",
    "abx = {'AMK': 'AMK', \n",
    "       'TEL': '', \n",
    "       'AZM': 'AZI', \n",
    "       'CLR': 'CLA', \n",
    "       'LZD': 'LIN', \n",
    "       'MXF': 'MOX', \n",
    "       'LXF': 'LEV', \n",
    "       'SPR': '', \n",
    "       'NFT': 'NIT', \n",
    "       'RIF': 'RIF', \n",
    "       'RFB': '', \n",
    "       'AMX': 'AMX', \n",
    "       'VAN': 'VAN', \n",
    "       'EMB': '', \n",
    "       'CER': 'CER', \n",
    "       'FOX': 'CEF', \n",
    "       'BDQ': '', \n",
    "       'TZ': 'THZ'\n",
    "        }\n",
    "out['Van']['abx'] = abx\n",
    "\n",
    "# Read image (Figure 2A)\n",
    "image = cv2.imread('./data/Van_2023_fig2a_hm.jpg', 1)\n",
    "\n",
    "# Convert to RGB\n",
    "image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n",
    "out['Van']['original'] = image\n",
    "\n",
    "# Get individual cells\n",
    "n = len(abx.keys())\n",
    "hm = np.empty((n, n, 3), dtype=int)\n",
    "rm = np.empty((n, n), dtype=int)\n",
    "gm = np.empty((n, n), dtype=int)\n",
    "bm = np.empty((n, n), dtype=int)\n",
    "zm = np.empty((n, n), dtype=float)\n",
    "for i in range(0, n, 1): \n",
    "      for j in range(0, n, 1): \n",
    "        # Define image index\n",
    "        x, y = 52*i+26, 52*j+26\n",
    "        # Extract RGB value\n",
    "        hm[i, j] = image[x, y]\n",
    "        # Calculate score\n",
    "        if all(hm[i, j]==[169, 169, 169]): \n",
    "            zm[i, j] = np.nan\n",
    "        else: \n",
    "            zm[i, j] = ((hm[i, j, 0] - 255) - (hm[i, j, -1] - 255)) / hm[i, j, 1]\n",
    "out['Van']['reduced'] = hm\n",
    "out['Van']['scores'] = zm\n",
    "\n",
    "# Sanity check #1\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(6, 3), constrained_layout=True)\n",
    "ax[0].imshow(image)\n",
    "ax[0].set_title('Original heatmap ({} x {})'.format(image.shape[0], image.shape[1]))\n",
    "ax[1].imshow(hm)\n",
    "ax[1].set_title('Reduced heatmap ({} x {})'.format(hm.shape[0], hm.shape[1]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7.2 Inspect and process inferred scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Absolute min: -1.9787234042553192\n",
      "Absolute max: 10.555555555555555\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Inferred score heatmap')"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmoAAADTCAYAAAAxiD/rAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8pXeV/AAAACXBIWXMAAA9hAAAPYQGoP6dpAAA6yUlEQVR4nO3deVhU9f4H8DcgzAzbgLILIuKKigvmhuaGomLlletW3RBNrcA1q0s3RbkqpZmUYmgWdkuv+5YLKuRSpmS4JKmIiokQuCDgBijz/f3hj7keGWQbmGF4v57nPA/z/Z7lcw6HD585q5EQQoCIiIiI9I6xrgMgIiIiIs1YqBERERHpKRZqRERERHqKhRoRERGRnmKhRkRERKSnWKgRERER6SkWakRERER6ioUaERERkZ5ioUZERESkp1io1VPjxo1D06ZNtT7fpk2bYty4cVqfb3kOHToEIyMjHDp0qNaXre+GDh2KiRMn1uoy586dCyMjo0pNc+7cOTRo0ADJyck1FBXpE+ag+kMXOUjXnt0P4+LiYGlpiZs3b1Z6XvW2UFuzZg2MjIzw22+/6ToUohpz9OhR7N+/Hx988IGuQymXl5cXAgICMGfOHF2HUiuYg6g+qEs5qCYNHjwYzZs3R2RkZKWnrbeFGtWMlJQUfPXVV7oOg/7f4sWLMWDAADRv3rxWl/vRRx/h4cOHlZ7urbfewrZt23D58uUaiIrqA+Yg/aKrHKSPJk+ejJUrV+Lu3buVmo6FGlWbEEL9T1kmk8HU1FTHEem3+/fv18pybty4gd27d2PUqFG1srynNWjQAHK5vNLT+fn5wdbWFt9++20NREWGijmocupDDtJHgYGBKCwsxKZNmyo1HQu1p4wbNw6Wlpa4du0ahg0bBktLSzRu3BjR0dEAgLNnz6J///6wsLCAu7s71q1bJ5k+JycHs2bNQvv27WFpaQlra2sMGTIEZ86cKbWsP//8Ey+//DIsLCzg4OCAGTNmYN++fRqvcUhMTMTgwYOhVCphbm6OPn364OjRoxVer+3bt6Ndu3aQy+Vo164dtm3bpnE8lUqFqKgotG3bFnK5HI6Ojpg8eTLu3LkjGa9p06YYNmwY9u3bhy5dukChUGDlypXqvpLz8r/99huMjIw0/tMtWdddu3ap2zIyMjB+/Hg4OjpCJpOhbdu2+Oabb0pNe/36dQwfPlyy7QoLCyu0Le7evYvp06ejadOmkMlkcHBwwMCBA3Hy5EnJeImJiRg6dChsbW1hYWEBb29vfP7555JxfvzxR/Tu3RsWFhawsbHBK6+8gvPnz0vGKblW69y5c3j11Vdha2uLXr16qfu///57+Pj4QKFQoGHDhhgzZgzS09Ml80hNTUVgYCCcnJwgl8vh6uqKMWPGIC8v77nrunv3bjx+/Bh+fn6S9srsp8uWLUPbtm1hbm4OW1tbdOnSpdR+r4mma9QOHDiAXr16wcbGBpaWlmjVqhU+/PBDyTimpqbo27cvduzYUe4yDBFzEHNQCUPOQQCQm5uLGTNmqLeDq6sr3njjDdy6dQsAUFRUhDlz5sDHxwdKpRIWFhbo3bs3Dh48KJnP1atXYWRkhE8//RRLly6Fu7s7FAoF+vTpU+p615K/rytXrsDf3x8WFhZwcXFBREQEhBCScSu6LwohMH/+fLi6usLc3Bz9+vXDH3/8oXF7ODg4wNvbu9L5rUGlxq4HiouLMWTIELz44otYtGgR1q5di9DQUFhYWOBf//oXXnvtNYwYMQIxMTF444030KNHD3h4eAAArly5gu3bt2PkyJHw8PBAdnY2Vq5ciT59+uDcuXNwcXEB8OTbTP/+/fHXX39h2rRpcHJywrp160rtgMCTP8QhQ4bAx8cH4eHhMDY2RmxsLPr374+ffvoJXbt2fe767N+/H4GBgfDy8kJkZCRu376N4OBguLq6lhp38uTJWLNmDYKDgzF16lSkpaVh+fLlOHXqFI4ePSr5lpqSkoKxY8di8uTJmDhxIlq1alVqfl26dEGzZs2wceNGBAUFSfo2bNgAW1tb+Pv7AwCys7PRvXt3GBkZITQ0FPb29ti7dy8mTJiA/Px8TJ8+HQDw8OFDDBgwANeuXcPUqVPh4uKC7777Dj/++ONzt0OJt956C5s3b0ZoaCi8vLxw+/Zt/Pzzzzh//jw6d+4M4ElBMWzYMDg7O6t/P+fPn8euXbswbdo0AEB8fDyGDBmCZs2aYe7cuXj48CGWLVsGX19fnDx5stRF0iNHjkSLFi2wcOFCdUJYsGABZs+ejVGjRuHNN9/EzZs3sWzZMrz44os4deoUbGxsUFRUBH9/fxQWFmLKlClwcnJCRkYGdu3ahdzcXCiVyjLX9ZdffkGjRo3g7u4uaa/ofvrVV19h6tSp+Pvf/45p06ahoKAAv//+OxITE/Hqq69WaHuX+OOPPzBs2DB4e3sjIiICMpkMly5d0vjP3sfHBzt27EB+fj6sra0rtRxDwBzEHGToOejevXvo3bs3zp8/j/Hjx6Nz5864desWdu7cievXr8POzg75+flYvXo1xo4di4kTJ+Lu3bv4+uuv4e/vj19//RUdO3aUzPM///kP7t69i5CQEBQUFODzzz9H//79cfbsWTg6OqrHKy4uxuDBg9G9e3csWrQIcXFxCA8Px+PHjxEREaEer6L74pw5czB//nwMHToUQ4cOxcmTJzFo0CAUFRVp3CY+Pj7Yvn172TuIJqKeio2NFQDEiRMn1G1BQUECgFi4cKG67c6dO0KhUAgjIyOxfv16dfuFCxcEABEeHq5uKygoEMXFxZLlpKWlCZlMJiIiItRtS5YsEQDE9u3b1W0PHz4UrVu3FgDEwYMHhRBCqFQq0aJFC+Hv7y9UKpV63AcPHggPDw8xcODActezY8eOwtnZWeTm5qrb9u/fLwAId3d3ddtPP/0kAIi1a9dKpo+LiyvV7u7uLgCIuLi4Ustzd3cXQUFB6s9hYWHC1NRU5OTkqNsKCwuFjY2NGD9+vLptwoQJwtnZWdy6dUsyvzFjxgilUikePHgghBAiKipKABAbN25Uj3P//n3RvHlzybYri1KpFCEhIWX2P378WHh4eAh3d3dx584dSd/Tv4OOHTsKBwcHcfv2bXXbmTNnhLGxsXjjjTfUbeHh4QKAGDt2rGReV69eFSYmJmLBggWS9rNnz4oGDRqo20+dOiUAiE2bNj13vTTp1auX8PHxKdVe0f30lVdeEW3btq30coX433qXWLp0qQAgbt68We6069atEwBEYmJilZZdVzAHMQdpUh9y0Jw5cwQAsXXr1lJ9Jev4+PFjUVhYKOm7c+eOcHR0lPze0tLSBAChUCjE9evX1e2JiYkCgJgxY4a6reTva8qUKZLlBQQECDMzM3V+qui+eOPGDWFmZiYCAgIkv5sPP/xQAJDshyUWLlwoAIjs7OzSG6wMPPWpwZtvvqn+2cbGBq1atYKFhYXkPHurVq1gY2ODK1euqNtkMhmMjZ9s0uLiYty+fVt9iufpw9pxcXFo3LgxXn75ZXWbXC4vdfvy6dOnkZqaildffRW3b9/GrVu3cOvWLdy/fx8DBgzAkSNHoFKpylyPv/76C6dPn0ZQUJDkW8/AgQPh5eUlGXfTpk1QKpUYOHCgejm3bt2Cj48PLC0tS33T9vDwUH8TfZ7Ro0fj0aNH2Lp1q7pt//79yM3NxejRowE8OXS8ZcsWvPTSSxBCSJbv7++PvLw89fbbs2cPnJ2d8fe//109P3Nzc0yaNKncWIAnv8/ExERkZmZq7D916hTS0tIwffp02NjYSPpKTuWVbNdx48ahYcOG6n5vb28MHDgQe/bsKTXft956S/J569atUKlUGDVqlGR9nZyc0KJFC/X2Lvm97du3Dw8ePKjQOpa4ffs2bG1tS7VXdD+1sbHB9evXceLEiUotV5OSbbljx47n7rMA1DGXnAKpj5iDmIMMOQdt2bIFHTp0wN/+9rdSfSXraGJiAjMzMwBPTkPm5OTg8ePH6NKlS6nTxAAwfPhwNG7cWP25a9eu6Natm8ZtERoaKlleaGgoioqKEB8fD6Di+2J8fDyKioowZcoUyaUeJUdfNalKfmOh9gy5XA57e3tJm1KphKura6lrbpRKpeR8tUqlwtKlS9GiRQvIZDLY2dnB3t4ev//+u+Rc/p9//glPT89S83v2rpjU1FQAQFBQEOzt7SXD6tWrUVhYiLy8PBQVFSErK0syFBcX488//wQAtGjRotR6PnuaIDU1FXl5eXBwcCi1rHv37uHGjRuS8UtOtZSnQ4cOaN26NTZs2KBu27BhA+zs7NC/f38AwM2bN5Gbm4tVq1aVWnZwcDAAqJf/559/onnz5qW2nabTHposWrQIycnJcHNzQ9euXTF37lzJP7qSuw3btWtX5jxKtqumZbZp00b9j+xpz26v1NRUCCHQokWLUut8/vx59fp6eHhg5syZWL16Nezs7ODv74/o6Ohyrw0pIZ657gKo+H76wQcfwNLSEl27dkWLFi0QEhIiOVVZ1n6nyejRo+Hr64s333wTjo6OGDNmDDZu3Kjxn3xJzJV9DpuhYA5iDgIMOwddvnz5uetX4ttvv4W3tzfkcjkaNWoEe3t77N69W+OyNe1jLVu2xNWrVyVtxsbGaNasWanxAKjHrei+WNb+bW9vr7FABaqW33iN2jNMTEwq1f70Trhw4ULMnj0b48ePx7///W80bNgQxsbGmD59erlHETQpmWbx4sWlzseXsLS0xNGjR9GvXz9Je1paWqWX5eDggLVr12rsf/Yfh0KhqPC8R48ejQULFuDWrVuwsrLCzp07MXbsWDRo0EC9bAB4/fXXS11HUsLb27vCy3ueUaNGoXfv3ti2bRv279+PxYsX45NPPsHWrVsxZMgQrSxDk2e3l0qlgpGREfbu3atx37K0tFT/vGTJEowbNw47duzA/v37MXXqVERGRuL48eMar/Mp0ahRo1IXvgIV30/btGmDlJQU7Nq1C3FxcdiyZQtWrFiBOXPmYN68efjll1807neaHmKqUChw5MgRHDx4ELt370ZcXBw2bNiA/v37Y//+/ZJtUBKznZ1dmetmyJiDmINqgj7loIr4/vvvMW7cOAwfPhzvvfceHBwcYGJigsjIyBp/fE9l98XKqEp+Y6GmRZs3b0a/fv3w9ddfS9pzc3MlvxR3d3ecO3cOQghJVX3p0iXJdJ6engAAa2trjXfNlOjQoQMOHDggaXNycoJMJgPwv2/FT0tJSSm1rPj4ePj6+lYqAVbE6NGjMW/ePGzZsgWOjo7Iz8/HmDFj1P329vawsrJCcXHxc9cTeLLtkpOTS227Z9fneZydnfHOO+/gnXfewY0bN9C5c2csWLAAQ4YMUW/z5OTkMmMpuTBW0zIvXLgAOzs7WFhYPDcGT09PCCHg4eGh/jb3PO3bt0f79u3x0Ucf4ZdffoGvry9iYmIwf/78Mqdp3bo1tmzZUqq9ovspAFhYWGD06NEYPXo0ioqKMGLECCxYsABhYWFl7ndlMTY2xoABAzBgwAB89tlnWLhwIf71r3/h4MGDkm2dlpYGY2PjCm0XkmIO0ow5qDRd5iBPT89y30CyefNmNGvWDFu3bpVs5/DwcI3ja9rHLl68WOqLo0qlwpUrVyTrfPHiRQBQj1vRfbHk95Camio5Snfz5s0yC9S0tDT1ke6K4qlPLTIxMSl1mHfTpk3IyMiQtPn7+yMjIwM7d+5UtxUUFJR6SKOPjw88PT3x6aef4