gene_level_task_table_1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "cdec6a12",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "sns.set_style(\"whitegrid\")\n",
    "import pandas as pd\n",
    "import numpy as np \n",
    "import scipy.stats as stats\n",
    "from collections import Counter\n",
    "import matplotlib.pyplot as plt\n",
    "import umap\n",
    "import matplotlib\n",
    "import mygene\n",
    "%matplotlib inline\n",
    "import pickle\n",
    "import sklearn\n",
    "import random\n",
    "from sklearn.model_selection import StratifiedKFold\n",
    "from sklearn.metrics import roc_curve, auc\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import cross_validate\n",
    "from xgboost import XGBClassifier\n",
    "# import sentence_transformers\n",
    "plt.style.use('ggplot')\n",
    "#plt.style.use('seaborn-v0_8-dark-palette')\n",
    "plt.rcParams['axes.facecolor'] = 'white'\n",
    "plt.rcParams.update({\n",
    "    \"text.usetex\": True,\n",
    "    \"font.family\": \"Helvetica\"\n",
    "})\n",
    "import matplotlib_inline\n",
    "matplotlib_inline.backend_inline.set_matplotlib_formats('retina')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "370b03b7",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(f\"/Users/yiquntchen/Downloads/data_embedding/GPT_3_5_gene_embeddings.pickle\", \"rb\") as fp:\n",
    "    GPT_3_5_gene_embeddings = pickle.load(fp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "25391584",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create random embeddings as negative controls\n",
    "np.random.seed(2023)\n",
    "N_DIM = 1536\n",
    "fake_pretrain_embed = dict()\n",
    "for key in GPT_3_5_gene_embeddings.keys():\n",
    "    fake_pretrain_embed[key] = np.random.normal(size=(N_DIM))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d890098f",
   "metadata": {},
   "source": [
    "### Long- vs short- range TFs\n",
    "The input data used here are downloaded from Chen et al. (2020) (link: https://www-nature-com.stanford.idm.oclc.org/articles/s41467-020-16106-x)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "a7e106aa",
   "metadata": {},
   "outputs": [],
   "source": [
    "long_short_range_tf = pd.read_csv('~/Downloads/41467_2020_16106_MOESM4_ESM.csv')\n",
    "long_range_tf_gene = list(long_short_range_tf[long_short_range_tf['assignment']=='long-range TF']\\\n",
    "                                ['Unnamed: 0'])\n",
    "short_range_tf_gene = list(long_short_range_tf[long_short_range_tf['assignment']=='short-range TF']\\\n",
    "                                ['Unnamed: 0'])\n",
    "x_long_range_tf = [GPT_3_5_gene_embeddings[x] for x in long_range_tf_gene \\\n",
    "               if x in GPT_3_5_gene_embeddings]\n",
    "x_short_range_tf =  [GPT_3_5_gene_embeddings[x] for x in short_range_tf_gene\\\n",
    "               if x in GPT_3_5_gene_embeddings]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "850ddec9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Logistic Regression ROC AUC: 0.726 +/- 0.261\n",
      "Random Forest ROC AUC: 0.635 +/- 0.070\n"
     ]
    }
   ],
   "source": [
    "# bivalent versus non-methylated\n",
    "np.random.seed(2023)\n",
    "random.seed(2023)\n",
    "\n",
    "# Assuming x and y are your data\n",
    "# For demonstration, let's create some dummy data.\n",
    "# Ensure your data is in NumPy array format for compatibility\n",
    "X_array = np.concatenate((x_long_range_tf,x_short_range_tf))\n",
    "y_array =  np.concatenate((np.repeat(1,len(x_long_range_tf)),np.repeat(0,len(x_short_range_tf))))\n",
    "\n",
    "# Set up Stratified K-Folds cross-validator\n",
    "# It provides train/test indices to split data into train/test sets\n",
