06_correlation_analysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Univariate correlation analysis\n",
    "\n",
    "We wanted to look at whether correlations between gene expression and mutation status are primarily driven by a strong correlation in a single cancer type, or by weak correlations across all cancer types. To make the analysis simpler, we'll just look at univariate correlations between the expression of a single gene and mutation status in a given driver.\n",
    "\n",
    "More specifically, we want to identify genes that have a strong univariate pan-cancer correlation, and classify them (roughly) as one of the following:\n",
    "\n",
    "* Driven mostly by a single cancer type (one cancer type with a large correlation f-statistic, others fairly small)\n",
    "* Driven mostly by 2+ cancer types (two or more cancer types with large correlations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys\n",
    "from pathlib import Path\n",
    "import warnings\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "import pancancer_evaluation.config as cfg\n",
    "import pancancer_evaluation.utilities.data_utilities as du\n",
    "import pancancer_evaluation.utilities.tcga_utilities as tu\n",
    "\n",
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "gene = 'TP53'\n",
    "mad_threshold = 100"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load expression data and mutation label data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Loading gene label data...\n",
      "Loading sample info...\n",
      "Loading pan-cancer data from cached pickle file...\n",
      "Loading gene expression data...\n",
      "Standardizing columns of expression data...\n"
     ]
    }
   ],
   "source": [
    "print('Loading gene label data...', file=sys.stderr)\n",
    "genes_df = du.load_top_50()\n",
    "sample_info_df = du.load_sample_info(verbose=True)\n",
    "\n",
    "# this returns a tuple of dataframes, unpack it below\n",
    "pancancer_data = du.load_pancancer_data(verbose=True)\n",
    "(sample_freeze_df,\n",
    " mutation_df,\n",
    " copy_loss_df,\n",
    " copy_gain_df,\n",
    " mut_burden_df) = pancancer_data\n",
    "\n",
    "rnaseq_df = du.load_expression_data(verbose=True)\n",
    "\n",
    "# standardize columns of expression dataframe\n",
    "print('Standardizing columns of expression data...', file=sys.stderr)\n",
    "rnaseq_df[rnaseq_df.columns] = StandardScaler().fit_transform(rnaseq_df[rnaseq_df.columns])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(11060, 16148)\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>1</th>\n",
       "      <th>10</th>\n",
       "      <th>100</th>\n",
       "      <th>1000</th>\n",
       "      <th>10000</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",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>TCGA-02-0047-01</th>\n",
       "      <td>-0.144100</td>\n",
       "      <td>-0.136450</td>\n",
       "      <td>-0.207065</td>\n",
       "      <td>1.049402</td>\n",
       "      <td>0.644625</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-0055-01</th>\n",
       "      <td>-0.124925</td>\n",
       "      <td>-0.197893</td>\n",
       "      <td>-0.132694</td>\n",
       "      <td>0.704438</td>\n",
       "      <td>0.154763</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-2483-01</th>\n",
       "      <td>-0.133543</td>\n",
       "      <td>-0.174587</td>\n",
       "      <td>-0.103291</td>\n",
       "      <td>1.473420</td>\n",
       "      <td>0.669303</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-2485-01</th>\n",
       "      <td>-0.147052</td>\n",
       "      <td>-0.072888</td>\n",
       "      <td>-0.213119</td>\n",
       "      <td>4.405612</td>\n",
       "      <td>11.503035</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-2486-01</th>\n",
       "      <td>-0.145321</td>\n",
       "      <td>-0.181076</td>\n",
       "      <td>-0.147395</td>\n",
       "      <td>1.013468</td>\n",
       "      <td>0.117745</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        1        10       100      1000      10000\n",
       "sample_id                                                         \n",
       "TCGA-02-0047-01 -0.144100 -0.136450 -0.207065  1.049402   0.644625\n",
       "TCGA-02-0055-01 -0.124925 -0.197893 -0.132694  0.704438   0.154763\n",
       "TCGA-02-2483-01 -0.133543 -0.174587 -0.103291  1.473420   0.669303\n",
       "TCGA-02-2485-01 -0.147052 -0.072888 -0.213119  4.405612  11.503035\n",
       "TCGA-02-2486-01 -0.145321 -0.181076 -0.147395  1.013468   0.117745"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(rnaseq_df.shape)\n",
    "rnaseq_df.iloc[:5, :5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>PATIENT_BARCODE</th>\n",
       "      <th>SAMPLE_BARCODE</th>\n",
       "      <th>DISEASE</th>\n",
       "      <th>SUBTYPE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>TCGA-OR-A5J1</td>\n",
       "      <td>TCGA-OR-A5J1-01</td>\n",
       "      <td>ACC</td>\n",
       "      <td>Not_Applicable</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>TCGA-OR-A5J2</td>\n",
       "      <td>TCGA-OR-A5J2-01</td>\n",
       "      <td>ACC</td>\n",
       "      <td>Not_Applicable</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>TCGA-OR-A5J3</td>\n",
       "      <td>TCGA-OR-A5J3-01</td>\n",
       "      <td>ACC</td>\n",
       "      <td>Not_Applicable</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>TCGA-OR-A5J5</td>\n",
       "      <td>TCGA-OR-A5J5-01</td>\n",
       "      <td>ACC</td>\n",
       "      <td>Not_Applicable</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>TCGA-OR-A5J6</td>\n",
       "      <td>TCGA-OR-A5J6-01</td>\n",
       "      <td>ACC</td>\n",
       "      <td>Not_Applicable</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  PATIENT_BARCODE   SAMPLE_BARCODE DISEASE         SUBTYPE\n",
       "0    TCGA-OR-A5J1  TCGA-OR-A5J1-01     ACC  Not_Applicable\n",
       "1    TCGA-OR-A5J2  TCGA-OR-A5J2-01     ACC  Not_Applicable\n",
       "2    TCGA-OR-A5J3  TCGA-OR-A5J3-01     ACC  Not_Applicable\n",
       "3    TCGA-OR-A5J5  TCGA-OR-A5J5-01     ACC  Not_Applicable\n",
       "4    TCGA-OR-A5J6  TCGA-OR-A5J6-01     ACC  Not_Applicable"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_freeze_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>5S_rRNA</th>\n",
       "      <th>A1BG</th>\n",
       "      <th>A1CF</th>\n",
       "      <th>A2M</th>\n",
       "      <th>A2ML1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SAMPLE_BARCODE</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>TCGA-02-0047-01</th>\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>TCGA-02-0055-01</th>\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>TCGA-02-2483-01</th>\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>TCGA-02-2485-01</th>\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>TCGA-02-2486-01</th>\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": [
       "                 5S_rRNA  A1BG  A1CF  A2M  A2ML1\n",
       "SAMPLE_BARCODE                                  \n",
       "TCGA-02-0047-01        0     0     0    0      0\n",
       "TCGA-02-0055-01        0     0     0    0      0\n",
       "TCGA-02-2483-01        0     0     0    0      0\n",
       "TCGA-02-2485-01        0     0     0    0      0\n",
       "TCGA-02-2486-01        0     0     0    0      0"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mutation_df.iloc[:5, :5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9074, 3)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>status</th>\n",
       "      <th>cancer_type</th>\n",
       "      <th>subtype</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SAMPLE_BARCODE</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>TCGA-02-0047-01</th>\n",
       "      <td>0</td>\n",
       "      <td>GBM</td>\n",
       "      <td>IDHwt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-0055-01</th>\n",
       "      <td>1</td>\n",
       "      <td>GBM</td>\n",
       "      <td>IDHwt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-2483-01</th>\n",
       "      <td>1</td>\n",
       "      <td>GBM</td>\n",
       "      <td>IDHmut-non-codel</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-2485-01</th>\n",
       "      <td>1</td>\n",
       "      <td>GBM</td>\n",
       "      <td>IDHwt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-2486-01</th>\n",
       "      <td>0</td>\n",
       "      <td>GBM</td>\n",
       "      <td>IDHwt</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 status cancer_type           subtype\n",
       "SAMPLE_BARCODE                                       \n",
       "TCGA-02-0047-01       0         GBM             IDHwt\n",
       "TCGA-02-0055-01       1         GBM             IDHwt\n",
       "TCGA-02-2483-01       1         GBM  IDHmut-non-codel\n",
       "TCGA-02-2485-01       1         GBM             IDHwt\n",
       "TCGA-02-2486-01       0         GBM             IDHwt"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_df = (mutation_df\n",
    "    .loc[:, [gene]]\n",
    "    .merge(sample_freeze_df, left_index=True, right_on='SAMPLE_BARCODE')\n",
    "    .drop(columns='PATIENT_BARCODE')\n",
    "    .set_index('SAMPLE_BARCODE')\n",
    "    .rename(columns={gene: 'status',\n",
    "                     'DISEASE': 'cancer_type',\n",
    "                     'SUBTYPE': 'subtype'})\n",
