PRADA.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# PRADA"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prioritization of Regulatory Pathways based on Analysis of RNA Dynamics Alterations"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Dysregulation of RNA stability plays an important role in cancer progression. Key regulators of RNA turnover, such as miRNAs and RNA-binding proteins, have been implicated in a variety of cancers - however, the list of annotated regulatory programs that govern the RNA lifecycle remains incomplete. The development of analytical frameworks for systematic discovery of post-transcriptional regulators is critical for a better understanding of regulatory networks that impact disease progression. For this purpose, we have developed a computational framework, named PRADA, to identify RNA-binding proteins that underlie pathologic gene expression modulations. Using this approach, we uncovered the RNA-binding protein RBMS1 as a novel suppressor of colon cancer progression. Our findings indicate that silencing RBMS1, which is achieved through epigenetic reprogramming, results in increased metastatic capacity in colon cancer cells. Restoring RBMS1 expression, in turn, blunts metastatic capacity. We have shown that RBMS1 functions as a post-transcriptional regulator of RNA stability by binding and stabilizing ~80 target mRNAs. Importantly, our analyses of colon cancer datasets as well as measurements in clinical samples have shown that RBMS1 silencing is associated with disease progression and poor survival. Our findings establish a previously unknown role for RBMS1 in mammalian gene expression regulation and its role in colon cancer metastasis."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1. Create and clean up input file (log-fold change)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this study, we are starting with Illumina arrays from GSE59857, which compares poorly and highly metastatic colon cancer cell lines."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys\n",
    "import os\n",
    "import pandas as pd\n",
    "import re\n",
    "import numpy as np\n",
    "import scipy as sp\n",
    "from collections import defaultdict\n",
    "from itertools import islice\n",
    "os.environ['KMP_DUPLICATE_LIB_OK']='True'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>RefSeq</th>\n",
       "      <th>CACO2</th>\n",
       "      <th>COLO201</th>\n",
       "      <th>LS123</th>\n",
       "      <th>SW480</th>\n",
       "      <th>SW1417</th>\n",
       "      <th>LS174T</th>\n",
       "      <th>COLO320</th>\n",
       "      <th>HCT116</th>\n",
       "      <th>HT29</th>\n",
       "      <th>WIDR</th>\n",
       "      <th>LOVO</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Probe</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>ILMN_3245919</td>\n",
       "      <td>XM_001714734</td>\n",
       "      <td>151.323411</td>\n",
       "      <td>148.142560</td>\n",
       "      <td>171.410158</td>\n",
       "      <td>167.456323</td>\n",
       "      <td>183.939424</td>\n",
       "      <td>41923.971670</td>\n",
       "      <td>146.239302</td>\n",
       "      <td>177.357721</td>\n",
       "      <td>176.733719</td>\n",
       "      <td>164.927448</td>\n",
       "      <td>158.597969</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1672148</td>\n",
       "      <td>NM_020299</td>\n",
       "      <td>146.968324</td>\n",
       "      <td>192.024693</td>\n",
       "      <td>123.071839</td>\n",
       "      <td>133.238433</td>\n",
       "      <td>140.839404</td>\n",
       "      <td>2105.803564</td>\n",
       "      <td>124.424634</td>\n",
       "      <td>302.047197</td>\n",
       "      <td>2930.576970</td>\n",
       "      <td>12003.061920</td>\n",
       "      <td>12739.901670</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1685387</td>\n",
       "      <td>NM_002644</td>\n",
       "      <td>161.659510</td>\n",
       "      <td>143.659146</td>\n",
       "      <td>167.381620</td>\n",
       "      <td>159.185124</td>\n",
       "      <td>129.949763</td>\n",
       "      <td>26836.516690</td>\n",
       "      <td>143.809830</td>\n",
       "      <td>144.222191</td>\n",
       "      <td>153.363798</td>\n",
       "      <td>207.071610</td>\n",