    "cv = StratifiedKFold(n_splits=5)\n",
    "\n",
    "# Lists to store ROC AUC scores for each fold\n",
    "roc_auc_logistic = []\n",
    "roc_auc_rf = []\n",
    "\n",
    "# Lists to store TPR and FPR for each fold\n",
    "tpr_logistic = []\n",
    "fpr_logistic = []\n",
    "tpr_rf = []\n",
    "fpr_rf = []\n",
    "\n",
    "for train_index, test_index in cv.split(X_array, y_array):\n",
    "    X_train, X_test = X_array[train_index], X_array[test_index]\n",
    "    y_train, y_test = y_array[train_index], y_array[test_index]\n",
    "\n",
    "    # Logistic Regression\n",
    "    logistic_model = LogisticRegression()\n",
    "    logistic_model.fit(X_train, y_train)\n",
    "    y_score_logistic = logistic_model.predict_proba(X_test)[:, 1]\n",
    "    fpr, tpr, _ = roc_curve(y_test, y_score_logistic)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    roc_auc_logistic.append(roc_auc)\n",
    "    tpr_logistic.append(tpr)\n",
    "    fpr_logistic.append(fpr)\n",
    "\n",
    "    # Random Forest\n",
    "    random_forest_model = RandomForestClassifier()\n",
    "    random_forest_model.fit(X_train, y_train)\n",
    "    y_score_rf = random_forest_model.predict_proba(X_test)[:, 1]\n",
    "    fpr, tpr, _ = roc_curve(y_test, y_score_rf)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    roc_auc_rf.append(roc_auc)\n",
    "    tpr_rf.append(tpr)\n",
    "    fpr_rf.append(fpr)\n",
    "\n",
    "# Print ROC AUC scores\n",
    "print(f\"Logistic Regression ROC AUC: {np.mean(roc_auc_logistic):.3f} +/- {np.std(roc_auc_logistic):.3f}\")\n",
    "print(f\"Random Forest ROC AUC: {np.mean(roc_auc_rf):.3f} +/- {np.std(roc_auc_rf):.3f}\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2a80d457",
   "metadata": {},
   "source": [
    "## Dosage sensitive vs insensitive TFs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "ac3fa658",
   "metadata": {},
   "outputs": [],
   "source": [
    "link_file = \"https://huggingface.co/datasets/ctheodoris/Genecorpus-30M/raw/main/example_input_files/gene_classification/dosage_sensitive_tfs/dosage_sens_tf_labels.csv\"\n",
    "dosage_tfs = pd.read_csv(link_file, header=0)\n",
    "sensitive = dosage_tfs[\"dosage_sensitive\"].dropna()\n",
    "insensitive = dosage_tfs[\"dosage_insensitive\"].dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "cb7fb72a",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:biothings.client:querying 1-122...\n",
      "INFO:biothings.client:done.\n",
      "INFO:biothings.client:Finished.\n",
      "INFO:biothings.client:querying 1-368...\n",
      "INFO:biothings.client:done.\n",
      "INFO:biothings.client:Finished.\n",
      "WARNING:biothings.client:6 input query terms found dup hits:\t[('ENSG00000063587', 2), ('ENSG00000157429', 2), ('ENSG00000181135', 2), ('ENSG00000188629', 2), ('E\n",
      "WARNING:biothings.client:1 input query terms found no hit:\t['ENSG00000215271']\n",
      "INFO:biothings.client:Pass \"returnall=True\" to return complete lists of duplicate or missing query terms.\n"
     ]
    }
   ],
   "source": [
    "mg = mygene.MyGeneInfo()\n",
    "sensitive_query = mg.querymany(sensitive, species='human')\n",
    "in_sensitive_query = mg.querymany(insensitive, species='human')\n",
    "sensitive_gene_name = [x['symbol'] for x in sensitive_query]\n",
    "in_sensitive_gene_name = [x['symbol'] for x in in_sensitive_query if 'symbol' in x]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "21d8689e",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_sensitive = [GPT_3_5_gene_embeddings[name] for name in sensitive_gene_name\\\n",
    "               if name in GPT_3_5_gene_embeddings]\n",
    "x_insensitive = [GPT_3_5_gene_embeddings[name] for name in in_sensitive_gene_name \\\n",