t69e6WWV/IqCltbW/j5+UkGuVwOZ2dndOzYEd9++63kUPGBAwdw7tw5ybxGjRqF4uJi/Pvf/y61nMePHyM3N7diG0GDNm3aoH379tiwYQM2bNgAZ2dnvPjii+p+ExMTBAYGYsuWLRr/eJ9+5cbQoUORmZmJzZs3q9sePHiAVatWlRtHcXFxqUPmDg4OcHFxUd9a37lzZ3h4eCAqKqrUOpf8bp/erk+Pk5ycjP3792Po0KHlxjJixAiYmJhg3rx5pfYZIQRu374NAMjPz8fjx48l/e3bt4exsXG5jwPo0aMH7ty5IzmtAlR8Py2JoYSZmRm8vLwghMCjR4/K3O80ycnJKdVWcoTm2fVISkpC27Ztn3s3GWnGHKQZc1BpusxBgYGBOHPmjMbHtJTEUnKU7+nYEhMTcezYMY3L2r59u2Q///XXX5GYmKjxKOXy5csly1u+fDlMTU0xYMAAABXfF/38/GBqaoply5ZJ4oyKitIYI/Akv/Xo0aPMfk14RE2Lhg0bhoiICAQHB6Nnz544e/Ys1q5dW+p8+OTJk7F8+XKMHTsW06ZNg7OzM9auXav+J1fy7cHY2BirV6/GkCFD0LZtWwQHB6Nx48bIyMjAwYMHYW1tjR9++OG5MUVGRiIgIAC9evXC+PHjkZOTo3421tOJt0+fPpg8eTIiIyNx+vRpDBo0CKampkhNTcWmTZvw+eefSy6erazRo0djzpw5kMvlmDBhgvqC5xIff/wxDh48iG7dumHixInw8vJCTk4OTp48ifj4ePU/+okTJ2L58uV44403kJSUBGdnZ3z33XcwNzcvN4a7d+/C1dUVf//739GhQwdYWloiPj4eJ06cwJIlSwA82eZffvklXnrpJXTs2BHBwcFwdnbGhQsX8Mcff2Dfvn0AnpwKGjJkCHr06IEJEyaob41XKpWYO3duubF4enpi/vz5CAsLw9WrVzF8+HBYWVkhLS0N27Ztw6RJkzBr1iz8+OOPCA0NxciRI9GyZUs8fvwY3333nfofy/MEBASgQYMGiI+Pl1zoXNH9dNCgQXBycoKvry8cHR1x/vx5LF++HAEBAbCysip3HZ8WERGBI0eOICAgAO7u7rhx4wZWrFgBV1dXyTOdHj16hMOHD+Odd96p1PzpCeagsjEHSekyB7333nvYvHkzRo4cifHjx8PHxwc5OTnYuXMnYmJi0KFDBwwbNgxbt27F3/72NwQEBCAtLQ0xMTHw8vLS+KWhefPm6NWrF95++20UFhYiKioKjRo1wvvvvy8ZTy6XIy4uDkFBQejWrRv27t2L3bt348MPP1Qf5arovmhvb49Zs2YhMjISw4YNw9ChQ3Hq1Cns3btX46nNGzdu4Pfff0dISEi5vx+JCt8famDKujXewsKi1Lh9+vTR+JgCd3d3ERAQoP5cUFAg3n33XeHs7CwUCoXw9fUVx44dE3369BF9+vSRTHvlyhUREBAgFAqFsLe3F++++67YsmWLACCOHz8uGffUqVNixIgRolGjRkImkwl3d3cxatQokZCQUKF13bJli2jTpo2QyWTCy8tLbN26VQQFBUlujS+xatUq4ePjIxQKhbCyshLt27cX77//vsjMzCxzvZ/dJppuSU5NTRUABADx888/a5w2OztbhISECDc3N2FqaiqcnJzEgAEDxKpVqyTj/fnnn+Lll18W5ubmws7OTkybNk192/Tzbo0vLCwU7733nujQoYOwsrISFhYWokOHDmLFihWlxv3555/FwIED1eN5e3uLZcuWScaJj48Xvr6+QqFQCGtra/HSSy+Jc+fOScYpuTW+rMdSbNmyRfTq1UtYWFgICwsL0bp1axESEiJSUlKEEE/2k/HjxwtPT08hl8tFw4YNRb9+/UR8fHyZ6/m0l19+WQwYMEDSVtH9dOXKleLFF19U73eenp7ivffeE3l5eeUu99nHcyQkJIhXXnlFuLi4CDMzM+Hi4iLGjh0rLl68KJlu7969AoBITU2t0PrVZcxBzEH1NQcJIcTt27dFaGioaNy4sTAzMxOurq4iKChI/XgUlUolFi5cKNzd3YVMJhOdOnUSu3btKrXflDyeY/HixWLJkiXCzc1NyGQy0bt3b3HmzBnJMkv+vi5fviwGDRokzM3NhaOjowgPDy/1WBshKrYvFhcXi3nz5qn/5vr27SuSk5M17odffvmlMDc3F/n5+RXadiWMhNBwSwbpRFRUFGbMmIHr169LbjMmqqqffvoJffv2xYULFzTeFaVvhg8fDiMjozKfXE81izmItK2mc9DVq1fh4eGBxYsXY9asWc8dd9y4cdi8ebPGI3K1oVOnTujbty+WLl1aqel4jZqOPPvC6oKCAqxcuRItWrRggiSt6d27NwYNGoRFixbpOpRylTx5XdN1IaR9zEFUG+pSDqpJcXFxSE1NRVhYWKWn5TVqOjJixAg0adIEHTt2RF5eHr7//ntcuHChzNuBiapq7969ug6hQtq0aVPqomWqOcxBVFvqSg6qSYMHD67ykTwWajri7++P1atXY+3atSguLoaXlxfWr1+vflI2EVFNYg4iqht4jRoRERGRnuI1akRERER6ioUaERERkZ4y+GvUVCoVMjMzYWVlVW9f8kxkSIQQuHv3LlxcXEo9tLSuY74iMizayFcGX6hlZmbCzc1N12EQkZalp6c/94XQdRHzFZFhqk6+MvhCreRVN+np6bC2ttZxNERUXfn5+XBzc6v0a6zqAuYrIsOijXxl8IVayekDa2trJj4iA2KIpwaZr4gMU3XylWFd4EFERERkQFioEREREekpgz/1WRepVCpkZGQAABo3bmxwd7YRERFRxbBQ00MZGRkYv2IfAOCbd/x5FxgRUR3zysixyLyZU2a/i31D7Nj031qMiOoqFmp6SqG003UIRERURZk3c+D56twy+y+vK7uP6Gk8p0ZERESkp1ioERFVQEZGBl5//XU0atQICoUC7du3x2+//abuF0Jgzpw5cHZ2hkKhgJ+fH1JTU3UYMREZAhZqRETluHPnDnx9fWFqaoq9e/fi3LlzWLJkCWxtbdXjLFq0CF988QViYmKQmJgICwsL+Pv7o6CgQIeRE1Fdx2vUiIjK8cknn8DNzQ2xsbHqNg8PD/XPQghERUXho48+wiuvvAIA+M9//gNHR0ds374dY8aMqfWYicgw8IgaEVE5du7ciS5dumDkyJFwcHBAp06d8NVXX6n709LSkJWVBT8/P3WbUqlEt27dcOzYsTLnW1hYiPz8fMlARPQ0FmpEemTcuHEwMjLCxx9/LGnfvn17jb4yKScnB1OmTEGrVq2gUCjQpEkTTJ06FXl5eZLxjIyMSg3r16+XjHPo0CF07twZMpkMzZs3x5o1a2os7tpy5coVfPnll2jRogX27duHt99+G1OnTsW3334LAMjKygIAODo6SqZzdHRU92kSGRkJpVKpHvgoHjIkzGfawUKNSM/I5XJ88sknuHPnTq0tMzMzE5mZmfj000+RnJyMNWvWIC4uDhMmTCg1bmxsLP766y/1MHz4cHVfWloaAgIC0K9fP5w+fRrTp0/Hm2++iX379tXautQElUqFzp07Y+HChejUqRMmTZqEiRMnIiYmplrzDQsLQ15ennpIT0/XUsRE+oH5rPp4jRqRnvHz88OlS5cQGRmJRYsW1coy27Vrhy1btqg/e3p6YsGCBXj99dfx+PFjNGjwv1RhY2MDJycnjfOJiYmBh4cHlixZAgBo06YNfv75ZyxduhT+/v41uxI1yNnZGV5eXpK2Nm3aqLdZyfbIzs6Gs7Ozepzs7Gx07NixzPnKZDLIZDLtB0ykJ5jPqo9H1Ij0jImJCRYuXIhly5bh+vXrFZ5uyJAhsLS0LHNo27ZtpeLIy8uDtbW1JKkBQEhICOzs7NC1a1d88803EEKo+44dOya5TgsA/P39n3udVl3g6+uLlJQUSdvFixfh7u4O4MmNBU5OTkhISFD35+fnIzExET169KjVWIn0CfNZ9fGIGpEe+tvf/oaOHTsiPDwcX3/9dYWmWb16NR4+fFhmv6mpaYWXf+vWLfz73//GpEmTJO0RERHo378/zM3NsX//frzzzju4d+8epk6dCuDJtVqartPKz8/Hw4cPoVAoKhyDPpkxYwZ69uyJhQsXYtSoUfj111+xatUqrFq1CsCTa12mT5+O+fPno0WLFvDw8MDs2bPh4uIiOZVCVB8xn1UPCzUiPfXJJ5+gf//+mDVrVoXGb9y4sVaWm5+fj4CAAHh5eWHu3LmSvtmzZ6t/7tSpE+7fv4/FixerE5uheuGFF7Bt2zaEhYUhIiICHh4eiIqKwmuvvaYe5/3338f9+/cxadIk5ObmolevXoiLi4NcLtdh5ET6gfms6njqk0hPvfjii/D390dYWFiFxtfGqYK7d+9i8ODBsLKywrZt28r91tqtWzdcv34dhYWFAJ5cq5WdnS0ZJzs7G9bW1nX2aFqJYcOG4ezZsygoKMD58+cxceJESb+RkREiIiKQlZWFgoICxMfHo2XLljqKlki/MJ9VHY+oEemxjz/+GB07dkSrVq3KHbe6pwry8/Ph7+8PmUyGnTt3VuhI0OnTp2Fra6u+IL5Hjx7Ys2ePZJwDBw7wOi0iYj6rIhZqRHqsffv2eO211/DFF1+UO251ThXk5+dj0KBBePDgAb7//nvJw1ft7e1hYmKCH374AdnZ2ejevTvkcjkOHDiAhQsXSk5lvPXWW1i+fDnef/99jB8/Hj/++CM2btyI3bt3Vzk2IjIMzGdVw0KNSM9FRERgw4YNNbqMkydPIjExEQDQvHlzSV9aWhqaNm0KU1NTREdHY8aMGRBCoHnz5vjss88kpwA9PDywe/duzJgxA59//jlcXV2xevXqOv1oDiLSHuazyjMST9+LaoDy8/OhVCrVt+bWBenp6QhZmwQAiH7Nh08rJ3pKXfybrihDXrf65oW+/vB8dW6Z/ZfXzcWJQ3X7QdBUPm38TfNmAiIiIiI9xUKNiIiISE/pTaH28ccfqx8aWaKgoAAhISFo1KgRLC0tERgYWOpWWSIiIiJDpReF2okTJ7By5Up4e3tL2mfMmIEffvgBmzZtwuHDh5GZmYkRI0boKEoiIiKi2qXzQu3evXt47bXX8NVXX8HW1lbdnpeXh6+//hqfffYZ+vfvDx8fH8TGxuKXX37B8ePHdRgxERERUe3QeaEWEhKCgICAUi8+TUpKwqNHjyTtrVu3RpMmTZ77QtTCwkL1M1OefnYKERERUV2j0+eorV+/HidPnsSJEydK9WVlZcHMzAw2NjaSdkdHR2RlZZU5z8jISMybN0/boRIRERHVOp0dUUtPT8e0adOwdu1arb60OCwsDHl5eeohPT1da/MmIiIiqk06K9SSkpJw48YNdO7cGQ0aNECDBg1w+PBhfPHFF2jQoAEcHR1RVFSE3NxcyXTZ2dlwcnIqc74ymQzW1taSgYiIiKgu0tmpzwEDBuDs2bOStuDgYLRu3RoffPAB3NzcYGpqioSEBAQGBgIAUlJScO3aNb7gmYiIiOoFnRVqVlZWaNeunaTNwsICjRo1UrdPmDABM2fORMOGDWFtbY0pU6agR48e6N69uy5CJiIiIqpVev1S9qVLl8LY2BiBgYEoLCyEv78/VqxYoeuwiIiIiGqFXhVqhw4dknyWy+WIjo5GdHS0bgIiIiIi0iGdP0eNiIiIiDTTqyNqJCVUKmRmZqo/N27cGMbGrK2JiIjqCxZqeqwgPwfvb8yC0v4vPMy7hW/e8Yebm5uuwyIiIqJawkJNz8mVdjBv6KjrMIiIiEgHeB6NiIiISE+xUCMiIiLSUyzUiIiIiPQUCzUiIiIiPVWlQq1Zs2a4fft2qfbc3Fw0a9as2kHVRyqVCunp6UhPT3/ySA6h64iIDAPzFRHVZVW66/Pq1asoLi4u1V5YWIiMjIxqB1UfZWRkYPyKfVAo7XAn/SLMHZvqOiQig8B8RUR1WaUKtZ07d6p/3rdvH5RKpfpzcXExEhIS0LRpU60FV98o/v9RHA9zb+k6FKI6rybz1ccff4ywsDBMmzYNUVFRAICCggK8++67WL9+veTdxI6OfLwOEVVdpQq14cOHAwCMjIwQFBQk6TM1NUXTpk2xZMkSrQVHRFRVNZWvTpw4gZUrV8Lb21vSPmPGDOzevRubNm2CUqlEaGgoRowYgaNHj1Z5HYiIKlWoqVQqAICHhwdOnDgBOzu7GgmKiKi6aiJf3bt3D6+99hq++uorzJ8/X92el5eHr7/+GuvWrUP//v0BALGxsWjTpg2OHz+O7t27V3vZRFQ/VelmgrS0NBZpWsAbCIhqnjbzVUhICAICAuDn5ydpT0pKwqNHjyTtrVu3RpMmTXDs2LEy51dYWIj8/HzJQET0tCq/QiohIQEJCQm4ceOG+ptriW+++abagdUHmm4gMNd1UEQGSBv5av369Th58iROnDhRqi8rKwtmZmawsbGRtDs6OiIrK6vMeUZGRmLevHkVWj4R1U9VOqI2b948DBo0CAkJCbh16xbu3LkjGajiSm4gkFs11HUoRAZJG/kqPT0d06ZNw9q1ayGXy7UWW1hYGPLy8tRDenq61uZNRIahSkfUYmJisGbNGvzjH//QdjxERFqljXyVlJSEGzduoHPnzuq24uJiHDlyBMuXL8e+fftQVFSE3NxcyVG17OxsODk5lTlfmUwGmUxW5biIyPBVqVArKipCz549tR0LEZHWaSNfDRgwAGfPnpW0BQcHo3Xr1vjggw/g5uYGU1NTJCQkIDAwEACQkpKCa9euoUePHtVaNhHVb1U69fnmm29i3bp12o6FiEjrtJGvrKys0K5dO8lgYWGBRo0aoV27dlAqlZgwYQJmzpyJgwcPIikpCcHBwejRowfv+CSiaqnSEbWCggKsWrUK8fHx8Pb2hqmpqaT/s88+00pwRETVVVv5aunSpTA2NkZgYKDkgbdERNVRpULt999/R8eOHQEAycnJkj4jI6NqB0VEpC01la8OHTok+SyXyxEdHY3o6Ogqz5OI6FlVKtQOHjyo7TiIiGoE8xUR1WVVukaNiIiIiGpelY6o9evX77mnDH788ccqB0REpE3MV6SPLl9KxQt9/cvsd7FviB2b/luLEZG+qlKhVnK9R4lHjx7h9OnTSE5OLvXyYyIiXWK+In30WBjB89W5ZfZfXld2H9UvVSrUli5dqrF97ty5uHfvXrUCIiLSJuYrIqrLtHqN2uuvv873fBJRncB8RUR1gVYLtWPHjmn1PXhERDWF+YqI6oIqnfocMWKE5LMQAn/99Rd+++03zJ49WyuBERFpA/MVEdVlVSrUlEql5LOxsTFatWqFiIgIDBo0SCuBERFpA/MVEdVlVSrUYmNjtR0HEVGNYL4iorqsSoVaiaSkJJw/fx4A0LZtW3Tq1EkrQRERaRvzFRHVRVUq1G7cuIExY8bg0KFDsLGxAQDk5uaiX79+WL9+Pezt7bUZIxFRlTFfEVFdVqW7PqdMmYK7d+/ijz/+QE5ODnJycpCcnIz8/HxMnTpV2zESEVUZ8xUR1WVVOqIWFxeH+Ph4tGnTRt3m5eWF6OhoXpxLRHqF+YqI6rIqHVFTqVQwNTUt1W5qagqVSlXtoIiItIX5iojqsioVav3798e0adOQmZmpbsvIyMCMGTMwYMAArQVHRFRdzFdEVJdVqVBbvnw58vPz0bRpU3h6esLT0xMeHh7Iz8/HsmXLKjyfyMhIvPDCC7CysoKDgwOGDx+OlJQUyTgFBQUICQlBo0aNYGlpicDAQGRnZ1clbCKqh7SVr4iIdKFK16i5ubnh5MmTiI+Px4ULFwAAbdq0gZ+fX6Xmc/jwYYSEhOCFF17A48eP8eGHH2LQoEE4d+4cLCwsAAAzZszA7t27sWnTJiiVSoSGhmLEiBE4evRoVUInonpGW/mKiEgXKlWo/fjjjwgNDcXx48dhbW2NgQMHYuDAgQCAvLw8tG3bFjExMejdu3eF5hcXFyf5vGbNGjg4OCApKQkvvvgi8vLy8PXXX2PdunXo378/gCcPr2zTpg2OHz+O7t27VyZ8IqpHtJ2viIh0oVKnPqOiojBx4kRYW1uX6lMqlZg8eTI+++yzKgeTl5cHAGjYsCGAJw+ofPTokeSbb+vWrdGkSRMcO3ZM4zwKCwuRn58vGYio/qnpfEVEVBsqVaidOXMGgwcPLrN/0KBBSEpKqlIgKpUK06dPh6+vL9q1awcAyMrKgpmZmfohlSUcHR2RlZWlcT6RkZFQKpXqwc3NrUrxEFHdVpP5ioiotlSqUMvOztZ4m3uJBg0a4ObNm1UKJCQkBMnJyVi/fn2Vpi8RFhaGvLw89ZCenl6t+RFR3VST+YqIqLZUqlBr3LgxkpOTy+z//fff4ezsXOkgQkNDsWvXLhw8eBCurq7qdicnJxQVFSE3N1cyfnZ2NpycnDTOSyaTwdraWjIQUf1TU/mKiKg2VepmgqFDh2L27NkYPHgw5HK5pO/hw4cIDw/HsGHDKjw/IQSmTJmCbdu24dChQ/Dw8JD0+/j4wNTUFAkJCQgMDAQApKSk4Nq1a+jRo0dlQieiekbb+Yrql1dGjkXmzZwy+zOuXUXjJk3L7L+SdhWeNRAX1T+VKtQ++ugjbN26FS1btkRoaChatWoFALhw4QKio6NRXFyMf/3rXxWeX0hICNatW4cdO3bAyspKfd2ZUqmEQqGAUqnEhAkTMHPmTDRs2BDW1taYMmUKevTowTs+iei5tJ2vqH7JvJkDz1fnltmfMu/1cvuJtKFShZqjoyN++eUXvP322wgLC4MQAgBgZGQEf39/REdHw9HRscLz+/LLLwEAffv2lbTHxsZi3LhxAIClS5fC2NgYgYGBKCwshL+/P1asWFGZsImoHtJ2viIi0oVKP/DW3d0de/bswZ07d3Dp0iUIIdCiRQvY2tpWeuElifN55HI5oqOjER0dXen5E1H9ps18FRkZia1bt+LChQtQKBTo2bMnPvnkE/WROuDJm1TeffddrF+/XvLFkgUhEVVVlV4hBQC2trZ44YUX0LVr1yolPSKi2qKNfFXyJpXjx4/jwIEDePToEQYNGoT79++rx5kxYwZ++OEHbNq0CYcPH0ZmZiZGjBihrdUgonqoSq+QIv2gUqmQkZEB4MkdbsbGVa67iagcfJMKEekC/7PXYRkZGRi/Yh/Gr9inLtiIqHbwTSpEVBtYqNVxCqUdFEo7XYdBVK/wTSpEVFtYqBERVRLfpEJEtYXXqBERVULJm1SOHDlS5ptUnj6qVt6bVGQyWU2HTHXQ5UupeKGvf5n9LvYNsWPTf2sxItIVFmp1hFCpkJmZCYA3DhDpAt+kQrXpsTB67gN1L68ru48MCwu1OqIgPwfvb8yCmdlZfPOOP69lIaplfJMKEekCC7U6RK60g8zMTNdhENVLfJMKEekCCzUiogrgm1SISBd4oRMRERGRnuIRtTrm6ZsKMjMzAQEI8b82gDcbEBHVd6+MHIvMmzll9mdcu4rGTZqW2c+7SvUHC7U6puSmAqX9X7iTfhHmjk2hepCvbnuYd4s3GxAR1XOZN3Oee9doyrzXeVdpHcFCrQ6SK+1g3tARD3NvlWojIiIiw8HzY0RERER6ioUaERERkZ5ioUZERESkp1ioEREREekpFmpEREREeoqFGhEREZGeYqFGREREpKf4HLUaoFKpkJGRof7cuHFjAFC3qVQqAEBWVhZQ/usDiYiIqJ5ioVYDMjIyMH7FPiiUduo3BQBQt91JvwhjuSVUBfdg7tgU5jqOl4iIiPQTC7UaotDwpgDFU28UMDa3hsrUTEfRERERUV3Aa9SIiIiI9BQLNSIiIiI9xVOfRERkkF4ZORaZN3M09rnYN8SOTf+t5Yi05/KlVLzQ17/M/itpV+FZi/FQzWGhRkREBinzZg48X52rse/yOs3tdcVjYVTmugFAyrzXay8YqlE89UlERESkp3hEjYiI9NLzTl0C1Tt9yVOHVFewUHvK0w+qbdy4MYyNjSvVr4lQqZCZmfn/H7QbLxGRIXveqUugeqcveeqQ6goWak8peVAtAHzzjj/c3Nwq1a9JQX4O3t+YxYfbEhERUaWxUHuGQmlXrX5N5Eo7PtyWiIiIKo2FGhFRHVWT13DVxvyri9eZ1Zzytq2uf/c1TZ/2fRZqRER1VE1ew1Ub868uXmdWc8rbtrr+3dc0fdr3WagZGMnNC3hy0wMA9U0Qz2uryM0RJZ6+saIq0xMREVH5WKgZmJKbF5T2f+Fh3i18886TQ9fjV+yDQmn33LaK3BxRouTGiqpOT0REROWrE4VadHQ0Fi9ejKysLHTo0AHLli1D165ddR2W3pIr7WDe0FHSpqhgW2VUd3oyHFU5wlqVx93UBcxXROWr7jVg+nQNWU3T+0Jtw4YNmDlzJmJiYtCtWzdERUXB398fKSkpcHBw0HV4RISqHWGtyuNu9B3zFVHFVPcaMH26hqym6f1X2M8++wwTJ05EcHAwvLy8EBMTA3Nzc3zzzTe6Do2InlJyhLUyj7BRKO2q9MgbfcV8RUTaptdH1IqKipCUlISwsDB1m7GxMfz8/HDs2DGN0xQWFqKwsFD9OS8vDwCQn59f7vLu3r2LuzfSAQCpqda4e/eupD8rK+u5/c+O96jgAe7dzoTxvTyoCu/D+F4ejFWPn9tWXn9l5lOQfxupqdZP1u3/43leW1nrU946VmV6MixV2R+e/nu6e7dlhf5Ggf/9LQuhX6/6qO18BQDFjx/j0cP7z+2v6Lx0Mf/qLl+oVFXur8607Nf9vlVX9n2t5CuhxzIyMgQA8csvv0ja33vvPdG1a1eN04SHhws8eVkTBw4cDHhIT0+vjTRUYcxXHDhwKGuoTr7S6yNqVREWFoaZM2eqP6tUKuTk5KBRo0YwMjLSYWSVk5+fDzc3N6Snp8Pa2lrX4RgkbuOaVxPbWAiBu3fvwsXFRSvz06Wq5itD33e5fnUb1+9/tJGv9LpQs7Ozg4mJCbKzsyXt2dnZcHJy0jiNTCaDTCaTtNnY2NRUiDXO2traIHd0fcJtXPO0vY2VSqXW5qUtushXhr7vcv3qNq7fE9XNV3p9M4GZmRl8fHyQkJCgblOpVEhISECPHj10GBkRkRTzFRHVBL0+ogYAM2fORFBQELp06YKuXbsiKioK9+/fR3BwsK5DIyKSYL4iIm3T+0Jt9OjRuHnzJubMmYOsrCx07NgRcXFxcHQ07AetymQyhIeHlzotQtrDbVzz6ts2rq18ZejbletXt3H9tMtICD27x52IiIiIAOj5NWpERERE9RkLNSIiIiI9xUKNiIiISE+xUCMiIiLSUyzU9FR0dDSaNm0KuVyObt264ddff9V1SAZj7ty5MDIykgytW7fWdVh11pEjR/DSSy/BxcUFRkZG2L59u6RfCIE5c+bA2dkZCoUCfn5+SE1N1U2wBubq1auYMGECPDw8oFAo4OnpifDwcBQVFek6NK1ZsGABevbsCXNz8zr98PIShpzby8sFdVlkZCReeOEFWFlZwcHBAcOHD0dKSkqtLJuFmh7asGEDZs6cifDwcJw8eRIdOnSAv78/bty4oevQDEbbtm3x119/qYeff/5Z1yHVWffv30eHDh0QHR2tsX/RokX44osvEBMTg8TERFhYWMDf3x8FBQW1HKnhuXDhAlQqFVauXIk//vgDS5cuRUxMDD788ENdh6Y1RUVFGDlyJN5++21dh1Jthp7by8sFddnhw4cREhKC48eP48CBA3j06BEGDRqE+/fLfnG71lT5LaFUY7p27SpCQkLUn4uLi4WLi4uIjIzUYVSGIzw8XHTo0EHXYRgkAGLbtm3qzyqVSjg5OYnFixer23Jzc4VMJhP//e9/dRCh4Vu0aJHw8PDQdRhaFxsbK5RKpa7DqJb6lNufzQWG5saNGwKAOHz4cI0vi0fU9ExRURGSkpLg5+enbjM2Noafnx+OHTumw8gMS2pqKlxcXNCsWTO89tpruHbtmq5DMkhpaWnIysqS7M9KpRLdunXj/lxD8vLy0LBhQ12HQc9gbjcseXl5AFArf2ss1PTMrVu3UFxcXOpJ5o6OjsjKytJRVIalW7duWLNmDeLi4vDll18iLS0NvXv3xt27d