    ")\n",
    "display(y_df.shape, y_df.head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9074, 16148)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>1</th>\n",
       "      <th>10</th>\n",
       "      <th>100</th>\n",
       "      <th>1000</th>\n",
       "      <th>10000</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SAMPLE_BARCODE</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>TCGA-02-0047-01</th>\n",
       "      <td>-0.144100</td>\n",
       "      <td>-0.136450</td>\n",
       "      <td>-0.207065</td>\n",
       "      <td>1.049402</td>\n",
       "      <td>0.644625</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-0055-01</th>\n",
       "      <td>-0.124925</td>\n",
       "      <td>-0.197893</td>\n",
       "      <td>-0.132694</td>\n",
       "      <td>0.704438</td>\n",
       "      <td>0.154763</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-2483-01</th>\n",
       "      <td>-0.133543</td>\n",
       "      <td>-0.174587</td>\n",
       "      <td>-0.103291</td>\n",
       "      <td>1.473420</td>\n",
       "      <td>0.669303</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-2485-01</th>\n",
       "      <td>-0.147052</td>\n",
       "      <td>-0.072888</td>\n",
       "      <td>-0.213119</td>\n",
       "      <td>4.405612</td>\n",
       "      <td>11.503035</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-2486-01</th>\n",
       "      <td>-0.145321</td>\n",
       "      <td>-0.181076</td>\n",
       "      <td>-0.147395</td>\n",
       "      <td>1.013468</td>\n",
       "      <td>0.117745</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        1        10       100      1000      10000\n",
       "SAMPLE_BARCODE                                                    \n",
       "TCGA-02-0047-01 -0.144100 -0.136450 -0.207065  1.049402   0.644625\n",
       "TCGA-02-0055-01 -0.124925 -0.197893 -0.132694  0.704438   0.154763\n",
       "TCGA-02-2483-01 -0.133543 -0.174587 -0.103291  1.473420   0.669303\n",
       "TCGA-02-2485-01 -0.147052 -0.072888 -0.213119  4.405612  11.503035\n",
       "TCGA-02-2486-01 -0.145321 -0.181076 -0.147395  1.013468   0.117745"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "X_df = rnaseq_df.reindex(y_df.index)\n",
    "\n",
    "# make sure we didn't introduce any NA rows\n",
    "assert X_df.isna().sum().sum() == 0\n",
    "\n",
    "display(X_df.shape,\n",
    "        X_df.isna().sum().sum(),\n",
    "        X_df.iloc[:5, :5])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Subset genes by mean absolute deviation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>index</th>\n",
       "      <th>0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>126695</td>\n",
       "      <td>0.821090</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>147798</td>\n",
       "      <td>0.808853</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10053</td>\n",
       "      <td>0.808764</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>54845</td>\n",
       "      <td>0.803626</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>582</td>\n",
       "      <td>0.802497</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    index         0\n",
       "0  126695  0.821090\n",
       "1  147798  0.808853\n",
       "2   10053  0.808764\n",
       "3   54845  0.803626\n",
       "4     582  0.802497"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# first subset by MAD\n",
    "mad_genes_df = (rnaseq_df\n",
    "    .mad(axis=0)\n",
    "    .sort_values(ascending=False)\n",
    "    .reset_index()\n",
    ")\n",
    "mad_genes_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['126695' '147798' '10053' '54845' '582']\n"
     ]
    }
   ],
   "source": [
    "mad_genes_df.columns=['gene_id', 'mad']\n",
    "mad_genes = (mad_genes_df\n",
    "  .head(mad_threshold)\n",
    "  .gene_id\n",
    "  .astype(str)\n",
    "  .values\n",
    ")\n",
    "\n",
    "print(mad_genes[:5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9074, 100)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>126695</th>\n",
       "      <th>147798</th>\n",
       "      <th>10053</th>\n",
       "      <th>54845</th>\n",
       "      <th>582</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SAMPLE_BARCODE</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>TCGA-02-0047-01</th>\n",
       "      <td>-1.307892</td>\n",
       "      <td>-0.977865</td>\n",
       "      <td>-1.119059</td>\n",
       "      <td>-1.116767</td>\n",
       "      <td>0.458711</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-0055-01</th>\n",
       "      <td>-1.305820</td>\n",
       "      <td>-0.964695</td>\n",
       "      <td>-1.111900</td>\n",
       "      <td>-1.114346</td>\n",
       "      <td>-0.551517</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-2483-01</th>\n",
       "      <td>-1.307892</td>\n",
       "      <td>-0.984642</td>\n",
       "      <td>-1.119059</td>\n",
       "      <td>-1.116851</td>\n",
       "      <td>-0.470321</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-2485-01</th>\n",
       "      <td>-1.306134</td>\n",
       "      <td>-0.984678</td>\n",
       "      <td>-1.118344</td>\n",
       "      <td>-1.113535</td>\n",
       "      <td>1.100724</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCGA-02-2486-01</th>\n",
       "      <td>-1.305639</td>\n",
       "      <td>-0.983574</td>\n",
       "      <td>-1.119059</td>\n",
       "      <td>-1.117114</td>\n",
       "      <td>1.006310</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   126695    147798     10053     54845       582\n",
       "SAMPLE_BARCODE                                                   \n",
       "TCGA-02-0047-01 -1.307892 -0.977865 -1.119059 -1.116767  0.458711\n",
       "TCGA-02-0055-01 -1.305820 -0.964695 -1.111900 -1.114346 -0.551517\n",
       "TCGA-02-2483-01 -1.307892 -0.984642 -1.119059 -1.116851 -0.470321\n",
       "TCGA-02-2485-01 -1.306134 -0.984678 -1.118344 -1.113535  1.100724\n",
       "TCGA-02-2486-01 -1.305639 -0.983574 -1.119059 -1.117114  1.006310"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "X_df = X_df.reindex(mad_genes, axis='columns')\n",
    "display(X_df.shape, X_df.iloc[:5, :5])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Calculate univariate feature correlations with mutation labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 276.60698975,   70.28722435,   34.75394368,  753.51928171,\n",
       "        744.36087967,  446.19287169, 1170.54889718,  168.5785283 ,\n",
       "          5.27937846,  150.45691553])"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.feature_selection import f_classif, mutual_info_classif\n",
    "\n",
    "f_stats_pancan = f_classif(X_df, y_df.status)[0]\n",
    "display(f_stats_pancan[:10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(100, 33)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>GBM</th>\n",
       "      <th>OV</th>\n",
       "      <th>LUAD</th>\n",
       "      <th>LUSC</th>\n",
       "      <th>PRAD</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>9341</th>\n",
       "      <td>2.723151</td>\n",
       "      <td>1.077021</td>\n",
       "      <td>0.078672</td>\n",
       "      <td>0.037524</td>\n",
       "      <td>0.089605</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>65982</th>\n",
       "      <td>0.047316</td>\n",
       "      <td>0.047248</td>\n",
       "      <td>19.764914</td>\n",
       "      <td>10.070783</td>\n",
       "      <td>5.333496</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1656</th>\n",
       "      <td>10.898580</td>\n",
       "      <td>0.304855</td>\n",
       "      <td>0.053062</td>\n",
       "      <td>2.943339</td>\n",
       "      <td>2.253619</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57573</th>\n",
       "      <td>0.073949</td>\n",
       "      <td>3.367391</td>\n",
       "      <td>9.698130</td>\n",
       "      <td>5.831784</td>\n",
       "      <td>5.983377</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51652</th>\n",
       "      <td>0.508308</td>\n",
       "      <td>1.382821</td>\n",
       "      <td>23.220050</td>\n",
       "      <td>1.692238</td>\n",
       "      <td>0.182759</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             GBM        OV       LUAD       LUSC      PRAD\n",
       "9341    2.723151  1.077021   0.078672   0.037524  0.089605\n",
       "65982   0.047316  0.047248  19.764914  10.070783  5.333496\n",
       "1656   10.898580  0.304855   0.053062   2.943339  2.253619\n",
       "57573   0.073949  3.367391   9.698130   5.831784  5.983377\n",
       "51652   0.508308  1.382821  23.220050   1.692238  0.182759"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def get_f_stats_for_cancer_types(X_df, y_df):\n",
    "    f_stats_df = {}\n",
    "    for cancer_type in y_df.cancer_type.unique():\n",
    "        ct_samples = y_df[y_df.cancer_type == cancer_type].index\n",
    "        X_ct_df = X_df.reindex(ct_samples)\n",
    "        y_ct_df = y_df.reindex(ct_samples)\n",
    "        \n",
    "        with warnings.catch_warnings():\n",
    "            warnings.filterwarnings('error')\n",
    "            try:\n",
    "                f_stats_df[cancer_type] = f_classif(X_ct_df, y_ct_df.status)[0]\n",
    "            except RuntimeWarning:\n",
    "                # this can happen if there are no mutated samples in the cancer type\n",
