COVID-19 Host Genetic Predictor Model.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## COVID-19 Host Genetic Severity Predictor Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Introduction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There is still a lot we are yet to unravel when it comes to the severity manifested by different SARS-CoV-2 patients for example; why are certain patients even though not advanced in age and with no comorbidities susceptible severity to the disease while others are not. Vital information to why the disease differs greatly between people might lie in their DNA (e.g. variations in immune-related genes). Gene expression identifies patterns within human immune cells may also play a key role in determining how the host immune system interacts with the virus. Examining genomes of patients who have a severe response to COVID-19 becomes necessary to further understand these complex interactions that are crucial to shed more light on understanding the biology of the disease, selecting drugs for repurposing — and knowing patients who are most at risk or providing some sorts of protection against the infection.\n",
    "\n",
    "In this study, we made predictions using identified 16 candidate genetic variants that are likely determinants of COVID-19 severity outcomes in patients from a prior study we carried out using the 2000 cohort Whole Exome Sequencing (WES) genetic and clinical datasets (https://doi.org/10.21203/rs.3.rs-1062190/v1, Onoja, et al., 2022). \n",
    "\n",
    "The outline of this analysis is as follows: \n",
    "\n",
    "1) First, we carried out an external validation of the model developed from the prior study of 2000 cohort dataset using a follow-up 3000 cohort dataset. \n",
    "\n",
    "2) Secondly, we further utilized the explainer dashboard python library to carry out an explanation of our model predictions particularly considering the new cohort dataset information. \n",
    "\n",
    "3) Lastly, we implored the Phenome-wide Association technique by leveraging the Bioinformatics tool of OpenTarget web tool (an open-source web browser Bioinformatics tool) to associate our identified genetic variants with disease traits which could lead to plausible clinical trajectories of the COVID-19 disease in patients."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### External Validation of COVID-19 Host Genetic Predictor Model "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Hp\\anaconda3\\lib\\site-packages\\numpy\\_distributor_init.py:32: UserWarning: loaded more than 1 DLL from .libs:\n",
      "C:\\Users\\Hp\\anaconda3\\lib\\site-packages\\numpy\\.libs\\libopenblas.TXA6YQSD3GCQQC22GEQ54J2UDCXDXHWN.gfortran-win_amd64.dll\n",
      "C:\\Users\\Hp\\anaconda3\\lib\\site-packages\\numpy\\.libs\\libopenblas.WCDJNK7YVMPZQ2ME2ZZHJJRJ3JIKNDB7.gfortran-win_amd64.dll\n",
      "  stacklevel=1)\n",
      "\n",
      "The dash_html_components package is deprecated. Please replace\n",
      "`import dash_html_components as html` with `from dash import html`\n",
      "\n",
      "The dash_core_components package is deprecated. Please replace\n",
      "`import dash_core_components as dcc` with `from dash import dcc`\n",
      "\n",
      "The dash_table package is deprecated. Please replace\n",
      "`import dash_table` with `from dash import dash_table`\n",
      "\n",
      "Also, if you're using any of the table format helpers (e.g. Group), replace \n",
      "`from dash_table.Format import Group` with \n",
      "`from dash.dash_table.Format import Group`\n"
     ]
    }
   ],
   "source": [
    "# import numpy as np\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn.preprocessing import StandardScaler, label_binarize\n",
    "import joblib\n",
    "\n",
    "from explainerdashboard import ClassifierExplainer, ExplainerDashboard\n",
    "from sklearn.ensemble import VotingClassifier\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Load Ensemble Model trained from 2000 cohort dataset "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style type='text/css'>\n",
       ".datatable table.frame { margin-bottom: 0; }\n",
       ".datatable table.frame thead { border-bottom: none; }\n",
       ".datatable table.frame tr.coltypes td {  color: #FFFFFF;  line-height: 6px;  padding: 0 0.5em;}\n",
       ".datatable .bool    { background: #DDDD99; }\n",
       ".datatable .object  { background: #565656; }\n",
       ".datatable .int     { background: #5D9E5D; }\n",
       ".datatable .float   { background: #4040CC; }\n",
       ".datatable .str     { background: #CC4040; }\n",
       ".datatable .row_index {  background: var(--jp-border-color3);  border-right: 1px solid var(--jp-border-color0);  color: var(--jp-ui-font-color3);  font-size: 9px;}\n",