3UdmsEp2We5P9eOS5cuYdmyZZg8ebKuQ6FnMLcbDpVKhenTp8PX1xft2rWr8eWxUKN6Z8iQIRg5ciS8vb3h7++PPXv2IDc3Fxs3btR1aEQAgH/+85+lbnh5drhw4YJkmoyMDAwePBgjR47ExIkTdRR5xVRl/Yj0RUhICJKTk7F+/fpaWZ7ev+uzvrGzs4OJiQmys7Ml7dnZ2XByctJRVIbNxsYGLVu2xKVLl3QdisEp2Wezs7Ph7Oysbs/OzkbHjh11FJX+e/fddzFu3LjnjtOsWTP1z5mZmejXrx969uyJVatW1XB01VfZ9TMEzO2GITQ0FLt27cKRI0fg6upaK8tkoaZnzMzM4OPjg4SEBAwfPhzAk8OsCQkJCA0N1W1wBurevXu4fPky/vGPf+g6FIPj4eEBJycnJCQkqAuz/Px8JCYmGsRdfDXF3t4e9vb2FRo3IyMD/fr1g4+PD2JjY2FsrP8nSiqzfoaCub1uE0JgypQp2LZtGw4dOgQPD49aWzYLNT00c+ZMBAUFoUuXLujatSuioqJw//59BAcH6zo0gzBr1iy89NJLcHd3R2ZmJsLDw2FiYoKxY8fqOrQ66d69e5KjkWlpaTh9+jQaNmyIJk2aYPr06Zg/fz5atGgBDw8PzJ49Gy4uLup/VlR1GRkZ6Nu3L9zd3fHpp5/i5s2b6j5DOUpz7do15OTk4Nq1ayguLsbp06cBAM2bN4elpaVug6skQ8/t5eWCuiwkJATr1q3Djh07YGVlpb6uUKlUQqFQ1OzCa/y+UqqSZcuWiSZNmggzMzPRtWtXcfz4cV2HZDBGjx4tnJ2dhZmZmWjcuLEYPXq0uHTpkq7DqrMOHjwoAJQagoKChBBPHtExe/Zs4ejoKGQymRgwYIBISUnRbdAGIjY2VuO2N6TUHhQUpHH9Dh48qOvQqsSQc3t5uaAuK+vvLDY2tsaXbfT/ARARERGRntH/ixmIiIiI6ikWakRERER6ioUaERERkZ5ioUZERESkp1ioEREREekpFmpEREREeoqFGhEREZGeYqFGREREpKdYqBERERHpKRZqpBdu3ryJt99+G02aNIFMJoOTkxP8/f1x9OhRXYdGRCTBfEW1iS9lJ70QGBiIoqIifPvtt2jWrBmys7ORkJCA27dv18jyioqKYGZmViPzJiLDxnxFtarG3yZKVI47d+4IAOLQoUPPHWfSpEnCwcFByGQy0bZtW/HDDz+o+zdv3iy8vLyEmZmZcHd3F59++qlkend3dxERESH+8Y9/CCsrK/VLgn/66SfRq1cvIZfLhaurq5gyZYq4d+9ejawnEdV9zFdU21iokc49evRIWFpaiunTp4uCgoJS/cXFxaJ79+6ibdu2Yv/+/eLy5cvihx9+EHv27BFCCPHbb78JY2NjERERIVJSUkRsbKxQKBQiNjZWPQ93d3dhbW0tPv30U3Hp0iX1YGFhIZYuXSouXrwojh49Kjp16iTGjRtXW6tORHUM8xXVNhZqpBc2b94sbG1thVwuFz179hRhYWHizJkzQggh9u3bJ4yNjUVKSorGaV999VUxcOBASdt7770nvLy81J/d3d3F8OHDJeNMmDBBTJo0SdL2008/CWNjY/Hw4UNtrBYRGSDmK6pNvJmA9EJgYCAyMzOxc+dODB48GIcOHULnzp2xZs0anD59Gq6urmjZsqXGac+fPw9fX19Jm6+vL1JTU1FcXKxu69Kli2ScM2fOYM2aNbC0tFQP/v7+UKlUSEtL0/5KEpFBYL6i2sSbCUhvyOVyDBw4EAMHDsTs2bPx5ptvIjw8HLNmzdLK/C0sLCSf7927h8mTJ2Pq1Kmlxm3SpIlWlklEhon5imoLCzXSW15eXti+fTu8vb1x/fp1XLx4UeO31DZt2pS6Lf7o0aNo2bIlTExMypx/586dce7cOTRv3lzrsRNR/cJ8RTVG1+deiW7duiX69esnvvvuO3HmzBlx5coVsXHjRuHo6CjGjx8vhBCib9++ol27dmL//v3iypUrYs+ePWLv3r1CCCGSkpIkF+euWbNG48W5S5culSz3zJkzQqFQiJCQEHHq1Clx8eJFsX37dhESElJbq05EdQzzFdU2FmqkcwUFBeKf//yn6Ny5s1AqlcLc3Fy0atVKfPTRR+LBgwdCCCFu374tgoODRaNGjYRcLhft2rUTu3btUs+j5HZ3U1NT0aRJE7F48WLJMjQlPiGE+PXXX8XAgQOFpaWlsLCwEN7e3mLBggU1ur5EVHcxX1FtMxJCCF0f1SMiIiKi0njXJxEREZGeYqFGREREpKdYqBERERHpKRZqRERERHqKhRoRERGRnmKhRkRERKSnWKgRERER6SkWakRERER6ioUaERERkZ5ioUZERESkp1ioEREREemp/wP2jVkj8z1RKgAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 600x200 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 900x300 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Inspect min and max values\n",
    "print('Absolute min: {}'.format(np.nanmin(zm)))\n",
    "print('Absolute max: {}'.format(np.nanmax(zm)))\n",
    "\n",
    "# Visualize score distribution (as-is)\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(6, 2), constrained_layout=True)\n",
    "scores = out['Van']['scores'].reshape(-1)\n",
    "sns.histplot(scores, ax=ax[0])\n",
    "ax[0].set(xlabel='Score', ylabel='Count', title='Image-derived scores (as-is)')\n",
    "ax[0].text(7, 43, 'N = {}'.format(sum(~np.isnan(scores))))\n",
    "\n",
    "# Cap upper end\n",
    "zm[zm > 2] = 2\n",
    "\n",
    "# Visualize score distribution (capped)\n",
    "scores = zm.reshape(-1)\n",
    "sns.histplot(scores, ax=ax[1])\n",
    "ax[1].set(xlabel='Score', ylabel='Count', title='Image-derived scores (capped)')\n",
    "ax[1].text(1, 55, 'N = {}'.format(sum(~np.isnan(scores))))\n",
    "\n",
    "# Sanity check #2\n",
    "fig, ax = plt.subplots(nrows=1, ncols=3, figsize=(9, 3), constrained_layout=True)\n",
    "ax[0].imshow(image)\n",
    "ax[0].set_title('Original heatmap ({} x {})'.format(image.shape[0], image.shape[1]))\n",
    "ax[1].imshow(hm)\n",
    "ax[1].set_title('Reduced heatmap ({} x {})'.format(hm.shape[0], hm.shape[1]))\n",
    "nan_mask = np.isnan(zm)\n",
    "cmap = sns.color_palette('vlag', as_cmap=True).copy()\n",
    "cmap.set_bad('gray')\n",
    "sns.heatmap(zm, center=0, cmap=cmap, mask=nan_mask, ax=ax[2])\n",