    "                # in that case, just skip it\n",
    "                continue\n",
    "        \n",
    "    return pd.DataFrame(f_stats_df, index=X_df.columns)\n",
    "\n",
    "f_stats_df = get_f_stats_for_cancer_types(X_df, y_df)\n",
    "f_stats_df['pancan'] = f_stats_pancan\n",
    "\n",
    "# make sure we didn't introduce any NA rows\n",
    "assert f_stats_df.isna().sum().sum() == 0\n",
    "\n",
    "display(f_stats_df.shape,\n",
    "        f_stats_df.isna().sum().sum(),\n",
    "        f_stats_df.iloc[-5:, :5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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>max - min</th>\n",
       "      <th>pancan</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>24137</th>\n",
       "      <td>290.781887</td>\n",
       "      <td>1215.285958</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23397</th>\n",
       "      <td>302.217636</td>\n",
       "      <td>1170.548897</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5347</th>\n",
       "      <td>351.053548</td>\n",
       "      <td>1124.041606</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>699</th>\n",
       "      <td>277.920176</td>\n",
       "      <td>1061.147200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9493</th>\n",
       "      <td>225.278353</td>\n",
       "      <td>923.233478</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29127</th>\n",
       "      <td>128.586952</td>\n",
       "      <td>862.842986</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55325</th>\n",
       "      <td>100.424225</td>\n",
       "      <td>823.754151</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79172</th>\n",
       "      <td>279.389949</td>\n",
       "      <td>821.075427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>146909</th>\n",
       "      <td>183.466143</td>\n",
       "      <td>807.072450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57659</th>\n",
       "      <td>299.319467</td>\n",
       "      <td>789.777604</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>83464</th>\n",
       "      <td>248.412445</td>\n",
       "      <td>764.606987</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54845</th>\n",
       "      <td>127.231178</td>\n",
       "      <td>753.519282</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>891</th>\n",
       "      <td>186.226963</td>\n",
       "      <td>749.973212</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79891</th>\n",
       "      <td>94.776546</td>\n",
       "      <td>744.611032</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>582</th>\n",
       "      <td>188.980197</td>\n",
       "      <td>744.360880</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         max - min       pancan\n",
       "24137   290.781887  1215.285958\n",
       "23397   302.217636  1170.548897\n",
       "5347    351.053548  1124.041606\n",
       "699     277.920176  1061.147200\n",
       "9493    225.278353   923.233478\n",
       "29127   128.586952   862.842986\n",
       "55325   100.424225   823.754151\n",
       "79172   279.389949   821.075427\n",
       "146909  183.466143   807.072450\n",
       "57659   299.319467   789.777604\n",
       "83464   248.412445   764.606987\n",
       "54845   127.231178   753.519282\n",
       "891     186.226963   749.973212\n",
       "79891    94.776546   744.611032\n",
       "582     188.980197   744.360880"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# get difference between max and min single-cancer f-statistic\n",
    "max_df = f_stats_df.drop(columns='pancan').max(axis='columns')\n",
    "min_df = f_stats_df.drop(columns='pancan').min(axis='columns')\n",
    "min_max_df = (\n",
    "    pd.DataFrame(max_df - min_df,\n",
    "                 columns=['max - min'])\n",
    "      .merge(f_stats_df.loc[:, ['pancan']],\n",
    "             left_index=True, right_index=True)\n",
    ")\n",
    "min_max_df.sort_values(by='pancan', ascending=False).head(15)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Calculate \"outlier-ness\" of correlation distributions\n",
    "\n",
    "To partition genes with strong univariate correlations into \"driven by a single cancer type\" and \"driven by two or more cancer types, we can use the [local outlier factor](https://en.wikipedia.org/wiki/Local_outlier_factor) of the cancer type with the maximum correlation.\n",
    "\n",
    "A more negative LOF means the distribution is more \"outlier-ish\", and a LOF closer to 1 means the distribution is more uniform/less likely to contain an outlier."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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>max_cancer_type</th>\n",
       "      <th>max_f_statistic</th>\n",
       "      <th>max_lof</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gene</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>126695</th>\n",
       "      <td>STAD</td>\n",
       "      <td>26.357617</td>\n",
       "      <td>-43.547103</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>147798</th>\n",
       "      <td>BRCA</td>\n",
       "      <td>148.020744</td>\n",
       "      <td>-10.573249</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10053</th>\n",
       "      <td>BRCA</td>\n",
       "      <td>24.230451</td>\n",
       "      <td>-2.920684</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54845</th>\n",
       "      <td>BRCA</td>\n",
       "      <td>127.231395</td>\n",
       "      <td>-5.035958</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>582</th>\n",
       "      <td>BRCA</td>\n",
       "      <td>188.994801</td>\n",
       "      <td>-28.765280</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       max_cancer_type  max_f_statistic    max_lof\n",
       "gene                                              \n",
       "126695            STAD        26.357617 -43.547103\n",
       "147798            BRCA       148.020744 -10.573249\n",
       "10053             BRCA        24.230451  -2.920684\n",
       "54845             BRCA       127.231395  -5.035958\n",
       "582               BRCA       188.994801 -28.765280"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.neighbors import LocalOutlierFactor\n",
    "\n",
    "lof_rows = []\n",
    "for gene_id, row in f_stats_df.iterrows():\n",
    "    row = row.drop(index='pancan')\n",
    "    max_ix = row.idxmax()\n",
    "    max_f_statistic = row.max()\n",
    "    lof = LocalOutlierFactor(n_neighbors=2,\n",
    "                             contamination='auto')\n",
    "    lof.fit_predict(\n",
    "        row.values.reshape(-1, 1)\n",
    "    )\n",
    "    max_lof = lof.negative_outlier_factor_[\n",
    "        row.index.get_loc(max_ix)\n",
    "    ]\n",
    "    lof_rows.append([gene_id, max_ix, max_f_statistic, max_lof])\n",
    "    \n",
    "lof_df = (\n",
    "    pd.DataFrame(lof_rows,\n",
    "                 columns=['gene', 'max_cancer_type',\n",
    "                          'max_f_statistic', 'max_lof'])\n",
    "      .set_index('gene')\n",
    ")\n",
    "lof_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "rank_df = (min_max_df\n",
    "    .merge(lof_df, left_index=True, right_index=True)\n",
    "    .sort_values(by='pancan', ascending=False)\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'TP53 pan-cancer f-statistic vs. max individual LOF, per gene')"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set({'figure.figsize': (8, 6)})\n",
    "\n",
    "rank_df['abs_max_lof'] = rank_df.max_lof.abs()\n",
    "\n",
    "sns.scatterplot(data=rank_df, x='pancan', y='abs_max_lof')\n",
    "plt.xlabel('pan-cancer f-statistic')\n",
    "plt.ylabel('abs(max LOF)')\n",
    "plt.title('{} pan-cancer f-statistic vs. max individual LOF, per gene'\n",
    "            .format(gene))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'TP53 max single-cancer f-statistic vs. max individual LOF, per gene')"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set({'figure.figsize': (8, 6)})\n",
    "\n",
    "sns.scatterplot(data=rank_df, x='max_f_statistic', y='abs_max_lof')\n",
    "plt.xlabel('highest single-cancer f-statistic')\n",
    "plt.ylabel('abs(max LOF)')\n",
    "plt.title('{} max single-cancer f-statistic vs. max individual LOF, per gene'\n",
    "          .format(gene))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# want to look at correlation of f-statistics with sample size\n",
    "ss_df = (y_df\n",
    "    .groupby('cancer_type')\n",
    "    .count()\n",
    "    .loc[:, ['subtype']]\n",
    "    .rename(columns={'subtype': 'count'})\n",
    ")\n",
    "\n",
    "def plot_f_dist(plot_gene):\n",
    "    dist_df = f_stats_df.loc[\n",
    "        plot_gene, ~(f_stats_df.columns == 'pancan')\n",
    "    ]\n",
    "    f_ss_df = (f_stats_df\n",
    "        .T\n",
    "        .loc[f_stats_df.T.index != 'pancan', [plot_gene]]\n",
    "        .rename(columns={plot_gene: 'f_statistic'})\n",
    "        .merge(ss_df, left_index=True, right_index=True)\n",
    "    )\n",
    "    return dist_df, f_ss_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['24137', '23397', '5347', '699', '9493', '29127', '55325', '79172',\n",
       "       '146909', '57659'],\n",
       "      dtype='object')"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sorted_genes = min_max_df.pancan.sort_values(ascending=False).index\n",