       ".datatable .frame tr.coltypes .row_index {  background: var(--jp-border-color0);}\n",
       ".datatable th:nth-child(2) { padding-left: 12px; }\n",
       ".datatable .hellipsis {  color: var(--jp-cell-editor-border-color);}\n",
       ".datatable .vellipsis {  background: var(--jp-layout-color0);  color: var(--jp-cell-editor-border-color);}\n",
       ".datatable .na {  color: var(--jp-cell-editor-border-color);  font-size: 80%;}\n",
       ".datatable .footer { font-size: 9px; }\n",
       ".datatable .frame_dimensions {  background: var(--jp-border-color3);  border-top: 1px solid var(--jp-border-color0);  color: var(--jp-ui-font-color3);  display: inline-block;  opacity: 0.6;  padding: 1px 10px 1px 5px;}\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "path=\"/Users/Hp/Desktop/3000_cohort_analysis/\"\n",
    "with open(path + 'ensemble_model_2000_cohort_16_full_support_with_covariates_1', 'rb') as f1: \n",
    "    ensemble_model = joblib.load(f1)\n",
    "model_ensemble = ensemble_model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Load validation dataset from 3000 cohort dataset "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>age</th>\n",
       "      <th>gender</th>\n",
       "      <th>grouping</th>\n",
       "      <th>BMS1P1;FRMPD2B</th>\n",
       "      <th>CEP131</th>\n",
       "      <th>CNTFR</th>\n",
       "      <th>GFM1</th>\n",
       "      <th>GOLGA6L3</th>\n",
       "      <th>HDGFL2</th>\n",
       "      <th>LOC100996720</th>\n",
       "      <th>MIR933</th>\n",
       "      <th>PCSK5</th>\n",
       "      <th>PLEC</th>\n",
       "      <th>SECISBP2L</th>\n",
       "      <th>SPATA6</th>\n",
       "      <th>TRIM72</th>\n",
       "      <th>ZBED3</th>\n",
       "      <th>ZBTB3</th>\n",
       "      <th>ZRANB3</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sample_ID</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AR-COVID3-USC-AREZZO_hg38</th>\n",
       "      <td>29</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>BS-COV-56_hg38</th>\n",
       "      <td>32</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>BS-COV-83_hg38</th>\n",
       "      <td>34</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>COV10001-3742_hg38</th>\n",
       "      <td>67</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>COV10013-3746_hg38</th>\n",
       "      <td>71</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                           age  gender  grouping  BMS1P1;FRMPD2B  CEP131  \\\n",
       "sample_ID                                                                  \n",
       "AR-COVID3-USC-AREZZO_hg38   29       1         0               0       0   \n",
       "BS-COV-56_hg38              32       0         1               0       0   \n",
       "BS-COV-83_hg38              34       0         1               0       0   \n",
       "COV10001-3742_hg38          67       1         1               0       0   \n",
       "COV10013-3746_hg38          71       0         1               0       0   \n",
       "\n",
       "                           CNTFR  GFM1  GOLGA6L3  HDGFL2  LOC100996720  \\\n",
       "sample_ID                                                                \n",
       "AR-COVID3-USC-AREZZO_hg38      0     0         0       0             2   \n",
       "BS-COV-56_hg38                 0     0         0       0             0   \n",
       "BS-COV-83_hg38                 0     0         0       1             0   \n",
       "COV10001-3742_hg38             0     0         0       0             1   \n",
       "COV10013-3746_hg38             0     0         0       0             2   \n",
       "\n",
       "                           MIR933  PCSK5  PLEC  SECISBP2L  SPATA6  TRIM72  \\\n",
       "sample_ID                                                                   \n",
       "AR-COVID3-USC-AREZZO_hg38       0      0     0          2       0       0   \n",
       "BS-COV-56_hg38                  0      0     0          2       0       0   \n",
       "BS-COV-83_hg38                  0      0     0          2       0       0   \n",
       "COV10001-3742_hg38              0      0     0          2       0       0   \n",
       "COV10013-3746_hg38              1      0     0          1       0       0   \n",
       "\n",
       "                           ZBED3  ZBTB3  ZRANB3  \n",
       "sample_ID                                        \n",
       "AR-COVID3-USC-AREZZO_hg38      0      0       0  \n",
       "BS-COV-56_hg38                 0      0       0  \n",
       "BS-COV-83_hg38                 0      0       0  \n",