    "ax[2].set_title('Inferred score heatmap')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7.3 Extract TACTIC predictions\n",
    "\n",
    "Extract predictions for M. abscessus that match drugs evaluated in Van et al. study."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define combinations\n",
    "drugs = sorted(list(set(list(abx.values()))))[1:]\n",
    "combos = []\n",
    "for k in range(2, 3):\n",
    "    combos.extend(combinations(drugs, k))\n",
    "combos = [', '.join(i) for i in combos]\n",
    "\n",
    "# Extract predictions\n",
    "df = pd.DataFrame(index=combos)\n",
    "df['Prediction'] = [out['landscape']['lb'].loc[i + ', LB', 'M. abscessus'] for i in combos]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7.4 Determine additional predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:337: UserWarning: Interaction entries do not match profile names, but key provided\n",
      "  warn('Interaction entries do not match profile names, but key provided')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:351: UserWarning: Not all feature names are unique. Enforcing unique entries for output\n",
      "  warn('Not all feature names are unique. Enforcing unique entries for output')\n",
      "c:\\Users\\carol\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\indigopy\\core.py:406: UserWarning: Drug profile information missing for 42 interactions\n",
      "  warn('Drug profile information missing for {} interactions'.format(len(interactions) - len(ixn_list)))\n",
      "Defining INDIGO features: 100%|██████████| 42/42 [00:00<?, ?it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 1/1 [00:01<00:00,  1.85s/it]\n",
      "Defining INDIGO features: 100%|██████████| 42/42 [00:00<00:00, 256.28it/s]\n",
      "Mapping orthologous genes: 100%|██████████| 1/1 [00:03<00:00,  3.46s/it]\n"
     ]
    }
   ],
   "source": [
    "# Update drug dictionary to include BDQ, EMBx, and RIFP\n",
    "new_drugs = ['BDQ', 'EMBx', 'RIFP']\n",
    "abx.update({'BDQ': new_drugs[0], 'EMB': new_drugs[1], 'RFB': new_drugs[2]})\n",
    "\n",
    "# Define interaction list\n",
    "new_combos = []\n",
    "for k in range(2, 3):\n",
    "    new_combos.extend(combinations(sorted(drugs + new_drugs), k))\n",
    "new_combos = [', '.join(i) for i in new_combos]\n",
    "interactions = [i for i in new_combos if i not in combos]\n",
    "ixn_list = [list(i.split(', ')) for i in interactions]\n",
    "str_list = ['M. abscessus'] * len(ixn_list)\n",
    "\n",
    "# Define new predictions\n",
    "p1, p2  = ecoli['omics'].iloc[:, 1:].to_dict('list'), mtb['omics'].iloc[:, 1:].to_dict('list')\n",
    "g1, g2  = ecoli['omics']['Gene'].tolist(), mtb['omics']['Gene'].tolist()\n",
    "f1      = featurize(ixn_list, p1, feature_names=g1, key=ecoli['key'], strains=str_list, orthology_map=ecoli['map'])\n",
    "f2      = featurize(ixn_list, p2, feature_names=g2, key=mtb['key'], strains=str_list, orthology_map=mtb['map'])\n",
    "d1, d2  = f1['feature_df'].to_numpy().transpose(), f2['feature_df'].to_numpy().transpose()\n",
    "x1      = np.zeros([d2.shape[0], d1.shape[1]], dtype=d2.dtype)\n",
    "x1[np.array(f1['idx'])] = d1[np.array(f1['idx'])]\n",
    "idx1    = [f in master['df'].index for f in list(f1['feature_df'].index)]\n",
    "idx2    = [f in master['df'].index for f in list(f2['feature_df'].index)]\n",
    "s1, s2  = sparse.csr_matrix(x1[:, np.array(idx1)]), sparse.csr_matrix(d2[:, np.array(idx2)])\n",
    "X       = sparse.hstack((s1, s2))\n",
    "mab_pred = out['model']['TACTIC'].predict(X)\n",
    "\n",
    "# Add new predictions to prediction dataframe\n",
    "df2 = pd.DataFrame({'Prediction': mab_pred}, index=interactions)\n",
    "df = pd.concat([df, df2])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7.5 Re-create heatmap based on predictions\n",
    "\n",
    "Mimic Figure 2A based on TACTIC predictions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'TACTIC prediction heatmap')"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 700x300 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Re-create heatmaps\n",