    "display(sorted_genes[:10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "symbol_to_entrez, _ = tu.get_symbol_map()\n",
    "entrez_to_symbol = {v: k for k, v in symbol_to_entrez.items()}\n",
    "assert entrez_to_symbol[673] == 'BRAF'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "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>entrez_id</th>\n",
       "      <th>max - min</th>\n",
       "      <th>pancan</th>\n",
       "      <th>max_cancer_type</th>\n",
       "      <th>max_f_statistic</th>\n",
       "      <th>max_lof</th>\n",
       "      <th>abs_max_lof</th>\n",
       "      <th>symbol</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>symbol</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>KIF4A</th>\n",
       "      <td>24137</td>\n",
       "      <td>290.781887</td>\n",
       "      <td>1215.285958</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>290.782276</td>\n",
       "      <td>-15.805231</td>\n",
       "      <td>15.805231</td>\n",
       "      <td>KIF4A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NCAPH</th>\n",
       "      <td>23397</td>\n",
       "      <td>302.217636</td>\n",
       "      <td>1170.548897</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>302.217668</td>\n",
       "      <td>-7.904053</td>\n",
       "      <td>7.904053</td>\n",
       "      <td>NCAPH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PLK1</th>\n",
       "      <td>5347</td>\n",
       "      <td>351.053548</td>\n",
       "      <td>1124.041606</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>351.056898</td>\n",
       "      <td>-28.995296</td>\n",
       "      <td>28.995296</td>\n",
       "      <td>PLK1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>BUB1</th>\n",
       "      <td>699</td>\n",
       "      <td>277.920176</td>\n",
       "      <td>1061.147200</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>277.925279</td>\n",
       "      <td>-7.312019</td>\n",
       "      <td>7.312019</td>\n",
       "      <td>BUB1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>KIF23</th>\n",
       "      <td>9493</td>\n",
       "      <td>225.278353</td>\n",
       "      <td>923.233478</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>225.280224</td>\n",
       "      <td>-19.845620</td>\n",
       "      <td>19.845620</td>\n",
       "      <td>KIF23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>RACGAP1</th>\n",
       "      <td>29127</td>\n",
       "      <td>128.586952</td>\n",
       "      <td>862.842986</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>128.612580</td>\n",
       "      <td>-4.382810</td>\n",
       "      <td>4.382810</td>\n",
       "      <td>RACGAP1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UFSP2</th>\n",
       "      <td>55325</td>\n",
       "      <td>100.424225</td>\n",
       "      <td>823.754151</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>100.424832</td>\n",
       "      <td>-17.404699</td>\n",
       "      <td>17.404699</td>\n",
       "      <td>UFSP2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CENPO</th>\n",
       "      <td>79172</td>\n",
       "      <td>279.389949</td>\n",
       "      <td>821.075427</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>279.403738</td>\n",
       "      <td>-7.746309</td>\n",
       "      <td>7.746309</td>\n",
       "      <td>CENPO</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>KIF18B</th>\n",
       "      <td>146909</td>\n",
       "      <td>183.466143</td>\n",
       "      <td>807.072450</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>183.467642</td>\n",
       "      <td>-19.750423</td>\n",
       "      <td>19.750423</td>\n",
       "      <td>KIF18B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ZBTB4</th>\n",
       "      <td>57659</td>\n",
       "      <td>299.319467</td>\n",
       "      <td>789.777604</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>299.323167</td>\n",
       "      <td>-15.644646</td>\n",
       "      <td>15.644646</td>\n",
       "      <td>ZBTB4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>APH1B</th>\n",
       "      <td>83464</td>\n",
       "      <td>248.412445</td>\n",
       "      <td>764.606987</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>248.413530</td>\n",
       "      <td>-81.341113</td>\n",
       "      <td>81.341113</td>\n",
       "      <td>APH1B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ESRP1</th>\n",
       "      <td>54845</td>\n",
       "      <td>127.231178</td>\n",
       "      <td>753.519282</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>127.231395</td>\n",
       "      <td>-5.035958</td>\n",
       "      <td>5.035958</td>\n",
       "      <td>ESRP1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CCNB1</th>\n",
       "      <td>891</td>\n",
       "      <td>186.226963</td>\n",
       "      <td>749.973212</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>186.244108</td>\n",
       "      <td>-12.098043</td>\n",
       "      <td>12.098043</td>\n",
       "      <td>CCNB1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ZNF671</th>\n",
       "      <td>79891</td>\n",
       "      <td>94.776546</td>\n",
       "      <td>744.611032</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>94.851271</td>\n",
       "      <td>-3.827293</td>\n",
       "      <td>3.827293</td>\n",
       "      <td>ZNF671</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>BBS1</th>\n",
       "      <td>582</td>\n",
       "      <td>188.980197</td>\n",
       "      <td>744.360880</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>188.994801</td>\n",
       "      <td>-28.765280</td>\n",
       "      <td>28.765280</td>\n",
       "      <td>BBS1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>EZH1</th>\n",
       "      <td>2145</td>\n",
       "      <td>257.595577</td>\n",
       "      <td>736.515306</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>257.740057</td>\n",
       "      <td>-32.180675</td>\n",
       "      <td>32.180675</td>\n",
       "      <td>EZH1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NICN1</th>\n",
       "      <td>84276</td>\n",
       "      <td>117.028257</td>\n",
       "      <td>659.651258</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>117.055612</td>\n",
       "      <td>-1.888055</td>\n",
       "      <td>1.888055</td>\n",
       "      <td>NICN1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>KIF11</th>\n",
       "      <td>3832</td>\n",
       "      <td>156.123663</td>\n",
       "      <td>616.061787</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>156.234640</td>\n",
       "      <td>-7.043517</td>\n",
       "      <td>7.043517</td>\n",
       "      <td>KIF11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>FAM219B</th>\n",
       "      <td>57184</td>\n",
       "      <td>64.069128</td>\n",
       "      <td>611.847961</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>64.070699</td>\n",
       "      <td>-1.410630</td>\n",
       "      <td>1.410630</td>\n",
       "      <td>FAM219B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>EID1</th>\n",
       "      <td>23741</td>\n",
       "      <td>93.355556</td>\n",
       "      <td>593.133112</td>\n",
       "      <td>BRCA</td>\n",
       "      <td>93.377861</td>\n",
       "      <td>-4.417216</td>\n",
       "      <td>4.417216</td>\n",
       "      <td>EID1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        entrez_id   max - min       pancan max_cancer_type  max_f_statistic  \\\n",
       "symbol                                                                        \n",
       "KIF4A       24137  290.781887  1215.285958            BRCA       290.782276   \n",
       "NCAPH       23397  302.217636  1170.548897            BRCA       302.217668   \n",
       "PLK1         5347  351.053548  1124.041606            BRCA       351.056898   \n",
       "BUB1          699  277.920176  1061.147200            BRCA       277.925279   \n",
       "KIF23        9493  225.278353   923.233478            BRCA       225.280224   \n",
       "RACGAP1     29127  128.586952   862.842986            BRCA       128.612580   \n",
       "UFSP2       55325  100.424225   823.754151            BRCA       100.424832   \n",
       "CENPO       79172  279.389949   821.075427            BRCA       279.403738   \n",
       "KIF18B     146909  183.466143   807.072450            BRCA       183.467642   \n",
       "ZBTB4       57659  299.319467   789.777604            BRCA       299.323167   \n",
       "APH1B       83464  248.412445   764.606987            BRCA       248.413530   \n",
       "ESRP1       54845  127.231178   753.519282            BRCA       127.231395   \n",
       "CCNB1         891  186.226963   749.973212            BRCA       186.244108   \n",
       "ZNF671      79891   94.776546   744.611032            BRCA        94.851271   \n",
       "BBS1          582  188.980197   744.360880            BRCA       188.994801   \n",
       "EZH1         2145  257.595577   736.515306            BRCA       257.740057   \n",
       "NICN1       84276  117.028257   659.651258            BRCA       117.055612   \n",
       "KIF11        3832  156.123663   616.061787            BRCA       156.234640   \n",
       "FAM219B     57184   64.069128   611.847961            BRCA        64.070699   \n",
       "EID1        23741   93.355556   593.133112            BRCA        93.377861   \n",
       "\n",