       "COV10001-3742_hg38             0      0       0  \n",
       "COV10013-3746_hg38             0      0       0  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# import dataset for validation\n",
    "path=\"/Users/Hp/Desktop/3000_cohort_analysis/\"\n",
    "df_2 = pd.read_csv(path+\"feature_count_16_full_supported_variants_latest_3000_cohort_arranged_updated.csv\", delimiter=',', quotechar='\"', index_col='sample_ID')\n",
    "\n",
    "df_2.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# prepare the dataset\n",
    "\n",
    "# define target variable \n",
    "y_val = df_2['grouping']\n",
    "\n",
    "# define target variable \n",
    "target = df_2['grouping']\n",
    "# features \n",
    "X_val = df_2.drop('grouping', 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Binarize the target variable and scale\n",
    "\n",
    "scaler = StandardScaler()\n",
    "\n",
    "# training set\n",
    "scale = StandardScaler()\n",
    "X_va = pd.DataFrame(scale.fit_transform(X_val))\n",
    "X_va.columns = X_val.columns.values\n",
    "X_va.index = X_val.index.values\n",
    "X_val = X_va\n",
    "\n",
    "y_val = label_binarize(y_val, classes=[0, 1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['age', 'gender', 'BMS1P1;FRMPD2B', 'CEP131', 'CNTFR', 'GFM1',\n",
       "       'GOLGA6L3', 'HDGFL2', 'LOC100996720', 'MIR933', 'PCSK5', 'PLEC',\n",
       "       'SECISBP2L', 'SPATA6', 'TRIM72', 'ZBED3', 'ZBTB3', 'ZRANB3'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_val.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Make Predictions "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy score 0.7799352750809061\n"
     ]
    }
   ],
   "source": [
    "# calculate model accuracy decision tree like models fold_1_5, 2000 cohort \n",
    "from sklearn.metrics import accuracy_score\n",
    "y_pred = ensemble_model.predict(X_val)\n",
    "print('Test Accuracy score %s' % accuracy_score(y_val, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The extenal model validation accuracy score is 77.99% which implies the model we developed on the 2000 cohort dataset was able to recaptulate relevant information that predicts the COVID-19 outcome severity in new patients cohort. However, we will further explore other performance classification metrics such as the f1, recall, precision, and AUC-ROC scores to build more trust on the model. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Logistic: f1=0.845 auc=0.968\n"
     ]
    }
   ],
   "source": [
    "# predict probabilities\n",
    "model = ensemble_model\n",
    "from sklearn.metrics import accuracy_score, f1_score, auc, roc_auc_score, roc_curve, precision_recall_curve, matthews_corrcoef\n",
    "es_probs = model.predict_proba(X_val)\n",
    "# keep probabilities for the positive outcome only\n",
    "es_probs = es_probs[:, 1]\n",
    "\n",
    "# predict class values\n",
    "y_pred = model.predict(X_val)\n",
    "# calculate precision and recall for each threshold\n",
    "es_precision, es_recall, _ = precision_recall_curve(y_val, es_probs)\n",
    "# calculate scores\n",
    "es_f1, es_auc = f1_score(y_val, y_pred), auc(es_recall, es_precision)\n",
    "# summarize scores\n",
    "print('Logistic: f1=%.3f auc=%.3f' % (es_f1, es_auc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The f1 score is 84.5% and the Precision-Recall AUC score is 96.8%. This shows the model's accuracy is not been influenced by some sorts of random chance. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Visualize Precision-Recall ROC-Curve "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the precision-recall curves\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib import pyplot\n",
    "\n",
    "fig = plt.figure(figsize=(8, 6))\n",
    "ax = fig.add_subplot(111)\n",
    "no_severe = len(y_val[y_val==1]) / len(y_val)\n",
    "pyplot.plot([0, 1], [no_severe, no_severe], linestyle='--')\n",
    "pyplot.plot(es_recall, es_precision, marker='.', label='Voting Classifier (AP = 0.97)')\n",
    "\n",
    "#fig.suptitle('VotingClassifier Classification Report', fontsize=20)\n",
    "#[t.set_color('black') for t in ax.xaxis.get_ticklines()]\n",
    "#[t.set_color('black') for t in ax.xaxis.get_ticklabels()]\n",
    "matplotlib.rcParams.update({'font.size': 18})\n",
    "matplotlib.rc('xtick', labelsize=18) \n",
    "matplotlib.rc('ytick', labelsize=18) \n",
    "# axis labels\n",
    "pyplot.xlabel('Recall (Positive label: 1)')\n",
    "pyplot.ylabel('Precision (Positive label: 1)')\n",
    "# show the legend\n",
    "pyplot.legend()\n",
    "plt.legend(loc='lower left')\n",
    "#plt.savefig(path+'PR_ROC_3000_Cohort_decision_tree_like_models.pdf', bbox_inches='tight')\n",