    "phm = pd.DataFrame(np.nan, index=abx.keys(), columns=abx.keys())\n",
    "\n",
    "for i in df.index: \n",
    "    # Extract individual drugs\n",
    "    d1, d2 = i.split(', ')\n",
    "    # Map to Figure 2A drug names\n",
    "    row = list(abx.keys())[list(abx.values()).index(d1)]\n",
    "    col = list(abx.keys())[list(abx.values()).index(d2)]\n",
    "    # Populate dataframe\n",
    "    phm.loc[row, col] = df.loc[i, 'Prediction']\n",
    "    phm.loc[col, row] = df.loc[i, 'Prediction']\n",
    "out['Van']['predictions'] = phm\n",
    "\n",
    "# Visualize score vs. prediction heatmaps\n",
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(7, 3), width_ratios=(1, 1), constrained_layout=True)\n",
    "cmap = sns.color_palette('vlag', as_cmap=True).copy()\n",
    "cmap.set_bad('gray')\n",
    "nan_mask = np.isnan(zm)\n",
    "sns.heatmap(zm, center=0, cmap=cmap, mask=nan_mask, ax=ax[0])\n",
    "ax[0].set_title('Inferred score heatmap')\n",
    "nan_mask = phm.isna()\n",
    "sns.heatmap(phm, center=0, cmap=cmap, mask=nan_mask, ax=ax[1])\n",
    "ax[1].set_title('TACTIC prediction heatmap')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7.6 Define dataframe of unique interactions\n",
    "\n",
    "Create a dataframe including score + prediction for all unique interactions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total number of unique interactions: 97\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>Interaction</th>\n",
       "      <th>Score</th>\n",
       "      <th>Prediction</th>\n",
       "      <th>Class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>AMK, AZM</td>\n",
       "      <td>1.442105</td>\n",
       "      <td>0.098155</td>\n",
       "      <td>Antagonism</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>AMK, LZD</td>\n",
       "      <td>0.740506</td>\n",
       "      <td>-0.094703</td>\n",
       "      <td>Antagonism</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>AMK, MXF</td>\n",
       "      <td>0.195556</td>\n",
       "      <td>0.404542</td>\n",
       "      <td>Antagonism</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>AMK, LXF</td>\n",
       "      <td>0.045643</td>\n",
       "      <td>0.427073</td>\n",
       "      <td>Antagonism</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>AMK, NFT</td>\n",
       "      <td>-0.121212</td>\n",
       "      <td>0.107741</td>\n",
       "      <td>Synergy</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Interaction     Score  Prediction       Class\n",
       "1    AMK, AZM  1.442105    0.098155  Antagonism\n",
       "3    AMK, LZD  0.740506   -0.094703  Antagonism\n",
       "4    AMK, MXF  0.195556    0.404542  Antagonism\n",
       "5    AMK, LXF  0.045643    0.427073  Antagonism\n",
       "7    AMK, NFT -0.121212    0.107741     Synergy"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Convert score matrix to dataframe\n",
    "zdf = pd.DataFrame(zm, index=abx.keys(), columns=abx.keys())\n",
    "\n",
    "# Extract unique values\n",
    "idx = np.triu_indices(zdf.shape[0], k=1)\n",
    "ixns = zdf.index.values[idx[0]] + ', ' + zdf.columns.values[idx[1]]\n",
    "scores, pred = zdf.values[idx], phm.values[idx]\n",
    "\n",
    "# Define dataframe\n",
    "pdf = pd.DataFrame({'Interaction': ixns, 'Score': scores, 'Prediction': pred})\n",
    "pdf['Class'] = classify(pdf['Score'].tolist(), thresholds=(0, 0))\n",
    "pdf.dropna(inplace=True)\n",
    "print('Total number of unique interactions: {}'.format(pdf.shape[0]))\n",
    "display(pdf.head())\n",
    "\n",
    "# Save in out variable\n",
    "out['Van']['table'] = pdf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Checkpoint\n",
    "with open('./TACTIC_data.pkl', 'wb') as f: \n",
    "    pickle.dump([ecoli, mtb, master, out], f)"
   ]
  }
 ],
 "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": "3.10.8"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}