       "           max_lof  abs_max_lof   symbol  \n",
       "symbol                                    \n",
       "KIF4A   -15.805231    15.805231    KIF4A  \n",
       "NCAPH    -7.904053     7.904053    NCAPH  \n",
       "PLK1    -28.995296    28.995296     PLK1  \n",
       "BUB1     -7.312019     7.312019     BUB1  \n",
       "KIF23   -19.845620    19.845620    KIF23  \n",
       "RACGAP1  -4.382810     4.382810  RACGAP1  \n",
       "UFSP2   -17.404699    17.404699    UFSP2  \n",
       "CENPO    -7.746309     7.746309    CENPO  \n",
       "KIF18B  -19.750423    19.750423   KIF18B  \n",
       "ZBTB4   -15.644646    15.644646    ZBTB4  \n",
       "APH1B   -81.341113    81.341113    APH1B  \n",
       "ESRP1    -5.035958     5.035958    ESRP1  \n",
       "CCNB1   -12.098043    12.098043    CCNB1  \n",
       "ZNF671   -3.827293     3.827293   ZNF671  \n",
       "BBS1    -28.765280    28.765280     BBS1  \n",
       "EZH1    -32.180675    32.180675     EZH1  \n",
       "NICN1    -1.888055     1.888055    NICN1  \n",
       "KIF11    -7.043517     7.043517    KIF11  \n",
       "FAM219B  -1.410630     1.410630  FAM219B  \n",
       "EID1     -4.417216     4.417216     EID1  "
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rank_df['symbol'] = rank_df.index.to_series().astype(int).map(entrez_to_symbol)\n",
    "rank_df.index.name = 'entrez_id'\n",
    "rank_df = rank_df.reset_index().set_index('symbol', drop=False)\n",
    "rank_df.head(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "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>f_statistic</th>\n",
       "      <th>count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>BRCA</th>\n",
       "      <td>351.056898</td>\n",
       "      <td>981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UCEC</th>\n",
       "      <td>72.698743</td>\n",
       "      <td>507</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>LIHC</th>\n",
       "      <td>57.952473</td>\n",
       "      <td>348</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>LUAD</th>\n",
       "      <td>55.548095</td>\n",
       "      <td>502</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>BLCA</th>\n",
       "      <td>49.620312</td>\n",
       "      <td>398</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ACC</th>\n",
       "      <td>36.862054</td>\n",
       "      <td>76</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>KIRC</th>\n",
       "      <td>34.460291</td>\n",
       "      <td>352</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>STAD</th>\n",
       "      <td>23.312115</td>\n",
       "      <td>383</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GBM</th>\n",
       "      <td>15.623279</td>\n",
       "      <td>119</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PRAD</th>\n",
       "      <td>15.503404</td>\n",
       "      <td>479</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      f_statistic  count\n",
       "BRCA   351.056898    981\n",
       "UCEC    72.698743    507\n",
       "LIHC    57.952473    348\n",
       "LUAD    55.548095    502\n",
       "BLCA    49.620312    398\n",
       "ACC     36.862054     76\n",
       "KIRC    34.460291    352\n",
       "STAD    23.312115    383\n",
       "GBM     15.623279    119\n",
       "PRAD    15.503404    479"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# this is an example of a fairly skewed distribution\n",
    "# (the cancer type with the top f-statistic, BRCA, has a large LOF)\n",
    "plot_gene = sorted_genes[2]\n",
    "dist_df, f_ss_df = plot_f_dist(plot_gene)\n",
    "f_ss_df.sort_values(by='f_statistic', ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Sample count vs. f-statistic, per cancer type')"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtYAAAGDCAYAAAAPu1cfAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy86wFpkAAAACXBIWXMAAAsTAAALEwEAmpwYAABV+klEQVR4nO3deVxU5f4H8M8sDDuiSWYXIVzILa5bmrmFV8NccgUFQ6+oqVmGK4hr4gIuLfpzr+xeXJBcylveum7FzYXKwgUly3A3xQVhBhiGmef3h3kuo+yeWYDP+/Xy9Zo56/c5M3P88Mwz5yiEEAJERERERPRYlLYugIiIiIioOmCwJiIiIiKSAYM1EREREZEMGKyJiIiIiGTAYE1EREREJAMGayIiIiIiGTBYExERERHJgMGaiIiIiEgGDNZEJbhz5w5GjRqF559/HjExMbYuh4iIiOyc2tYFENmr9evXw9fXF5s2bbJ1KURERFQFsMfagnbs2IHg4GD07t0bPXr0wKhRo3DixAmL7/fKlSto1qwZ+vfvL/179dVXsWPHDml+69atS1z34Xl79+5Fhw4dcPToUWmaEAJRUVH46KOPSq1l2bJl+O677x6zRbZx5MgR9OrVq0LrRERE4M6dO+Ve7tSpU5g0aZLsy5emvK9dSct9/vnnePXVV9G/f38MGzYMp06demTd/fv3l/geq4hvvvkG/fr1Q1BQECZNmgStVivN27x5M/r06YO+fftiwoQJuH37NgBAq9VizJgxyM/Pf+z9U/WXmpqK8PBw9OvXD3379sWYMWPw66+/WmXfzz77bLnOF7ZQ3nPZ4zh79ix69OiBQYMG4cqVK+VaJycnByNGjKjwcv3790d2drZsy9P/lPc1qTEEWcSKFSvEsGHDxJUrV6RpR44cER06dBBXr1616L4vX74sWrVqZTbtjz/+EO3atRNnz54tdn5J627btk106dJFnDlzRpr222+/ifDwcPHXv/5VfPjhhyXW8fPPP4tx48Y9ZmusT6/XizZt2gh/f3/RqlUr0bdv33Kv6+/vL27fvi3bcpVdviTlfe1KWu78+fOiU6dO4saNG0IIIb755hvRrVs3s3UzMjJEjx49SnyPldft27fFCy+8IDIyMoQQQixdulTMmzdPCCHEqVOnRGBgoMjOzhZCCBEXFyfmzJkjrbt7924RFxf3WPun6k+v14v27duL06dPS9M+++wz0a1bN1FYWGjx/cv1ubYEa9S2atUqERMTU6F1Svv/qzLLVXZ5+h8eO3McCmIBt27dwj/+8Q/s27cPTz75pDS9Y8eOiI6ORl5eHgDg4MGDWLt2LQwGA5ycnBAVFYXWrVsjJSUF7733Hho0aIBff/0VhYWFeOedd9C2bdsS1ylLvXr14OvriwsXLqBly5blaseGDRuwa9cubN26Fd7e3tL0LVu2IDg4GE8//XSp669atQqvvfYaACAlJQXLly/H008/jd9//x1OTk6Ii4tDo0aNYDKZsHjxYpw4cQI6nQ5CCCxcuBCFhYUlHgfg/jcCmzZtglKpRO3atREfH4969eoVu622bduWelyL0mg02L59O0aMGIEjR4480i6dToeZM2fi4sWLUCqVaNGiBRYsWIBZs2YBAEaOHIl169Zh06ZNxdYxc+ZMabnRo0fjww8/xBdffFGu7RZdvqRjUL9+fYwdOxbDhg3D3/72N7Pay/valbScRqPBwoULpfd1y5YtcevWLRQUFECj0SAvLw/Tp09HdHQ0pk2bZrZuWe/dZ599Fr/88ov0/LvvvsNzzz2HZ555BgAQGhqK/v37Y968eWjZsiW+/vprODg4QK/X48aNG2bv0VdeeQXLly/H6NGjUbdu3VLbSjVXXl4ecnJykJubK0179dVX4ebmBqPRCIVCUeL5JDo6Gk5OTjh37hxu376N7t27w9PTE4cOHUJmZiYWLlwonfMdHR2Rnp6O27dvo1OnTpg9ezYcHBzMavn000+xbds2mEwmeHp6Ys6cOWjUqNEjNZf0md++fTsSEhKgVCpRt25dzJkzB35+foiOjkaTJk0wevRoADB73r17dwwcOBBHjx7F9evX0b9/f0RGRpqdozZs2ID69esDAIxGI7p3747Vq1dL/49ERkaiffv26NChA2bNmoWCggIIITBkyBAMHz68xGO/Z88ebNu2DUajEfn5+VixYoXZ/MzMTERFReHu3bsAgG7dukm15efno3///ti1axd2796N7du3w2Aw4N69exg7dizCwsIeWa558+Y4evQojEZjubb7YPk6depg/fr12L17N9RqNXx9fREXFwd3d/cS27Zq1SpcvHgRf/zxBzIzM9G0aVMsWrQIbm5uuHHjBhYsWIDr16/DYDCgT58+GD9+PK5cuYLhw4ejUaNGuHr1KhISEszyQ2ZmJubNm4fff/8dSqUSw4YNw4gRI5Camoply5ahoKAAmZmZePHFF7F48WJcuXIFf//739GtWzecOHEC2dnZmD59Onr27InCwkIsW7YM33zzDVQqFVq3bo158+ZBo9Fg7dq1+M9//gOTyYS//OUvmDdvHurVq4fw8HDUqlULv//+O0JDQxEeHi7VVvTYjR49Glu3bkViYiIA4Nq1awgJCcHBgwfRqlUrjB07Fv/973+Rm5uLKVOm4OWXX67Q+79KsGmsr6b27dsnBg4cWOoyGRkZom/fvuLOnTtCCCHOnTsnOnXqJHQ6nTh27Jho1qyZ1Ev80UcfieHDh5e6TlHF/fX4008/ieeff15cu3atXD3W8fHxwt/fX2zevLnENkRFRZXY63nv3j3x17/+Vej1eiGEEMeOHRNNmzYVP/zwgxBCiK1bt0rH6KeffhJvvfWWMBqNQggh1q9fL8aNG1ficRBCiLNnz4oOHTqIa9euCSGE2LRpk5gzZ06J23pQQ0nbe9iePXvEqFGjip23e/duERERIYQQorCwUMyaNUtcuHBBCPG/Xp7S6ii63LFjx0SfPn3Kvd2iy5d0DMqjtNeuvMuZTCYxdepU8dZbb0nTpk2bJj799NNH3mPlee/6+/ubbX/9+vVm7TEYDMLf31/k5ORI0/bt2yfat28vOnfuLPVsPzBu3DixY8eOMttINdvHH38sAgICRPfu3aX3b25urhCi5HOTEPc/G8HBwaKgoEDcvHlT+Pv7i3/+859CCCE++eQT6fwRFRUlBgwYILRardDr9WL48OEiISFBCPG/z3VKSooICwuT9vvf//5X9OrV65FaS/rMHzlyRPTo0UPqYd65c6d45ZVXhMlkeuQzXPR5YGCg9M3OH3/8IZ577jlx6dIls9oe9sEHH4h33nlHCCFEVlaWaN++vcjOzhYzZ84U69evF0IIcfPmTREZGSkdt5KsXLlS2tbD/u///k/6/Ot0OhEZGSmys7PNzi1arVaEhIRI55Wff/5ZmvfwOehBe8qz3aLL79+/X7z88ssiKytLCCHE4sWLxZo1a8psV9euXUVmZqYwGo1iypQp0nEODw8XBw4cEEIIkZ+fL8LDw8WXX34pLl++LPz9/aX/Ix82ceJEER8fL4QQIjs7W/Tp00dcuHBBTJ48WRw7dkw6Hh06dBCnTp2Stnfw4EEhhBBfffWVeOmll4QQQvzjH/8Qw4cPF3l5ecJoNIq3335b7N69W+zevVtERkYKg8EghBAiMTFRjBkzRgghxGuvvSZmzpxZbG1Fj51erxcdO3YU586dE0II8f7774vly5dLx3Tt2rVCiPvv5bZt21bo/V9VsMfaAoQQZs+1Wq30l3tubi5eeeUV1KtXDzdv3sTf//53aTmFQoFLly4BAJ5++mk0a9YMANC8eXPs3r0bhw8fLnGdpk2bmu3zwV+PwP1ehtq1a2PZsmWoX79+mWPZcnNzce7cOWzYsAGTJ09G69at0bx58wodg4sXL8LLywsajUaa1rRpU7Rr1w4AMHjwYCxYsAB3795F69atUatWLSQmJuLy5ctISUmBq6triccBAI4ePYrOnTtLPSlFj0lJ2yptew87e/Ysnn322WLntW3bFu+99x7Cw8Px4osvYuTIkfD19TVbprQ2laQ82y2qtGNgabm5uYiOjsYff/yBDz/8EMD9Xm61Wo0hQ4Y88h4r6b376aefYteuXdK0B+/ZadOmwWQyQaFQPLJvpfJ/Pw3p0aMHevTogaSkJIwePRr79u2T5nt7eyMjI0O2NlP1NGrUKAQHB+OHH37ADz/8gI0bN2Ljxo3YsWNHmZ/jwMBAODg4wMvLCy4uLujSpQsAwMfHB1lZWdJyAwcOlNbr378/Dhw4IH2bB9z/LcHFixcxbNgwaVp2djaysrLg6ekpTSvpM7906VL07t0bderUAQAMGjQIixYtKte45QffatWrVw9PPPEE7t27hwYNGpS4/ODBgzFkyBBER0fjiy++QPfu3eHu7o6ePXsiKioKJ0+eRMeOHTF79myzz2pFdenSBa+//jquX7+OF198EVOnToW7uzvu3bsnLePq6op169bh22+/xYULF5Cenm727UNlt1vU0aNH0atXL9SqVQsApN78svTq1Uv6tmzIkCFYvHgx3nrrLfzwww+4d+8ePvjgAwD3z6Xp6ekICAiAWq1Gq1atit3ekSNHMH36dACAu7u79K1lXFwckpOTsW7dOvz+++/Q6/XIzc2Fp6cnHBwc0K1bNwD3/7978J48cuQI+vfvDycnJwDA+++/DwB4++23cerUKQwePBgAYDKZpG/YAUj/f5dGo9EgODgYn376KaKiorB7924kJCRI8x+875s2bQp/f3/88MMPOHHiRLne/1UFg7UFBAQEICMjA3fv3kXt2rXh5uaGzz//HMD9r4ju3r0LLy8vdOzYUXpDA8D169fx5JNP4scff5Te8MD9ACKEgMlkKnGdhzk5OUn7rCgnJyesXbsWDg4OGDduHN58803s2rWrQm9whUIBk8lkNk2lUj2ynEqlwjfffINFixZh1KhR+Nvf/oaGDRtiz549Ui1Ft/ngjxaVSmUWuvLz83H16lVcvny5xG2Vtr2H/fLLL3j11VcBAB988AEOHjwIAOjevTvefvtt7Nu3DykpKTh27BhGjRqFBQsWoHv37tL6pbWpJA0aNChzuw8fu+KOgaW/Prt27RrGjx+PRo0a4Z///Kd0THfv3i39QWcwGKTHGzZsKPW9O3LkSAD3h4IUfc/euXPH7Me+N27cQK1ateDi4oKLFy8iMzPT7A+1efPm4d69e6hduzYAwMHBodj3HNEDx48fx88//4wxY8YgMDAQgYGBmDJlCvr27YvDhw/Dycmp1M9x0Y4DAFCri/8vtej7UAjxSOA0mUzo37+/FJxMJhNu3rwphbmi2ynuM//wufbBfgoLCx85zxkMBrPlHB0dpcelnRMf+Mtf/oLmzZvjm2++wa5du6RLkQYGBuLrr7/GkSNHcPToUaxevRq7du3CU089Ver2Hpg1axZOnz4NABg2bBhCQ0Nx4MABHD16FMeOHUNwcDA2btxo9v/QH3/8gaFDhyIkJARt27ZFr169cOjQoVL3ExAQUOZ2i3r4mGdnZyM7O9ts6FlJ6z1gMpmgVCphMpkghEBiYiKcnZ0B3D/POTo64u7du9BoNCW+h9RqtVkdly9fRu3atREREYFnn30WXbp0wSuvvIITJ05Ir6GDg4P0Xiu67sP7uHXrFkwmE0wmE8aMGYOwsDAAQEFBgdkfHC4uLqW2+YFhw4ZhyJAhaN++PZo0aWL2h9rDx0WlUpX