    "#plt.savefig(path+'PR_ROC_3000_Cohort_decision_tree_like_models.jpg', bbox_inches='tight')\n",
    "# show the plot\n",
    "pyplot.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Visualizing the model's precision-recall curve help us to understand how well the model classified the patients COVID-19 severity outcome. However, it does it differently from the way an ROC curve does. Precision-recall curve plots true positive rate (recall or sensitivity) against the positive predictive value (precision). "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Visualize ROC Curve "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Asymptomatic: ROC AUC=0.500\n",
      "Voting Classifier: ROC AUC=0.864\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "### ROC Curve \n",
    "from sklearn.metrics import roc_auc_score, roc_curve\n",
    "# generate a no ensemble voting classifier prediction (majority class)\n",
    "ns_probs = [0 for _ in range(len(y_val))]\n",
    "# calculate scores\n",
    "ns_auc = roc_auc_score(y_val, ns_probs)\n",
    "es_auc = roc_auc_score(y_val, es_probs)\n",
    "# summarize scores\n",
    "print('Asymptomatic: ROC AUC=%.3f' % (ns_auc))\n",
    "print('Voting Classifier: ROC AUC=%.3f' % (es_auc))\n",
    "# calculate roc curves\n",
    "ns_fpr, ns_tpr, _ = roc_curve(y_val, ns_probs)\n",
    "es_fpr, es_tpr, _ = roc_curve(y_val, es_probs)\n",
    "# plot the roc curve for the model\n",
    "fig = plt.figure(figsize=(8, 6))\n",
    "pyplot.plot(ns_fpr, ns_tpr, linestyle='--')\n",
    "pyplot.plot(es_fpr, es_tpr, marker='.', label='Voting Classifier AUC = %0.2f' % es_auc)\n",
    "# axis labels\n",
    "pyplot.xlabel('False Positive Rate')\n",
    "pyplot.ylabel('True Positive Rate')\n",
    "# show the legend\n",
    "pyplot.legend()\n",
    "#plt.legend(loc='lower left')\n",
    "#plt.savefig(path+'AUC_ROC_3000_Cohort_decision_tree_like_models.pdf', bbox_inches='tight')\n",
    "#plt.savefig(path+'AUC_ROC_3000_Cohort_decision_tree_like_models.jpg', bbox_inches='tight')\n",
    "# show the plot\n",
    "pyplot.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The receiver operating characteristic (ROC) curve displayed above was used to know how well the model predicts the COVID-19 severity outcome in patients from the external dataset. The ROC curve is generated by plotting the false positive rate of a model against its true positive rate, for each possible cutoff value."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Visualize the Confusion Matrix "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix, plot_confusion_matrix\n",
    "\n",
    "class_name = ['Asymptomatic', 'Severe']\n",
    "fig = plt.figure(figsize=(8, 6))\n",
    "\n",
    "confusion_matrix(y_val, y_pred)\n",
    "conf_matrix = confusion_matrix(y_val, y_pred)\n",
    "\n",
    "ax= plt.subplot()\n",
    "sns.heatmap(conf_matrix, annot=True, fmt='g', cmap=plt.cm.bwr, ax=ax);  #annot=True to annotate cells, ftm='g' to disable scientific notation\n",
    "\n",
    "# labels, title and ticks\n",
    "ax.set_xlabel('Predicted labels');ax.set_ylabel('True labels'); \n",
    "ax.set_title('Confusion Matrix'); \n",
    "ax.xaxis.set_ticklabels(['Asymptomatic', 'Severe']); ax.yaxis.set_ticklabels(['Asymptomatic', 'Severe']);\n",
    "\n",
    "#plt.savefig(path+'confusion_matrix_3000_Cohort_decision_tree_like_models.pdf', bbox_inches='tight')\n",
    "#plt.savefig(path+'confusion_matrix_3000_Cohort_decision_tree_like_models.jpg', bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The confusion matrix heatmap showed the model was able to accurately predicts patients severity outcome considering the off-diagonal elements of the matrix. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Classifier Metrics Evaluation\n",
    "\n",
    "Classification models attempt to predict a target in a discrete space, that is assign an instance of dependent variables one or more categories. Classification score visualizers display the differences between classes as well as a number of classifier-specific visual evaluations.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from yellowbrick.classifier import ClassificationReport, ROCAUC, ClassBalance\n",
    "import matplotlib \n",
    "plt.figure(figsize=(12, 8))\n",
    "#fig = plt.figure()\n",
    "visualizer = ClassificationReport(model, classes=class_name)\n",
    "matplotlib.rc('xtick', labelsize=24) \n",
    "matplotlib.rc('ytick', labelsize=24) \n",
    "#fig.suptitle('VotingClassifier Classification Report', fontsize=20)\n",
    "matplotlib.rcParams.update({'font.size': 24})\n",
    "#visualizer.fit(X_train, y_train)  # Fit the training data to the visualizer\n",