    "                 if name in GPT_3_5_gene_embeddings]\n",
    "x_dosage = x_sensitive.copy()\n",
    "x_dosage.extend(x_insensitive)\n",
    "y_dosage = np.concatenate((np.repeat(1,len(x_sensitive)),np.repeat(0,len(x_insensitive))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "050c4a0f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Logistic Regression ROC AUC: 0.889 +/- 0.030\n",
      "Random Forest ROC AUC: 0.899 +/- 0.031\n"
     ]
    }
   ],
   "source": [
    "np.random.seed(2023)\n",
    "random.seed(2023)\n",
    "\n",
    "# Ensure your data is in NumPy array format for compatibility\n",
    "X_array = np.array(x_dosage)\n",
    "y_array = np.array(y_dosage)\n",
    "\n",
    "\n",
    "# Set up Stratified K-Folds cross-validator\n",
    "# It provides train/test indices to split data into train/test sets\n",
    "cv = StratifiedKFold(n_splits=5)\n",
    "\n",
    "# Lists to store ROC AUC scores for each fold\n",
    "roc_auc_logistic = []\n",
    "roc_auc_rf = []\n",
    "\n",
    "# Lists to store TPR and FPR for each fold\n",
    "tpr_logistic = []\n",
    "fpr_logistic = []\n",
    "tpr_rf = []\n",
    "fpr_rf = []\n",
    "\n",
    "for train_index, test_index in cv.split(X_array, y_array):\n",
    "    X_train, X_test = X_array[train_index], X_array[test_index]\n",
    "    y_train, y_test = y_array[train_index], y_array[test_index]\n",
    "\n",
    "    # Logistic Regression\n",
    "    logistic_model = LogisticRegression()\n",
    "    logistic_model.fit(X_train, y_train)\n",
    "    y_score_logistic = logistic_model.predict_proba(X_test)[:, 1]\n",
    "    fpr, tpr, _ = roc_curve(y_test, y_score_logistic)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    roc_auc_logistic.append(roc_auc)\n",
    "    tpr_logistic.append(tpr)\n",
    "    fpr_logistic.append(fpr)\n",
    "\n",
    "    # Random Forest\n",
    "    random_forest_model = RandomForestClassifier()\n",
    "    random_forest_model.fit(X_train, y_train)\n",
    "    y_score_rf = random_forest_model.predict_proba(X_test)[:, 1]\n",
    "    fpr, tpr, _ = roc_curve(y_test, y_score_rf)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    roc_auc_rf.append(roc_auc)\n",
    "    tpr_rf.append(tpr)\n",
    "    fpr_rf.append(fpr)\n",
    "\n",
    "# Print ROC AUC scores\n",
    "print(f\"Logistic Regression ROC AUC: {np.mean(roc_auc_logistic):.3f} +/- {np.std(roc_auc_logistic):.3f}\")\n",
    "print(f\"Random Forest ROC AUC: {np.mean(roc_auc_rf):.3f} +/- {np.std(roc_auc_rf):.3f}\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "20487f63",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 455,
       "width": 573
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot ROC curves\n",
    "plt.figure()\n",
    "plt.plot([0, 1], [0, 1], color='navy', lw=2, linestyle='--')\n",
    "plt.plot(fpr_logistic[0], tpr_logistic[0], color='darkorange', lw=2, label='Logistic Regression (AUC = %0.2f)' % np.mean(roc_auc_logistic))\n",
    "plt.plot(fpr_rf[0], tpr_rf[0], color='green', lw=2, label='Random Forest (AUC = %0.2f)' % np.mean(roc_auc_rf))\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('AUC for dosage sensitivity prediction, GPT')\n",
    "plt.legend(loc='lower right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8c81914b",
   "metadata": {},
   "source": [
    "## Methylation state prediction\n",
    "The csv files are downloaded from https://huggingface.co/datasets/ctheodoris/Genecorpus-30M/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "6be8835c",
   "metadata": {},
   "outputs": [],
   "source": [
    "bivalent_gene_labels = pd.read_csv('~/Downloads/bivalent_gene_labels.txt',header=None)\n",
    "no_methylation_gene_labels = pd.read_csv('~/no_methylation_gene_labels.txt',header=None)\n",