7/V9V8KogFlCvXj2MGDECb7/9Nq5duyZNv3r1Kn766ScolUp07NgRhw8fxvnz5wEA3377LV599dVSr2RQmXUqQ6lUSuP/Xn/9dTRu3BhTp04t9uRdEh8fH9y+fRt6vV6alp6ejvT0dADA9u3b0bp1a3h4eODw4cMIDAxEWFgYWrZsif3798NoNJa6/QdXKbl58yYAIDExEcuWLavUtoqTnp4ufQvw9ttv4/PPP8fnn3+Ot99+G1u3bsXMmTPRuXNnTJ8+HZ07d8aZM2cA3D9pFBYWllnHg+WKKs92y3MMLEmr1SI8PBwvv/wy3nvvPbM/VHbs2IEvvvgCn3/+OTZs2CD9cVevXr1KvXc7d+6MEydO4MKFC1L7HvSuZWZmYsqUKdJVC/71r3+hSZMmUqgG7l/hxs/PT+5DQNVInTp1sHbtWvz444/StMzMTGi1Wvj7+8t2Pvn3v/+NgoIC6PV67N69G4GBgWbzO3fujC+//FL6LG/btk36g7Ookj7zXbp0wd69e6XPw86dO+Hp6QlfX1/Url1bCqw3btzA999/X66aizvnPBASEoKNGzciLy9P+o3K1KlTsXfvXvTp0wfz5s2Dm5ub9A1seSxatEg6z4aGhmL58uVYs2YNevTogVmzZqFx48b49ddfoVarYTQaIYTA6dOnUadOHbzxxhvo3LmzFKqNRqPZckWVZ7tFvfjii9i3b590RaJVq1bhk08+KbM9Bw4cQE5ODkwmE5KSkhAYGAg3Nze0atVKuoRrdna29AdEWTp27IidO3cCuH8VjpEjR+LChQs4deoUpk2bhpdffhl//PEHLl26VOb/1R07dsQXX3yBgoICmEwmzJ8/H19++SU6d+6MHTt2SG394IMPMGPGjDJre/jY1a9fH61atcLixYsRGhpqtuxnn30GAEhLS0NGRgaef/75cr//qwr2WFvI5MmTsWfPHkydOlX6gUytWrXQu3dvDB8+HM7OzliwYAGmTJkCIQTUajXWrl1b6nCBxo0bV3idkuTm5j7yo8fExMRHtqVQKBAfH4+BAwfi/fffx5QpU8q1fQ8PD7Rt2xbHjh2TvoqqW7cu3n//fVy9ehV16tTB0qVLAdz/63bq1Kno168fCgsL0alTJ+nHEyV59tlnMX36dIwZMwYA4OXlhcWLF0Or1VZ4Ww/LzMzEvXv30LBhw2LnDxgwAN9//z169+4NZ2dn1K9fX/ohR69evRAeHo5Vq1ZhxowZxdahVCql5Yr+AKQ82y26fEnHAECJP14sTXnW2bJlC65du4Z9+/Zh37590vRPPvnELNQ+rDzv3aI/XASAJ554AkuWLMGkSZNgMBjg4+OD+Ph4APe/khw/fjxGjBgBlUqFJ598EqtXr5bWLSgoQGpqKhYtWlTu9lPN4+fnh9WrV+O9997DH3/8AUdHR7i7u2Px4sVo2LBhpc5NxXFyckJYWBiys7MRFBQkfdX+QOfOnTF27FhERERAoVDAzc0N//d///fIUKiSPvP16tXD3//+d4wcORImk0n6wZ1SqUR4eDimTZuGoKAgeHt744UXXihXzUXPZf7+/mbzunfvjnfeeQdjx46Vpr3xxhuYNWsWtm/fDpVKhR49euD555/HjRs38Prrr2PDhg2oV69euY/ZyJEjER0djb59+0Kj0eDZZ59Fnz59oFKpEBAQgD59+mDTpk2oV68eevXqBYVCgfbt26NOnTq4ePEifH19peW2bNlSoe0WXb5bt2747bffpIDYuHFjxMbGArg/rGfhwoV47rnnHqm/bt26GDt2LO7evYvnn38e48ePB3A/2MfGxqJfv34oKChA37598eqrr5Y5bGfu3LmYP38++vXrByEExo0bh5YtW+L111/HwIED4eLignr16qFNmza4ePFiqcN5hg0bhqtXr2LQoEEQQqB9+/YIDw+HUqnEjRs3EBISAoVCgfr16yMuLq7M18rLy8vs2NWuXRuDBg1CbGys9P//Az/99BOSkpJgMpnw3nvvoVatWuV+/1cVClHW9z5ElfTTTz9h3bp12LBhA1JSUhAbGyuNCyOypF27duHXX39FVFSUrUuhGu7hq3JQ9fHee+/h1VdffWT43YMhn3PnzrVRZbZlMpmwYMECPP3003j99del6c8++6x0pZXqjENByGLatGkDPz8/JCcn27oUqkF0Oh2++OILvPXWW7YuhYiqKSEE/vKXv1TdS8JZiFarRYcOHXD9+vUae9MY9lgTEREREcmAPdZERERERDJgsCYiIiIikgGDNRERERGRDKr05fZMJhOMxooPEVepFJVazxpYW+XZc332XBtg3/XZc21A5etzcKh5N68pes6299dVTjWlrTWlnQDbWh2V1c7ynrOrdLA2GgWyskq/fWlxPD1dKrWeNbC2yrPn+uy5NsC+67Pn2oDK1+fl5W6Bauxb0XO2vb+ucqopba0p7QTY1uqorHaW95zNoSBERERERDJgsCYiIiIikgGDNRERERGRDBisiYiIiIhkwGBNRERERCQDBmsiIiIiIhkwWBMRERERyYDBmoiIiIhIBgzWREREREQyYLAmIiIioupPAWTnGXApU4fs/EJAIf8uZL+lucFgQExMDK5evYqCggJMmDABjRs3RnR0NBQKBZo0aYJ58+ZBqfxfpjeZTJg/fz5++eUXaDQaLFy4EL6+vnKXRkREZTAajZg9ezYyMjKgUqmwZMkS5OTkYPz48XjmmWcAAKGhoejduzeSkpKQmJgItVqNCRMmIDAw0LbFExGVRAGcvXQPK5NSoTcY4eigwqSQVmjmUwsQ8u1G9mC9Z88eeHp6YtmyZbh79y4GDhyIpk2bIjIyEh06dMDcuXNx4MAB9OzZU1pn//79KCgowPbt25Gamoq4uDisXbtW7tKIiKgMhw4dAgAkJiYiJSUFS5YsQffu3TFq1ChERERIy2VmZiIhIQE7d+6EXq9HWFgYOnXqBI1GY6vSiYhKlJ1rkEI1AOgNRqxMSkX8Gy/Cw9lBtv3IHqx79eqFoKAg6blKpUJaWhrat28PAOjatSsOHz5sFqyPHz+OLl26AABatWqF06dPy10WERGVQ48ePfDSSy8BAK5du4a6devi9OnTyMjIwIEDB+Dr64uYmBicPHkSrVu3hkajgUajgY+PD9LT0xEQEGDbBhARFSNLWyCF6gf0BiOydAX2HaxdXV0BAFqtFpMmTUJkZCTi4+OhUCik+Tk5OWbraLVauLm5Sc9VKhUKCwuhVpdenkqlgKenS4VrNBhNcHRxLHm7SgU0atsMP1eplJVqkzXYc22Afddnz7UB9l2fPdcG2H99laFWqxEVFYV9+/Zh5cqVuHHjBoKDg9GyZUusXbsWq1evRtOmTeHu7i6t4+rqCq1WW+p2i56zq+NxK0lNaWtNaSfAtlZF+UYBRweVWbh2dFDBq7YLPD1dZGun7MEaAK5fv46JEyciLCwM/fr1w7Jly6R5Op0OHh4eZsu7ublBp9NJz00mU5mhGgCMRoGsrNwK1+fo4ojYD4+WOH9WRAfkavMrvF05eHq6VKpN1mDPtQH2XZ891wbYd332XBtQ+fq8vNzLXsiG4uPjMW3aNISEhCAxMRH16tUDAPTs2ROxsbFo166d2Xlbp9OZBe3iFD1n2/vrKqea0taa0k6Aba2KnNQKTApp9cgYa2e1AllZuWW2s7znbNm7ZW/duoWIiAhMnz4dQ4YMAQA0b94cKSkpAIDk5GS0a9fObJ02bdogOTkZAJCamgp/f3+5yyIionL47LPPsH79egCAs7MzFAoF3nzzTZw8eRIAcPToUbRo0QIBAQE4fvw49Ho9cnJycP78eZ67ich+CaCZTy3Ev/Ei5o/pgPg3XpT9h4uABXqs161bh+zsbKxZswZr1qwBAMyaNQsLFy7Eu+++i4YNG0pjsGfMmIHIyEj07NkThw8fxrBhwyCEwOLFi+Uui4iIyuHll1/GzJkzMXz4cBQWFiImJgb169dHbGwsHBwcULduXcTGxsLNzQ3h4eEICwuDEAKTJ0+Go2PJQ+yIiGxOAB7ODv8bUy1zqAYAhRDCApu1DoPBaLGhIIZ8w+OUVmn2/JWLPdcG2Hd99lwbYN/12XNtQPUdCmIJRc/Z9v66yqmmtLWmtBNgW6sjux0KQkRERERUEzFYExERERHJgMGaiIiIiEgGDNZERERERDJgsCYiIiIikgGDNRERERGRDBisiYiIiIhkwGBNRERERCQDBmsiIiIiIhkwWBMRERERyYDBmoiIiIhIBgzWREREREQyYLAmIiIiIpIBgzURERERkQwYrImIiIiIZMBgTUREREQkAwZrIiIiIiIZMFgTEREREcmAwZqIiIiISAYM1kREREREMmCwJiIiIiKSAYM1EREREZEMGKyJiIiIiGTAYE1EREREJAMGayIiIiIiGTBYExERERHJgMGaiIiIiEgGDNZERERERDJgsCYiIiIikgGDNRERERGRDBisiYiIiIhkoLbUhk+cOIHly5cjISEBkydPxq1btwAAV69exV//+le89957ZssPGDAA7u7uAABvb28sWbLEUqUREREREcnOIsF648aN2LNnD5ydnQFACtH37t3DiBEjMHPmTLPl9Xo9ACAhIcES5RARERERWZxFhoL4+Phg1apVj0xftWoVXnvtNTz55JNm09PT05GXl4eIiAiMGDECqampliiLiIiIiMhiLNJjHRQUhCtXrphNu337No4ePfpIbzUAODk5YfTo0QgODsaFCxcwduxYfPXVV1CrSy9PpVLA09OlwvXlFxihVqtKnK9UKiu1XTmoVLbbd1nsuTbAvuuz59oA+67PnmsD7L8+IiKyHouNsX7YV199hb59+0KlejTQ+vn5wdfXFwqFAn5+fvD09ERmZibq169f6jaNRoGsrNwK1+Lo4ojCQmOJ800mE7Ky9BXerhw8PV0q1SZrsOfaAPuuz55rA+y7PnuuDah8fV5e7hao5vEZjUbMnj0bGRkZUKlUWLJkCYQQiI6OhkKhQJMmTTBv3jwolUokJSUhMTERarUaEyZMQGBgoK3LJyKyKasF66NHj2LChAnFztuxYwfOnTuH+fPn48aNG9BqtfDy8rJWaURE9KdDhw4BABITE5GSkiIF68jISHTo0AFz587FgQMH0KpVKyQkJGDnzp3Q6/UICwtDp06doNFobNwCIiLbsdrl9jIyMtCgQQOzaTNmzMC1a9cwZMgQ5OTkIDQ0FJMnT8bixYvLHAZCRETy69GjB2JjYwEA165dQ926dZGWlob27dsDALp27YojR47g5MmTaN26NTQaDdzd3eHj44P09HRblk5EZHMWS6/e3t5ISkqSnn/55ZePLLN06VLp8YoVKyxVChERVYBarUZUVBT27duHlStX4tChQ1AoFAAAV1dX5OTkQKvVSpdIfTBdq9WWut2iv4upSWPTa0pba0o7Aba1OpKrnewWJiKiR8THx2PatGkICQmRLokKADqdDh4eHnBzc4NOpzObXjRoF6fo72Lsfey8nGpKW2tKOwG2tToqq53l/V0M77xIRESSzz77DOvXrwcAODs7Q6FQoGXLlkhJSQEAJCcno127dggICMDx48eh1+uRk5OD8+fPw9/f35alExHZHHusiYhI8vLLL2PmzJkYPnw4CgsLERMTg0aNGmHOnDl499130bBhQwQFBUGlUiE8PBxhYWEQQmDy5MlwdHS0dflERDbFYE1ERBIXFxd88MEHj0zfvHnzI9NCQkIQEhJijbKIiKoEDgUhIiIiIpIBgzURERERkQwYrImIiIiIZMBgTUREREQkAwZrIiIiIiIZMFgTEREREcmAwZqIiIiISAYM1kREREREMmCwJiIiIiKSAYM1EREREZEMGKyJiIiIiGTAYE1EREREJAMGayIiIiIiGTBYExERERHJgMGaiIiIiEgGDNZERERERDJgsCYiIiIikgGDNRERERGRDBisiYiIiIhkwGBNRERERCQDBmsiIiIiIhkwWBMRERERyYDBmoiIiIhIBgzWREREREQyYLAmIiIiIpIBgzURERERkQwYrImIiIiIZGCxYH3ixAmEh4cDANLS0tClSxeEh4cjPDwce/fuNVvWZDJh7ty5GDp0KMLDw3Hx4kVLlUVEREREZBFqS2x048aN2LNnD5ydnQEAZ86cwahRoxAREVHs8vv370dBQQG2b9+O1NRUxMXFYe3atZYojYiIiIjIIizSY+3j44NVq1ZJz0+fPo1vvvkGw4cPR0xMDLRardnyx48fR5cuXQAArVq1wunTpy1RFhERERGRxVikxzooKAhXrlyRngcEBCA4OBgtW7bE2rVrsXr1akRFRUnztVot3NzcpOcqlQqFhYVQq0svT6VSwNPTpcL15RcYoVarSpyvVCortV05qFS223dZ7Lk2wL7rs+faAPuuz55rA+y/PiIish6LBOuH9ezZEx4eHtLj2NhYs/lubm7Q6XTSc5PJVGaoBgCjUSArK7fC9Ti6OKKw0FjifJPJhKwsfYW3KwdPT5dKtcka7Lk2wL7rs+faAPuuz55rAypfn5eXuwWqISIiW7LKVUFGjx6NkydPAgCOHj2KFi1amM1v06YNkpOTAQCpqanw9/e3RllERERERLKxSo/1/PnzERsbCwcHB9StW1fqsZ4xYwYiIyPRs2dPHD58GMOGDYMQAosXL7ZGWUREREREsrFYsPb29kZSUhIAoEWLFkhMTHxkmaVLl0qPFyxYYKlSiIioHAwGA2JiYnD16lUUFBRgwoQJeOqppzB+/Hg888wzAIDQ0FD07t0bSUlJSExMhFqtxoQJExAYGGjb4omI7IBVeqyJiMj+7dmzB56enli2bBnu3r2LgQMHYuLEiY9cLjUzMxMJCQnYuXMn9Ho9wsLC0KlTJ2g0GhtWT0RkewzWREQEAOjVqxeCgoKk5yqVCqdPn0ZGRgYOHDgAX19fxMTE4OTJk2jdujU0Gg00Gg18fHyQnp6OgIAAG1ZPRGR7DNZERAQAcHV1BXD/EqiTJk1CZGQkCgoKHrlcatOmTeHu7m623sP3JyhO0Uuk1qTLFNaUttaUdgJsa3UkVzsZrImISHL9+nVMnDgRYWFh6NevH7Kzsx+5XGq7du3MLpGq0+nMgnZJil4i1d4voyinmtLWmtJOgG2tjspqZ3kvkWqVy+0REZH9u3XrFiIiIjB9+nQMGTIEQPGXSw0ICMDx48eh1+uRk5OD8+fP8zKpRERgjzUREf1p3bp1yM7Oxpo1a7BmzRoAQHR0NBYvXmx2uVQ3NzeEh4cjLCwMQghMnjwZjo6ONq6eiMj2GKyJiAgAMHv2bMyePfuR6cVdLjUkJAQhISHWKIuIqMrgUBAiIiIiIhkwWBMRERERyYDBmoiIiIhIBgzWREREREQyYLAmIiIiIpIBgzURERERkQwYrImIiIiIZMBgTUREREQkAwZrIiIiIiIZMFgTEREREcmAwZqIiIiISAYM1kREREREMmCwJiIiIiKSAYM1EREREZEMGKyJiIiIiGTAYE1EREREJAMGayIiIiIiGTBYExERERHJgMGaiIiIiEgGDNZERERERDJgsCYiIiIikgGDNRERERGRDBisiYiIiIhkwGBNRERERCQDtaU2fOLECSxfvhwJCQk4e/YsYmNjoVKpoNFoEB8fj7p165otP2DAALi7uwMAvL29sWTJEkuVRkREREQkO4sE640bN2LPnj1wdnYGACxatAhz5sxBs2bNkJiYiI0bN2LmzJnS8nq9HgCQkJBgiXKIiIiIiCzOIkNBfHx8sGrVKun5u+++i2bNmgEAjEYjHB0dzZZPT09HXl4eIiIiMGLECKSmplqiLCIiIiIii7FIj3VQUBCuXLkiPX/yyScBAD/99BM2b96MLVu2mC3v5OSE0aNHIzg4GBcuXMDYsWPx1VdfQa0uvTyVSgFPT5cK15dfYIRarSpxvlKprNR25aBS2W7fZbHn2gD7rs+eawPsuz57rg2w//qIiMh6LDbG+mF79+7F2rVrsWHDBtSpU8dsnp+fH3x9faFQKODn5wdPT09kZmaifv36pW7TaBTIysqtcC2OLo4oLDSWON9kMiErS1/h7crB09OlUm2yBnuuDbDv+uy5NsC+67Pn2oDK1+fl5W6BaoiIyJasclWQzz//HJs3b0ZCQgIaNGjwyPwdO3YgLi4OAHDjxg1otVp4eXlZozQiIiIiIllYPFgbjUYsWrQIOp0Ob731FsLDw7Fy5UoAwIwZM3Dt2jUMGTIEOTk5CA0NxeTJk7F48eIyh4EQEREREdkTi6VXb29vJCUlAQC+//77YpdZunSp9HjFihWWKoWIiIiIyOJ4gxgiIiIiIhkwWBMRERERyYADmYmICABgMBgQExODq1evoqCgABMmTEDjxo0RHR0NhUKBJk2aYN68eVAqlUhKSkJiYiLUajUmTJiAwMBAW5dPRGRzDNZERAQA2LNnDzw9PbFs2TLcvXsXAwcORNOmTREZGYkOHTpg7ty5OHDgAFq1aoWEhATs3LkTer0eYWFh6NSpEzQaja2bQERkUwzWREQEAOjVqxeCgoKk5yqVCmlpaWjfvj0AoGvXrjh8+DCUSiVat24NjUYDjUYDHx8fpKenIyAgwFalExHZBQZrIiICALi6ugIAtFotJk2ahMjISMTHx0OhUEjzc3JyoNVq4e7ubraeVqstc/tF75Zbk+5YWVPaWlPaCbCt1ZFc7WSwJiIiyfXr1zFx4kSEhYWhX79+WLZsmTRPp9PBw8MDbm5u0Ol0ZtOLBu2SFL1brr3fUVNONaWtNaWdANtaHZXVzvLeLZdXBSEiqgEMBkOZy9y6dQsRERGYPn06hgwZAgBo3rw5UlJSAADJyclo164dAgICcPz4cej1euTk5OD8+fPw9/e3aP1ERFUBgzURUTWVlJSExYsXAwDGjRuHzz77rNTl161bh+zsbKxZswbh4eEIDw9HZGQkVq1ahaFDh8JgMCAoKAheXl4IDw9HWFgYRo4cicmTJ8PR0dEKLSIism8cCkJEVE1t27YNiYmJAID169fjtddew4ABA0pcfvbs2Zg9e/Yj0zdv3vzItJCQEISEhMhWKxFRdcAeayKiakqpVEo9yQ4ODtKPEImIyDLYY01EVE397W9/Q1hYGAICApCWlobu3bvbuiQiomqNwZqIqJp64403EBgYiIyMDAwYMABNmza1dUlERNUah4IQEVUzn376KQBgxYoV+Pe//4309HTs3bsX7777ro0rIyKq3thjTURUzTz11FMAgIYNG5pN5xhrIiLLKleP9Zo1a8yer1ixwiLFEBHR4+vSpQsA4NSpUxg4cKD078iRIzaujIioeiu1x/rTTz/Fjh07cP78eSQnJwMAjEYjCgsLMXXqVKsUSEREFbNlyxasXbsW9+7dw3/+8x8AgBACjRs3tnFlRETVW6nBun///ujYsSPWr1+P8ePHA7h/+aYnnnjCKsUREVHFDR8+HMOHD8e6deukczcREVleqUNBNBoNvL298c477+D27du4du0arly5ghMnTlirPiIiqqRBgwbht99+Q0ZGBmJiYpCenm7rkoiIqrVy/Xhx0qRJuH37NurXrw/g/g9gnn/+eYsWRkREjycqKgrjxo3D1q1bERQUhEWLFiEhIcHWZRERVVvlCta3bt2SbotLRERVQ2FhIZ5//nmsW7cOffr0wdatW21dEhFRtVauq4L4+fnhxo0blq6FiIhkZDAYsGTJErRr1w7Hjh2D0Wi0dUlERNVauXqsjx8/jsDAQNSpU0ea9t1331msKCIienxxcXE4fPgwgoODsX//fixbtszWJRERVWvlCtYPLtdERET279SpU3juuedw5coV+Pr64vvvv4eHhwcuXryIBg0a2Lo8IqJqq1zBeubMmY9MW7JkiezFEBHR4zt69Ciee+45fPnll4/M69y5sw0qIiKqGcoVrHv37g3g/g0Gzpw5g5s3b1q0KCIiqrzXX38dANCmTRsEBwdL0//5z3/aqiQiohqhXMH6we1xAaBr166IiIiwWEFERPR4vvjiCxw8eBApKSk4duwYAMBkMuHcuXMYMWKEjasjIqq+yhWsi/5QMTMzE7du3bJYQURE9Hi6dOkCLy8vZGVlYejQoQDu3zWX46uJiCyrXMG66Dg9jUaDxYsXW6wgIiJ6PLVq1UKHDh3QoUMH3Lx5E4WFhRBC4Nq1a6hXr56tyyMiqrbKFayXLFmCc+fO4bfffoOfnx+aNWtm6bqIiOgxxcTEIDU1FXl5ecjLy4OPjw+SkpJsXRYRUbVVrhvEJCQkYM6cOfj5558xZ84cfPTRR5aui4iIHtPvv/+OL7/8Ep07d8bevXvh6Oho65KIiKq1cvVYf/HFF9iyZQvUajUMBgOGDRuG0aNHl7rOiRMnsHz5ciQkJODixYuIjo6GQqFAkyZNMG/ePCiV/8v0JpMJ8+fPxy+//AKNRoOFCxfC19f38VpGRFTDubq6QqFQIDc3F3Xq1IHBYLB1SURE1Vq5eqyFEFCr72dwBwcHODg4lLr8xo0bMXv2bOj1egD3h5JERkZi69atEELgwIEDZsvv378fBQUF2L59O6ZOnYq4uLjKtIWIiIpo0aIFPvroIzz55JOYPHkyCgsLbV0SEVG1Vq4e67Zt22LSpElo27Ytjh8/jtatW5e6vI+PD1atWoUZM2YAANLS0tC+fXsA9y/Xd/jwYfTs2VNa/vjx49Il/Vq1aoXTp0+Xq3iVSgFPT5dyLVtUfoERarWqxPlKpbJS25WDSmW7fZfFnmsD7Ls+e64NsO/67Lk2wL7rmzRpEvLz8+Hk5ITk5GQ899xzti6JiKhaKzNYb9++HVOmTMHhw4dx+vRptG/fHq+99lqp6wQFBeHKlSvScyEEFAoFgPtfTebk5Jgtr9Vq4ebmJj1XqVQoLCyUeslLYjQKZGXlltWERzi6OKKw0FjifJPJhKwsfYW3KwdPT5dKtcka7Lk2wL7rs+faAPuuz55rAypfn5eXuwWquS8zMxNarRZRUVFYunQphBDw9fXFhAkTsGPHDovtl4iopis1ua5atQq//vorXn31Vbz00kto3Lgx4uLicO/ePUycOLHcOyk6nlqn08HDw8NsvpubG3Q6nfTcZDKVGaqJiKh4J06cwD/+8Q9kZGRg7ty5EEJAqVTyduZERBZWanpNTk5GUlKS1Nvs7e2N9957D8OGDatQsG7evDlSUlLQoUMHJCcn44UXXjCb36ZNGxw6dAi9e/dGamoq/P39K9EUIiICgB49eqBHjx749ttv0a1bN1uXQ0RUY5T640UXFxcpVD/g4OAAV1fXCu0kKioKq1atwtChQ2EwGBAUFAQAmDFjBq5du4aePXtCo9Fg2LBhWLJkCWbOnFnBZhAR0cNcXFyQnJyMb7/9Fj169MC//vUvW5dERFStldpj7eTkhMuXL5vdBvfy5cuPhO3ieHt7Szci8PPzw+bNmx9ZZunSpdLjBQsWlLtoIiIq27Jly7B8+XK888472LZtGyIjI9GvXz9bl0VEVG2VGqynTZuGN954Ax07dkSDBg1w7do1fPfdd4iPj7dWfUREVEmOjo544oknoFar4eXlhYKCAluXRERUrZU6FKRJkybYunUrmjdvjry8PLRo0QLbtm1D8+bNrVUfERFVkpubG0aNGoVXXnkFW7ZsQf369W1dEhFRtVbmpTfc3d0xYMAAK5RCRERy+uCDD3Dp0iU0btwY586dQ3BwsK1LIiKq1sp150UiIqp6NBoNGjdujMjISPj7+0Oj0di6JCKiao3Bmoiomnn4Jly3b9+2USVERDULgzURUTUzfvx4AMC8efMAAM8880yF1j9x4gTCw8MBAGlpaejSpQvCw8MRHh6OvXv3AgCSkpIwaNAghISE4NChQ/IVT0RUhfH2hkRE1YyTkxMGDx6Mixcv4pdffoEQAkOHDoVCoUBiYmKp627cuBF79uyBs7MzAODMmTMYNWoUIiIipGUyMzORkJCAnTt3Qq/XIywsDJ06deJQEyKq8RisiYiqmY0bN+LmzZuYO3eu1GtdXj4+Pli1ahVmzJgBADh9+jQyMjJw4MAB+Pr6IiYmBidPnkTr1q2h0Wig0Wjg4+OD9PR0BAQEWKI5RERVBoM1EVE1o1Qq8dRTT2HDhg0VXjcoKAhXrlyRngcEBCA4OBgtW7bE2rVrsXr1ajRt2hTu7u7SMq6urtBqtWVuW6VSwNPT5c/HSulxdVdT2lpT2gmwrdWRXO1ksCYiohL17NkTHh4e0uPY2Fi0a9cOOp1OWkan05kF7ZIYjQJZWbkAAE9PF+lxdVdT2lpT2gmwrdVRWe308ir7HAfwx4tERFSK0aNH4+TJkwCAo0ePokWLFggICMDx48eh1+uRk5OD8+fPw9/f38aVEhHZHnusiYioRPPnz0dsbCwcHBxQt25dxMbGws3NDeHh4QgLC4MQApMnT4ajo6OtSyUisjkGayIiMuPt7Y2kpCQAQIsWLYq9kkhISAhCQkKsXRoRkV3jUBAiIiIiIhkwWBMRERERyYDBmoiIiIhIBgzWREREREQyYLAmIiIiIpIBgzURERERkQwYrImIiIiIZMBgTUREREQkAwZrIiIiIiIZMFgTEREREcmAwZqIiIiISAYM1kREREREMmCwJiIiIiKSAYM1EREREZEMGKyJiIiIiGTAYE1EREREJAMGayIiIiIiGaittaNdu3Zh9+7dAAC9Xo+zZ8/i8OHD8PDwAABs2rQJO3bsQJ06dQAA77zzDho2bGit8oiIiIiIHovVgvWgQYMwaNAgAPdD8+DBg6VQDQBpaWmIj49Hy5YtrVUSEREREZFsrD4U5NSpU/jtt98wdOhQs+lpaWnYsGEDQkNDsX79emuXRURERET0WKzWY/3A+vXrMXHixEem9+nTB2FhYXBzc8Obb76JQ4cOITAwsNRtqVQKeHq6VLiG/AIj1GpVifOVSmWltisHlcp2+y6LPdcG2Hd99lwbYN/12XNtgP3XR0RE1mPVYJ2dnY3ff/8dL7zwgtl0IQRGjhwJd3d3AEC3bt1w5syZMoO10SiQlZVb4TocXRxRWGgscb7JZEJWlr7C25WDp6dLpdpkDfZcG2Df9dlzbYB912fPtQGVr8/Ly90C1RARkS1ZdSjIDz/8gBdffPGR6VqtFn379oVOp4MQAikpKRxrTURERERVilV7rDMyMuDt7S09/9e//oXc3FwMHToUkydPxogRI6DRaNCxY0d069bNmqURERERET0WqwbrMWPGmD3v16+f9HjAgAEYMGCANcshIiIiIpINbxBDRERERCQDBmsiIiIiIhkwWBMRERERyYDBmoiIiIhIBgzWREREREQyYLAmIiIiIpIBgzURERERkQwYrImIiIiIZMBgTUREREQkAwZrIiIiIiIZMFgTEZGZEydOIDw8HABw8eJFhIaGIiwsDPPmzYPJZAIAJCUlYdCgQQgJCcGhQ4dsWS4Rkd1gsCYiIsnGjRsxe/Zs6PV6AMCSJUsQGRmJrVu3QgiBAwcOIDMzEwkJCUhMTMRHH32Ed999FwUFBTaunIjI9hisiYhI4uPjg1WrVknP09LS0L59ewBA165dceTIEZw8eRKtW7eGRqOBu7s7fHx8kJ6ebquSiYjshtrWBRARkf0ICgrClStXpOdCCCgUCgCAq6srcnJyoNVq4e7uLi3j6uoKrVZb5rZVKgU8PV3+fKyUHld3NaWtNaWdANtaHcnVTgZrIiIqkVL5vy82dTodPDw84ObmBp1OZza9aNAuidEokJWVCwDw9HSRHld3NaWtNaWdANtaHZXVTi+vss9xAIeCEBFRKZo3b46UlBQAQHJyMtq1a4eAgAAcP34cer0eOTk5OH/+PPz9/W1cKRGR7bHHmoiIShQVFYU5c+bg3XffRcOGDREUFASVSoXw8HCEhYVBCIHJkyfD0dHR1qUSEdkcgzUREZnx9vZGUlISAMDPzw+bN29+ZJmQkBCEhIRYuzQiIrvGoSBERERERDJgsCYiIiIikgGDNRERERGRDBisiYiIiIhkwGBNRERERCQDBmsiIiIiIhkwWBMRERERyYDBmoiIiIhIBgzWREREREQyYLAmIiIiIpIBgzURERERkQwYrImIiIiIZMBgTUREREQkA7U1dzZgwAC4u7sDALy9vbFkyRJp3sGDB7F69Wqo1WoMHjwYISEh1iyNiIiIiOixWC1Y6/V6AEBCQsIj8wwGA5YsWYIdO3bA2dkZoaGhCAwMhJeXl7XKIyIiIiJ6LFYbCpKeno68vDxERERgxIgRSE1NleadP38ePj4+qFWrFjQaDdq2bYsff/zRWqURERFRVaUAsvMMuJSpQ3Z+IaCwdUFUk1mtx9rJyQmjR49GcHAwLly4gLFjx+Krr76CWq2GVquVhogAgKurK7RabZnbVKkU8PR0qXAt+QVGqNWqEucrlcpKbVcOKpXt9l0We64NsO/67Lk2wL7rs+faAPuvj6haUwBnL93DyqRU6A1GODqoMCmkFZr51AKErYujmshqwdrPzw++vr5QKBTw8/ODp6cnMjMzUb9+fbi5uUGn00nL6nQ6s6BdEqNRICsrt8K1OLo4orDQWOJ8k8mErCx9hbcrB09Pl0q1yRrsuTbAvuuz59oA+67PnmsDKl+fl1fZ5zgiKl12rkEK1QCgNxixMikV8W+8CA9nBxtXRzWR1YaC7NixA3FxcQCAGzduQKvVSmOoGzVqhIsXLyIrKwsFBQX48ccf0bp1a2uVRkRERFVQlrZACtUP6A1GZOkKbFQR1XRW67EeMmQIZs6cidDQUCgUCixevBj//ve/kZubi6FDhyI6OhqjR4+GEAKDBw9GvXr1rFUaERERVUGe7o5wdFCZhWtHBxU8XTU2rIpqMqsFa41GgxUrVphNa9OmjfS4e/fu6N69u7XKISIioirOw1mNSSGtHhlj7eHiwDHWZBNWvY41ERERkWwE0MynFuLfeBFZugJ4umoYqsmmGKyJiIio6hKAh7PD/36syFBNNsRbmhMRERERyYDBmoiIiIhIBgzWREREREQyYLAmIiIiIpIBgzUREVFNpwCy8wy4lKlDdn4hoLB1QURVE68KQkREVJMpgLOX7j1yLehmPrV4hQ2iCmKPNRERUQ2WnWuQQjVw/5bgK5NSkZ1rsHFlZWAvO9kh9lgTERHVYFnaArNbggP3w3WWruB/14a2N+xlJzvFHmsiIqIazNPdEY4OKrNpjg4qeLpqbFRR2apsLztVewzWxVAplXBwcijxn6Oz/Z5siIiIKsLDWY1JIa2kcP2g99fDxU57q1F6LzuRLXEoSDFMQmDRxyklzp8V0cGK1RAREVmQAJr51EL8Gy8iS1cAT1fN/VBtx0MqHvSyFw3X9t7LTjUDgzUREZVpwIABcHd3BwB4e3tj/PjxiI6OhkKhQJMmTTBv3jwolfwStMoSgIezw//GVFckVCvuD824dvcO3JzV8HBWWzyUP+hlf3iMtb3/QUDVH4M1ERGVSq/XAwASEhKkaePHj0dkZCQ6dOiAuXPn4sCBA+jZs6etSiRbsdWPCKtgLzvVDOxeICKiUqWnpyMvLw8REREYMWIEUlNTkZaWhvbt2wMAunbtiiNHjti4SrIFm/6I8M9edp+6rvd72hmqyQ6wx5qIiErl5OSE0aNHIzg4GBcuXMDYsWMhhIBCcf/Cwa6ursjJySlzOyqVAp6eLn8+VkqPq7vq3NZrd+8U+yNCnd4In/q1bFSV5VXn1/RhNaWtcrWTwZqIiErl5+cHX19fKBQK+Pn5wdPTE2lpadJ8nU4HDw+PMrdjNApkZeUCADw9XaTH1V11bqubs7rYHxG6OqqqbZuB6v2aPqymtLWsdnp5uZdrOxwKQkREpdqxYwfi4uIAADdu3IBWq0WnTp2QknL/6knJyclo166dLUskG6mKl+ojsiT2WBMRUamGDBmCmTNnIjQ0FAqFAosXL0bt2rUxZ84cvPvuu2jYsCGCgoJsXSbZQpEfEer0Rrg6qvgjQqrRGKyJiKhUGo0GK1aseGT65s2bbVAN2Z0HPyKsX+v+V+kM1VSDcSgIEREREZEMGKyJiIiIiGTAYE1EREREJAMGayIiqh4UQHaeAZcydcjOLwQUti6IiGoa/niRiIiqPlvdWpuIqAj2WBMRUZVn01trVxfs8Sd6bOyxJiKiKi9LW1DsrbWzdAXwcObNSsrEHn8iWbDHmoiIqjxPd0fp7n8PODqo4OmqsVFFVQt7/InkwWBNRERVHm+t/XhK6/EnovLjUBAiIqr6itxaO0tXAE9XDW+tXQEPevyLhmv2+BNVnNWCtcFgQExMDK5evYqCggJMmDABf/vb36T5mzZtwo4dO1CnTh0AwDvvvIOGDRtaqzwiIqrq/ry1tjSmmqG63B70+D88xpp/nBBVjNWC9Z49e+Dp6Ylly5bh7t27GDhwoFmwTktLQ3x8PFq2bGmtkoiIiAhgjz+RTKwWrHv16oWgoCDpuUpl/iOTtLQ0bNiwAZmZmXjppZcwbtw4a5VGRERE7PEnemxWC9aurq4AAK1Wi0mTJiEyMtJsfp8+fRAWFgY3Nze8+eabOHToEAIDA0vdpkqlgKenS4VryS8wQq1WlbpMafOVSmWl9lseKpXltv247Lk2wL7rs+faAPuuz55rA+y/PiIish6r/njx+vXrmDhxIsLCwtCvXz9puhACI0eOhLu7OwCgW7duOHPmTJnB2mgUyMrKrXAdji6OKCw0lrpMafNNJhOysvQV3m95eHq6VKpN1mDPtQH2XZ891wbYd332XBtQ+fq8vNwtUA2RhSnuX5ovS1sAT3dHeDir2bNNVITVgvWtW7cQERGBuXPnomPHjmbztFot+vbti71798LFxQUpKSkYPHiwtUojIiKisvAmMkRlslqwXrduHbKzs7FmzRqsWbMGABAcHIy8vDwMHToUkydPxogRI6DRaNCxY0d069bNWqURERFRGUq6iUz8Gy/y7pZEf7JasJ49ezZmz55d4vwBAwZgwIAB1iqHiIiIKoC3jScqG++8SERERGXibeOJysZgTURERGXibeOJysZbmhMREVHZeBMZojIxWFuAo7MGJlHymUapUECfV2Cx9YmIqBx46biKs8VNZPg6URXCYG0BJiGw6OOUEufPiuhg0fWJiKgMvHRc1cDXiaoYjrEmIqIap6RLx2XnGmxcGRXF14mqGgZrIiKqcUq7dFyNogCy8wy4lKlDdn4hoLB1Qeb4OlFVw6EgRERU4zy4dFzR0FbjLh1X1jALOxjbzNeJqhr2WBMRUY3DS8eVMcziz9AdteYI5n+UgqjVh3H20j2r92jzdaKqhj3WRERU8/DScWUOs7CL25fzdaIqhsGaiIjsk6WHItji0nFF2XioRWnDLOzq9uW2fp2IKoDBmoiI7E95LrNWXDCFBcOqnEHYDi4j92CYxcM1eLg4AAoFxzYTVQKDNRER2Z2Sxv9KQxFKCKYatRLLt/4kb1hVANr8Qvx+PQdrd56UZdtlts8aShlmUWroZo8xUYn440UiIrI7ZY3/LSmYnr+WLe81j/8M8EdO35BCtRzbrvRl5OS+PN6fwyx86rreD/Tif9MfhO75Yzog/o0XeVMWonJgsCYiIrvzYPxvUUWHIpQUTE1CPDLtca55/CDAm4SQ9XrKZbWvWArg/LUcHDtzE2cu3sWxtBs4fy3HclfqKCl0W5KdX1ebqCwM1kREZHc8XNQYP+g5s8usjR/0HDxc7w+TKCmYKhWKR6YVG1bLGeCKBviSgnBlgmBlLiOnzS/ElUwtdn/zG5L2n8Pub37DlUwttPmF5dupvSvrEn8lvWYM42RHOMaaiIgqxWQyYf78+fjll1+g0WiwcOFC+Pr6yrLtbJ0BSfvPoX/XRveDkgCS9p+D31Nt/xznKzBpaCtcvqHF/h8uIkdnkMZYP/jRXYnjgv8McJu/OosurbyhVAJNfevAp54LYDRfztXZAY4OKhz88RKG9vTH9n3npG2/GfxXZN7Lx/mr2WY1lGvIRCUuI5edV4jEP/cP3O8xT9x3Dk18asPNser/d17S8J5lb3YCgOLHuPvWwtmLtv0RKFFRVf+TSERENrF//34UFBRg+/btSE1NRVxcHNauXSvLtrO0Bbh+OxdJB86ZTdfmG3D1Vq5ZkJowOAAN67vDzen+f2llhdXsXAM2f3UWPTv4mgXlCUMCEOBXGzDBLHw/CNR7j2Rg4EuN0aCeGzxcNFiz8wSu386Fo4MKQ3v6Y++RjIr9ALGCl5HL1xcWOxwlX189eqyLG97j7uqA36/n4I/budj9zW+PhO6F4zra/kegREVwKEglqJRKODg5lPhPoSj9e6jS1s8rMJa5fnk4OmtKrdHRmZdMIiqPsj5LBYUmW5doM8ePH0eXLl0AAK1atcLp06dl23ZJQz0cHdSPBKm1O0/CZBL3g2k5xgVnaQvQpZW3FKql7ew4idvZ5j+OvH47F3uPZKB/10bo8bwP2j7rhQZerli46Xtcv50rrbt93zl0b+vz2GO6S1O3llOxx6Suh6NF9mdtxb3mPZ73vf/6ljDG/XZ2vqxj34keF3usK8EkBBZ9nFLi/NmjX6j0+mq1CtEjn3+s+sraBwDMiujw2PsgqgnK+izNGdPRitXYF61WCzc3N+m5SqVCYWEh1Ori/2tRqRTw9HT587FSelwcD5NA5LDWeD/xZ6lHOXJYa+gNxmKDlE5vhE/9WuWqO98okH7pbrHbuZOjRyNvT1y7e0eafysrX+o5f65RXegNpmLXheJ+0PWq7WLWtrLaWl4lHRPveh5QKm0/sPhx21lc+3zquZmNcX/4utp1a7sUO/3h10Bucr2mVUFNaatc7WSwJiKiSnFzc4NOp5Oem0ymEkM1ABiNAllZ93t5PT1dpMcledbb45FhHdl5hcUGKVdHVZnbe8BJrUBT3zrFbqeOuyOysnLh5qwucT8l3TxFqVBgUkgrOKsVZrWUp63lVewxyc6TZduPS452Pty+B8e6uDHuk0JaoY5r8dfbfvg1kJucr6m9qyltLaudXl7u5doOh