    "visualizer.score(X_val, y_val)  # Evaluate the model on the test data\n",
    "g = visualizer.poof()  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### COVID-19 Host Severity Predictor Model Explanation\n",
    "\n",
    "We utilized the explainerdashboard python package. This approach makes it easy to quickly build an interactive dashboard that explains the inner workings of our developed ensemble COVID-19 host severity predictor model. This allows us to open up the 'black box' to further shed more light to our understanding of how exactly our developed ensemble COVID-19 host severity predictor model generates its predictions on the 3000 cohort dataset."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Define Feature Description Integrated with Phenome-wide Associated disease traits "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_descriptions = {\"age\" : \"Age (Years)\", \"gender\" : \"Gender\",\n",
    "                        \"ZBED3\" : \"ZBED3(rs531117283)[Hint: Genetic disease associated - BMI-adjusted waist-hip ratio, type 2 diabetes mellitus]\", \n",
    "                        \"PLEC\" : \"PLEC(rs140300753) [Hint: Genetic disease associated - Abnormalities of breathing, Epidermolysis bullosa simplex with muscular dystrophy, Epidermolysis bullosa simplex with pyloric atresia]\",\n",
    "                        \"TRIM72\" : \"TRIM72() [Hint: Genetic disease associated - breast adenocarcinoma, diabetes mellitus, coronary artery disease]\", \n",
    "                        \"HDGFL2\" : \"HDGFL2(rs146793578) [Hint: Genetic disease associated - mean corpuscular hemoglobin concentration, BMI-adjusted waist-hip ratio, smoking status measurement]\", \n",
    "                        \"SECISBP2L\" : \"SECISBP2L(rs75595801) [Hint: Genetic disease associated - calcium measurement, diastolic blood pressure, lung adenocarcinoma, alcohol consumption measurement]\",\n",
    "                        \"CEP131\" : \"CEP131(rs2659015)[Hint: Genetic disease associated - serum IgG glycosylation measurement, brain glioblastoma, IgG fucosylation measurement, vitamin D measurement]\",\n",
    "                        \"GOLGA6L3\" : \"GOLGA6L3(rs367838829) [Hint:Genetic disease - BMI-adjusted waist and hip circumference]\",\n",
    "                        \"PCSK5\" : \"PCSK5(rs72745135) [Hint: Genetic disease associated - Abnormalities of breathing, Currarino Syndrome and Avian Influenza.]\",\n",
    "                        \"GFM1\" : \"GFM1(rs370496368) [Hint:Genetic disease associated - Hepatoencephalopathy due to combined oxidative phosphorylation deficiency type 1, combined oxidative phosphorylation deficiency, systolic blood pressure]\", \n",
    "                        \"ZBTB3\" : \"ZBTB3(rs544641)[Hint: Genetic disease associated - alcohol consumption measurement, wellbeing measurement, chronotype measurement]\",\n",
    "                        \"BMS1P1;FRMPD2B\" : \"BMS1P1;FRMPD2B()\", \"SPATA6\" : \"SPATA6(rs77303590) [Hint: Genetic disease associated - urinary metabolite measurement, systolic blood pressure]\",\n",
    "                        \"CNTFR\" : \"CNTFR() [Hint: Genetic disease associated - blood protein measurement, heel bone mineral density, adolescent idiopathic scoliosis]\", \"MIR933\" : \"MIR933(rs79402775)\", \n",
    "                        \"ZRANB3\" : \"ZRANB3(rs1465146591)[Hint: Genetic disease associated - body height, gut microbiome measurement, body mass index]\",\n",
    "                        \"LOC100996720\" : \"LOC100996720()\"\n",
    "                       }\n",
    "                        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING: Parameter shap='guess', but failed to guess the type of shap explainer to use for VotingClassifier. Defaulting to the model agnostic shap.KernelExplainer (shap='kernel'). However this will be slow, so if your model is compatible with e.g. shap.TreeExplainer or shap.LinearExplainer then pass shap='tree' or shap='linear'!\n",
      "WARNING: For shap='kernel', shap interaction values can unfortunately not be calculated!\n",
      "Note: shap values for shap='kernel' normally get calculated against X_background, but paramater X_background=None, so setting X_background=shap.sample(X, 50)...\n",
      "Generating self.shap_explainer = shap.KernelExplainer(model, X, link='identity')\n"
     ]
    }
   ],
   "source": [
    "### Explainer model \n",
    "explainer = ClassifierExplainer(ensemble_model, X_val, y_val.ravel(), \n",
    "                                descriptions=feature_descriptions, # adds a table and hover labels to dashboard\n",
    "                                labels=['Asymptomatic', 'Severe'], # defaults to ['0', '1', etc]\n",