    "lysine_gene_labels = pd.read_csv('~/Downloads/lys4_only_gene_labels.txt',header=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "26f24ab1",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:biothings.client:querying 1-107...\n",
      "INFO:biothings.client:done.\n",
      "INFO:biothings.client:Finished.\n",
      "WARNING:biothings.client:10 input query terms found dup hits:\t[('ENSG00000007372', 2), ('ENSG00000110693', 2), ('ENSG00000117707', 2), ('ENSG00000120093', 2), ('E\n",
      "INFO:biothings.client:Pass \"returnall=True\" to return complete lists of duplicate or missing query terms.\n",
      "INFO:biothings.client:querying 1-42...\n",
      "INFO:biothings.client:done.\n",
      "INFO:biothings.client:Finished.\n",
      "WARNING:biothings.client:2 input query terms found dup hits:\t[('ENSG00000147488', 2), ('ENSG00000151322', 2)]\n",
      "INFO:biothings.client:Pass \"returnall=True\" to return complete lists of duplicate or missing query terms.\n",
      "INFO:biothings.client:querying 1-80...\n",
      "INFO:biothings.client:done.\n",
      "INFO:biothings.client:Finished.\n",
      "WARNING:biothings.client:2 input query terms found dup hits:\t[('ENSG00000196628', 2), ('ENSG00000198728', 2)]\n",
      "INFO:biothings.client:Pass \"returnall=True\" to return complete lists of duplicate or missing query terms.\n"
     ]
    }
   ],
   "source": [
    "bivalent_query = mg.querymany(list(bivalent_gene_labels[0]), species='human')\n",
    "no_methylation_query = mg.querymany(list(no_methylation_gene_labels[0]), species='human')\n",
    "lysine_query = mg.querymany(list(lysine_gene_labels[0]), species='human')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "308d2bd6",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_bivalent = [GPT_3_5_gene_embeddings[x['symbol']] for x in bivalent_query\\\n",
    "               if 'symbol' in x and x['symbol'] in GPT_3_5_gene_embeddings]\n",
    "x_no_methylation = [GPT_3_5_gene_embeddings[x['symbol']] for x in no_methylation_query\\\n",
    "               if 'symbol' in x and  x['symbol'] in GPT_3_5_gene_embeddings]\n",
    "x_lysine = [GPT_3_5_gene_embeddings[x['symbol']] for x in lysine_query\\\n",
    "                if 'symbol' in x and  x['symbol'] in GPT_3_5_gene_embeddings]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "adeba376",
   "metadata": {},
   "source": [
    "### Bivalent versus non-methylated"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "e6f83a44",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Logistic Regression ROC AUC: 0.912 +/- 0.062\n",
      "Random Forest ROC AUC: 0.922 +/- 0.059\n"
     ]
    }
   ],
   "source": [
    "# bivalent versus non-methylated\n",
    "np.random.seed(2023)\n",
    "random.seed(2023)\n",
    "\n",
    "X_array = np.concatenate((x_bivalent,x_no_methylation))\n",
    "y_array =  np.concatenate((np.repeat(1,len(x_bivalent)),np.repeat(0,len(x_no_methylation))))\n",
    "\n",
    "# Set up Stratified K-Folds cross-validator\n",
    "# It provides train/test indices to split data into train/test sets\n",
    "cv = StratifiedKFold(n_splits=5)\n",
    "\n",
    "# Lists to store ROC AUC scores for each fold\n",
    "roc_auc_logistic = []\n",
    "roc_auc_rf = []\n",
    "\n",
    "# Lists to store TPR and FPR for each fold\n",
    "tpr_logistic = []\n",
    "fpr_logistic = []\n",
    "tpr_rf = []\n",
    "fpr_rf = []\n",
    "\n",
    "for train_index, test_index in cv.split(X_array, y_array):\n",
    "    X_train, X_test = X_array[train_index], X_array[test_index]\n",
    "    y_train, y_test = y_array[train_index], y_array[test_index]\n",
    "\n",
    "    # Logistic Regression\n",