4IQEVGltGnTBsnJyQCA1NRU+Pv7y7uDYoZ1VOZqGsVt16eeCyYMCXjkqiNPeNwfJlfafoqbN2FwAF5sWc/yP5qzxSXwrOmh9j041jk6gzTGffprbf93XW0Tr7dN9oU91kREVCk9e/bE4cOHMWzYMAghsHjxYsvvtBJX0yiWEQjwq42F4zridnY+nvBwuh+qHwyZL2M/Jc5joJNXaa9DkZvZVORHoESWxGBNRESVolQqsWDBAuvvWK4gZQKecNPgCTeN9Lzc+2GYsx4ea6pCOBSEiIiIiEgGDNZERERERDJgsCYiIiIikgGDNRERERGRDBisiYiIiIhkwGBNRERERCQDBmsiIiIiIhlYLVibTCbMnTsXQ4cORXh4OC5evGg2/+DBgxg8eDCGDh2KpKQka5VFRERERCQLqwXr/fv3o6CgANu3b8fUqVMRFxcnzTMYDFiyZAk+/vhjJCQkYPv27cjMzLRWaUREREREj81qwfr48ePo0qULAKBVq1Y4ffq0NO/8+fPw8fFBrVq1oNFo0LZtW/z444/WKo2IiIiI6LEphBBWuTnorFmz8PLLL6Nbt24AgJdeegn79++HWq3Gjz/+iM2bN+P9998HAHzwwQd4+umnERwcbI3SiIiIiIgem9V6rN3c3KDT6aTnJpMJarW62Hk6nQ7u7u7WKo2IiIiI6LFZLVi3adMGycnJAIDU1FT4+/tL8xo1aoSLFy8iKysLBQUF+PHHH9G6dWtrlUZERERE9NisNhTEZDJh/vz5OHfuHIQQWLx4Mc6cOYPc3FwMHToUBw8exOrVqyGEwODBgzF8+HBrlEVEREREJAurBWsiIiIiouqMN4ghIiIiIpIBgzURERERkQzUti7AWh6M8f7ll1+g0WiwcOFC+Pr62rosDBgwQLoCire3N8aPH4/o6GgoFAo0adIE8+bNg1Jp3b9/Tpw4geXLlyMhIQEXL14stp6kpCQkJiZCrVZjwoQJCAwMtEl9aWlpGD9+PJ555hkAQGhoKHr37m31+gwGA2JiYnD16lUUFBRgwoQJaNy4sd0cu+Lqe+qpp+zi2BmNRsyePRsZGRlQqVRYsmQJhBB2c+yKqy8nJ8cujl1VZK/n4sdh759/ud2+fRuDBg3Cxx9/DLVaXW3buX79ehw8eBAGgwGhoaFo3759tWyrwWBAdHQ0rl69CqVSidjY2Gr3ulY21+Tn52P69Om4ffs2XF1dER8fjzp16pS+M1FDfP311yIqKkoIIcTPP/8sxo8fb+OKhMjPzxf9+/c3mzZu3Dhx7NgxIYQQc+bMEf/5z3+sWtOGDRtE3759RXBwcIn13Lx5U/Tt21fo9XqRnZ0tPbZFfUlJSeKjjz4yW8YW9e3YsUMsXLhQCCHEnTt3RLdu3ezq2BVXn70cu3379ono6GghhBDHjh0T48ePt6tjV1x99nLsqiJ7PBc/Lnv//MupoKBAvPHGG+Lll18Wv/32W7Vt57Fjx8S4ceOE0WgUWq1WrFy5stq2dd++fWLSpElCCCG+++478eabb1artj5Orvn444/FypUrhRBCfPHFFyI2NrbM/dWYoSCl3fnRVtLT05GXl4eIiAiMGDECqampSEtLQ/v27QEAXbt2xZEjR6xak4+PD1atWiU9L66ekydPonXr1tBoNHB3d4ePjw/S09NtUt/p06fxzTffYPjw4YiJiYFWq7VJfb169cLbb78tPVepVHZ17Iqrz16OXY8ePRAbGwsAuHbtGurWrWtXx664+uzl2FVF9nguflz2/vmXU3x8PIYNG4Ynn3wSgP39HyGX7777Dv7+/pg4cSLGjx+Pl156qdq21c/PD0ajESaTCVqtFmq1ulq19XFyTdHzVdeuXXH06NEy91djgrVWq4Wbm5v0XKVSobCw0IYVAU5OThg9ejQ++ugjvPPOO5g2bRqEEFAoFAAAV1dX5OTkWLWmoKAg6cY9AIqtR6vVmt3Ax9XVFVqt1ib1BQQEYMaMGdiyZQsaNGiA1atX26Q+V1dXuLm5QavVYtKkSYiMjLSrY1dcffZy7ABArVYjKioKsbGxCAoKsqtjV1x99nTsqhp7PBc/Lnv//Mtl165dqFOnjhQ0APv7P0Iud+/exenTp/HBBx+U+v9zdWiri4sLrl69ildeeQVz5sxBeHh4tWrr4+SaotPLm8lqTLAu7c6PtuLn54dXX30VCoUCfn5+8PT0xO3bt6X5Op0OHh4eNqwQZuO7H9RjT3fK7NmzJ1q2bCk9PnPmjM3qu379OkaMGIH+/fujX79+dnfsHq7Pno4dcL8n7Ouvv8acOXOg1+vNarD1sXu4vs6dO9vVsatK7PFcLAd7//zLYefOnThy5AjCw8Nx9uxZREVF4c6dO9L86tJOAPD09ETnzp2h0WjQsGFDODo6moWq6tTWTz75BJ07d8bXX3+Nzz//HNHR0TAYDNL86tRWoGK5puj08mayGhOsS7vzo63s2LEDcXFxAIAbN25Aq9WiU6dOSElJAQAkJyejXbt2tiwRzZs3f6SegIAAHD9+HHq9Hjk5OTh//rzNjufo0aNx8uRJAMDRo0fRokULm9R369YtREREYPr06RgyZAgA+zp2xdVnL8fus88+w/r16wEAzs7OUCgUaNmypd0cu+Lqe/PNN+3i2FVF9nguflz2/vmXy5YtW7B582YkJCSgWbNmiI+PR9euXatdOwGgbdu2+O9//wshBG7cuIG8vDx07NixWrbVw8NDCsi1atVCYWFhtXz/PlCRtrVp0wbffvuttGzbtm3L3H6NuUFMcXd+bNSokU1rKigowMyZM3Ht2jUoFApMmzYNtWvXxpw5c2AwGNCwYUMsXLgQKpXKqnVduXIFU6ZMQVJSEjIyMoqtJykpCdu3b4cQAuPGjUNQUJBN6ktLS0NsbCwcHBxQt25dxMbGws3Nzer1LVy4EP/+97/RsGFDadqsWbOwcOFCuzh2xdUXGRmJZcuW2fzY5ebmYubMmbh16xYKCwsxduxYNGrUyG7ed8XVV79+fbt431VF9nguflz2/vm3hPDwcMyfPx9KpdJuPqtyW7p0KVJSUiCEwOTJk+Ht7V0t26rT6RATE4PMzEwYDAaMGDECLVu2rFZtrWyuycvLQ1RUFDIzM+Hg4IAVK1bAy8ur1H3VmGBNRERERGRJNWYoCBERERGRJTFYExERERHJgMGaiIiIiEgGDNZERERERDJgsCYiIiIikgGDNVUbRqMRo0ePRmhoKO7du1fqsvv27cONGzdKnJ+VlYV//etfAIANGzZI1ywuzubNmwHcv8bl9u3bK1E5ERFZQ9FzO5ElMFhTtZGZmYm7d+9i27ZtqFWrVqnL/vOf/yz1Vqy//PILDh48CAB4/fXXERAQUOKya9euBQB07doVQ4cOrUTlRERkDUXP7USWUPXvI0v0pzlz5uDChQuYO3cuFixYAADQ6/V4++23odVqkZ+fj+nTpyMvL0+6He/WrVuxatUqnD59GjqdDo0aNcKSJUuwbt06pKenY/v27fj555/Ru3dvNGjQADNnzoRarYZKpcLSpUuxa9cu3Lt3D/Pnz0dAQAB+//13TJs2DWvWrMH+/fthNBoRGhqKYcOG2fjoEBFVLfn5+dJN1AwGA2JiYrB9+3ZcvnwZRqMRo0aNQu/evaWb1TRq1Ajbtm3DrVu3MHDgQEydOhVPPfUULl++jOeeew7vvPOO2bmdHSFkCeyxpmpj3rx5aNy4sRSqAeDSpUu4desW1q1bhxUrViA/Px8vvfSSdDvegoICeHh4YNOmTUhMTERqaipu3LiB8ePH44UXXjA78R45cgQtWrTApk2bMH78eNy7dw8TJkxArVq1MH/+fGm5M2fOIDk5GZ9++ikSExPx22+/gfdhIiKqmMTERPzlL3/B9u3bERcXh++//x61a9dGYmIiNm3ahPfffx937twpcf0LFy5g0aJF+PTTT5GcnIzMzMxiz+1EcmKPNVU7s2bNwqVLl1C7dm2sXLkSw4cPx5QpU1BYWIjw8HCzZR0dHXHnzh1MmTIFLi4uyM3NhcFgKHa7Q4YMwcaNGzFmzBi4u7tj8uTJxS6XkZGBgIAAqFQqODs7Y/bs2bK3kYiouvv999/RtWtXAIC/vz+2bduGF198EQDg5uaGRo0a4fLly2brFO3E8PHxgZubGwDAy8sLer3eSpVTTcYea6p2Fi1ahISEBKxcuRK//PILdDodNmzYgLi4OMTGxgIAFAoFhBBITk7G9evX8e6772LKlCnIz8+HEAJKpRImk8lsuwcOHEDbtm3xj3/8A7169cKHH34IAI/0Rjds2BBnzpyByWSCwWDAqFGjUFBQYJ3GExFVE40aNcKpU6cAAJcvX8aXX36JH3/8EQCg1Wpx7tw5eHt7Q6PRIDMzE8D9bwwfUCgUj2yzuHM7kZzYY03V2jPPPIPVq1fjs88+g4ODAyZNmgQAaN26NWbMmIG1a9dizZo1CAkJgUajQYMGDXDz5k34+Pjg3Llz+OSTT6RttWzZEtOnT8eqVaugVCoxc+ZMAPdP/tOmTZN6Upo1a4YuXbogNDQUJpMJoaGh0Gg0Vm87EVFVNmzYMMTExOC1116D0WjEhx9+iC1btiA0NBR6vR5vvvkmnnjiCYwYMQILFixA/fr18eSTT5a6zaLn9r///e/WaQjVKArBwZ9ERERERI+NQ0GIiIiIiGTAYE1EREREJAMGayIiIiIiGTBYExERERHJgMGaiIiIiEgGDNZERERERDJgsCYiIiIikgGDNRERERGRDP4fkQGh6RO9HLsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 864x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set({'figure.figsize': (12, 6)})\n",
    "\n",
    "fig, axarr = plt.subplots(1, 2)\n",
    "\n",
    "sns.histplot(dist_df, ax=axarr[0])\n",
    "axarr[0].set_xlabel('f-statistic')\n",
    "axarr[0].set_title(r'Gene {} (pancan $f$-statistic: {:.3e})'.format(\n",
    "    entrez_to_symbol[int(plot_gene)],\n",
    "    f_stats_df.loc[plot_gene, 'pancan']))\n",
    "\n",
    "sns.scatterplot(data=f_ss_df, x='count', y='f_statistic', ax=axarr[1])\n",
    "axarr[1].set_ylabel('f-statistic')\n",
    "axarr[1].set_title('Sample count vs. f-statistic, per cancer type')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "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>f_statistic</th>\n",
       "      <th>count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>BRCA</th>\n",
       "      <td>64.070699</td>\n",
       "      <td>981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UCEC</th>\n",
       "      <td>46.365732</td>\n",
       "      <td>507</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>LGG</th>\n",
       "      <td>32.065341</td>\n",
       "      <td>507</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>LUAD</th>\n",
       "      <td>21.327605</td>\n",
       "      <td>502</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>BLCA</th>\n",
       "      <td>17.633666</td>\n",
       "      <td>398</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>COAD</th>\n",
       "      <td>9.232498</td>\n",
       "      <td>341</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>OV</th>\n",
       "      <td>8.320908</td>\n",
       "      <td>176</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PRAD</th>\n",
       "      <td>7.205094</td>\n",
       "      <td>479</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>STAD</th>\n",
       "      <td>6.219720</td>\n",
       "      <td>383</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>KIRC</th>\n",
       "      <td>4.241350</td>\n",
       "      <td>352</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      f_statistic  count\n",
       "BRCA    64.070699    981\n",
       "UCEC    46.365732    507\n",
       "LGG     32.065341    507\n",
       "LUAD    21.327605    502\n",
       "BLCA    17.633666    398\n",
       "COAD     9.232498    341\n",
       "OV       8.320908    176\n",
       "PRAD     7.205094    479\n",
       "STAD     6.219720    383\n",
       "KIRC     4.241350    352"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# this is an example of less skewed distribution\n",
    "# (the cancer type with the top f-statistic, BRCA, has a much smaller LOF)\n",
    "plot_gene = sorted_genes[18]\n",
    "dist_df, f_ss_df = plot_f_dist(plot_gene)\n",
    "f_ss_df.sort_values(by='f_statistic', ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Sample count vs. f-statistic, per cancer type')"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set({'figure.figsize': (12, 6)})\n",
    "\n",
    "fig, axarr = plt.subplots(1, 2)\n",
    "\n",
    "sns.histplot(dist_df, ax=axarr[0])\n",
    "axarr[0].set_xlabel('f-statistic')\n",
    "axarr[0].set_title(r'Gene {} (pancan $f$-statistic: {:.3e})'.format(\n",
    "    entrez_to_symbol[int(plot_gene)],\n",
    "    f_stats_df.loc[plot_gene, 'pancan']))\n",
    "\n",
    "sns.scatterplot(data=f_ss_df, x='count', y='f_statistic', ax=axarr[1])\n",
    "axarr[1].set_ylabel('f-statistic')\n",
    "axarr[1].set_title('Sample count vs. f-statistic, per cancer type')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the plots above, each point is a cancer type. The left plot shows the distribution of f-statistics (higher = better correlation with labels) across cancer types, and the right plot shows f-statistic vs. sample size.\n",
    "\n",
    "For TP53, we can see that there might be a slight correlation with sample size (BRCA, the largest/best sampled cancer type, tends to have higher f-statistics for most genes) but this isn't necessarily the case for other genes. It's possible/likely that expression is generally just very predictive of TP53 mutation status in many genes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save univariate correlation results\n",
    "output_dir = cfg.data_dir / 'univariate_corrs'\n",
    "output_dir.mkdir(exist_ok=True)\n",
    "\n",
    "output_file = output_dir / '{}_{}_corrs.tsv'.format(gene, mad_threshold)\n",
    "rank_df.to_csv(output_file, sep='\\t')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:pancancer-evaluation]",
   "language": "python",
   "name": "conda-env-pancancer-evaluation-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}