    "                                idxs = X_val.index, # defaults to X.index\n",
    "                                index_name = \"sample_ID\", # defaults to X.index.name\n",
    "                                target = \"grouping\", # defaults to y.name\n",
    "                                )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Building ExplainerDashboard..\n",
      "For this type of model and model_output interactions don't work, so setting shap_interaction=False...\n",
      "The explainer object has no decision_trees property. so setting decision_trees=False...\n",
      "Generating layout...\n",
      "Calculating shap values...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5481b19c220a4781b8f35764541f5381",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=618.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Calculating prediction probabilities...\n",
      "Calculating metrics...\n",
      "Calculating confusion matrices...\n",
      "Calculating classification_dfs...\n",
      "Calculating roc auc curves...\n",
      "Calculating pr auc curves...\n",
      "Calculating liftcurve_dfs...\n",
      "Calculating dependencies...\n",
      "Calculating permutation importances (if slow, try setting n_jobs parameter)...\n",
      "Calculating predictions...\n",
      "Calculating pred_percentiles...\n",
      "Reminder: you can store the explainer (including calculated dependencies) with explainer.dump('explainer.joblib') and reload with e.g. ClassifierExplainer.from_file('explainer.joblib')\n",
      "Registering callbacks...\n",
      "Starting ExplainerDashboard inline (terminate it with ExplainerDashboard.terminate(8050))\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"1000\"\n",
       "            height=\"800\"\n",
       "            src=\"http://127.0.0.1:8050/\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x1b5b7e25208>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#db = ExplainerDashboard(explainer, \n",
    " # title=\"An Explainable Model of Host Genetic interactions linked to COVID-19 severity in patients\", # defaults to \"Model Explainer\"\n",
    " # whatif=False, mode='inline')\n",
    "#db.run(port=8050)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "from explainerdashboard import InlineExplainer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"1000\"\n",
       "            height=\"800\"\n",
       "            src=\"http://127.0.0.1:8050/\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x1b5b79db9c8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize the model SHAP Feature importance \n",
    "db = InlineExplainer(explainer)\n",
    "db.importances()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"1000\"\n",
       "            height=\"800\"\n",
       "            src=\"http://127.0.0.1:8050/\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x1b5b875ee08>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize explainer model stats\n",
    "db.model_stats()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"1000\"\n",
       "            height=\"800\"\n",
       "            src=\"http://127.0.0.1:8050/\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x1b5b86555c8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Check the explainer SHAP permutated features relationships\n",
    "db.shap.overview()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"1000\"\n",
       "            height=\"800\"\n",
       "            src=\"http://127.0.0.1:8050/\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x1b5b93d4bc8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "db.shap.contributions_table()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get and convert to dataframe SHAP Model Explainer Values \n",
    "df_shap = explainer.get_shap_values_df()\n",
    "#save as CSV file \n",
    "df_shap.to_csv(path+\"SHAP_Explainerdashboard_importance_values.csv\", index = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save model predicted probabilities  \n",
    "df_2 = pd.read_csv(path+\"feature_count_16_full_supported_variants_latest_3000_cohort_arranged_updated.csv\")\n",
    "df_new  = df_2[[\"sample_ID\",\"grouping\"]]\n",
    "pred_probs = model.predict_proba(X_val)\n",
    "pred_probs\n",
    "df_1 = pd.DataFrame(pred_probs, columns=['pred_proba_Asymptomatic', 'pred_proba_Severe'])\n",
    "#concatenate dataframes\n",
    "df_3 = pd.concat([df_new, df_1], axis=1)\n",
    "#df_3.sort_values(by='predicted_proba_RF', ascending=False).head(10)\n",
    "#save as CSV file \n",
    "df_3.to_csv(path + \"predicted_probabilities_3000_cohort_dataset.csv\", index = True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}