    "    logistic_model = LogisticRegression()\n",
    "    logistic_model.fit(X_train, y_train)\n",
    "    y_score_logistic = logistic_model.predict_proba(X_test)[:, 1]\n",
    "    fpr, tpr, _ = roc_curve(y_test, y_score_logistic)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    roc_auc_logistic.append(roc_auc)\n",
    "    tpr_logistic.append(tpr)\n",
    "    fpr_logistic.append(fpr)\n",
    "\n",
    "    # Random Forest\n",
    "    random_forest_model = RandomForestClassifier()\n",
    "    random_forest_model.fit(X_train, y_train)\n",
    "    y_score_rf = random_forest_model.predict_proba(X_test)[:, 1]\n",
    "    fpr, tpr, _ = roc_curve(y_test, y_score_rf)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    roc_auc_rf.append(roc_auc)\n",
    "    tpr_rf.append(tpr)\n",
    "    fpr_rf.append(fpr)\n",
    "\n",
    "# Print ROC AUC scores\n",
    "print(f\"Logistic Regression ROC AUC: {np.mean(roc_auc_logistic):.3f} +/- {np.std(roc_auc_logistic):.3f}\")\n",
    "print(f\"Random Forest ROC AUC: {np.mean(roc_auc_rf):.3f} +/- {np.std(roc_auc_rf):.3f}\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cac9e83c",
   "metadata": {},
   "source": [
    "### bivalent versus lys-4 methylated"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "e7cd1241",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Logistic Regression ROC AUC: 0.941 +/- 0.034\n",
      "Random Forest ROC AUC: 0.952 +/- 0.036\n"
     ]
    }
   ],
   "source": [
    "np.random.seed(2023)\n",
    "random.seed(2023)\n",
    "\n",
    "# Assuming x and y are your data\n",
    "# For demonstration, let's create some dummy data.\n",
    "# Ensure your data is in NumPy array format for compatibility\n",
    "X_array = np.concatenate((x_bivalent,x_lysine))\n",
    "y_array =  np.concatenate((np.repeat(1,len(x_bivalent)),np.repeat(0,len(x_lysine))))\n",
    "\n",
    "# Set up Stratified K-Folds cross-validator\n",
    "# It provides train/test indices to split data into train/test sets\n",
    "cv = StratifiedKFold(n_splits=5)\n",
    "\n",
    "# Lists to store ROC AUC scores for each fold\n",
    "roc_auc_logistic = []\n",
    "roc_auc_rf = []\n",
    "roc_auc_xgb = []\n",
    "\n",
    "# Lists to store TPR and FPR for each fold\n",
    "tpr_logistic = []\n",
    "fpr_logistic = []\n",
    "tpr_rf = []\n",
    "fpr_rf = []\n",
    "\n",
    "for train_index, test_index in cv.split(X_array, y_array):\n",
    "    X_train, X_test = X_array[train_index], X_array[test_index]\n",
    "    y_train, y_test = y_array[train_index], y_array[test_index]\n",
    "\n",
    "    # Logistic Regression\n",
    "    logistic_model = LogisticRegression()\n",
    "    logistic_model.fit(X_train, y_train)\n",
    "    y_score_logistic = logistic_model.predict_proba(X_test)[:, 1]\n",
    "    fpr, tpr, _ = roc_curve(y_test, y_score_logistic)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    roc_auc_logistic.append(roc_auc)\n",
    "    tpr_logistic.append(tpr)\n",
    "    fpr_logistic.append(fpr)\n",
    "\n",
    "    # Random Forest\n",
    "    random_forest_model = RandomForestClassifier()\n",
    "    random_forest_model.fit(X_train, y_train)\n",
    "    y_score_rf = random_forest_model.predict_proba(X_test)[:, 1]\n",
    "    fpr, tpr, _ = roc_curve(y_test, y_score_rf)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    roc_auc_rf.append(roc_auc)\n",
    "    tpr_rf.append(tpr)\n",
    "    fpr_rf.append(fpr)\n",
    "\n",
    "\n",
    "# Print ROC AUC scores\n",
    "print(f\"Logistic Regression ROC AUC: {np.mean(roc_auc_logistic):.3f} +/- {np.std(roc_auc_logistic):.3f}\")\n",
    "print(f\"Random Forest ROC AUC: {np.mean(roc_auc_rf):.3f} +/- {np.std(roc_auc_rf):.3f}\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}