       "      <td>427.244406</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1720998</td>\n",
       "      <td>NM_001218</td>\n",
       "      <td>205.546207</td>\n",
       "      <td>173.576579</td>\n",
       "      <td>171.661433</td>\n",
       "      <td>126.627450</td>\n",
       "      <td>198.623380</td>\n",
       "      <td>10232.277200</td>\n",
       "      <td>281.566402</td>\n",
       "      <td>318.708726</td>\n",
       "      <td>3647.612966</td>\n",
       "      <td>12367.323190</td>\n",
       "      <td>1912.260020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1666536</td>\n",
       "      <td>NM_014312</td>\n",
       "      <td>196.454316</td>\n",
       "      <td>169.608556</td>\n",
       "      <td>225.914542</td>\n",
       "      <td>173.089286</td>\n",
       "      <td>184.556284</td>\n",
       "      <td>387.418966</td>\n",
       "      <td>165.364108</td>\n",
       "      <td>213.687788</td>\n",
       "      <td>7966.161694</td>\n",
       "      <td>20813.031510</td>\n",
       "      <td>207.675049</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    RefSeq       CACO2     COLO201       LS123       SW480  \\\n",
       "Probe                                                                        \n",
       "ILMN_3245919  XM_001714734  151.323411  148.142560  171.410158  167.456323   \n",
       "ILMN_1672148     NM_020299  146.968324  192.024693  123.071839  133.238433   \n",
       "ILMN_1685387     NM_002644  161.659510  143.659146  167.381620  159.185124   \n",
       "ILMN_1720998     NM_001218  205.546207  173.576579  171.661433  126.627450   \n",
       "ILMN_1666536     NM_014312  196.454316  169.608556  225.914542  173.089286   \n",
       "\n",
       "                  SW1417        LS174T     COLO320      HCT116         HT29  \\\n",
       "Probe                                                                         \n",
       "ILMN_3245919  183.939424  41923.971670  146.239302  177.357721   176.733719   \n",
       "ILMN_1672148  140.839404   2105.803564  124.424634  302.047197  2930.576970   \n",
       "ILMN_1685387  129.949763  26836.516690  143.809830  144.222191   153.363798   \n",
       "ILMN_1720998  198.623380  10232.277200  281.566402  318.708726  3647.612966   \n",
       "ILMN_1666536  184.556284    387.418966  165.364108  213.687788  7966.161694   \n",
       "\n",
       "                      WIDR          LOVO  \n",
       "Probe                                     \n",
       "ILMN_3245919    164.927448    158.597969  \n",
       "ILMN_1672148  12003.061920  12739.901670  \n",
       "ILMN_1685387    207.071610    427.244406  \n",
       "ILMN_1720998  12367.323190   1912.260020  \n",
       "ILMN_1666536  20813.031510    207.675049  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "expfile = 'input/high-vs-low_metastatic_lines_GSE59857.txt'\n",
    "exp = pd.read_csv(expfile, sep='\\t', header=0, index_col=0)\n",
    "exp.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>logFC</th>\n",
       "      <th>pval</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Probe</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>ILMN_3245919</td>\n",
       "      <td>5.437057</td>\n",
       "      <td>0.363198</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1672148</td>\n",
       "      <td>5.095663</td>\n",
       "      <td>0.093406</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1685387</td>\n",
       "      <td>4.932240</td>\n",
       "      <td>0.357067</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1720998</td>\n",
       "      <td>4.773884</td>\n",
       "      <td>0.083028</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1666536</td>\n",
       "      <td>4.706519</td>\n",
       "      <td>0.221115</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 logFC      pval\n",
       "Probe                           \n",
       "ILMN_3245919  5.437057  0.363198\n",
       "ILMN_1672148  5.095663  0.093406\n",
       "ILMN_1685387  4.932240  0.357067\n",
       "ILMN_1720998  4.773884  0.083028\n",
       "ILMN_1666536  4.706519  0.221115"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from scipy.stats import ttest_ind\n",
    "#Poorly metastatic: CACO2,COLO201,LS123,SW480,SW1417\n",
    "#Highly metastatic: LS174T,COLO320,HCT116,HT29,WIDR,LOVO\n",
    "logFC = pd.DataFrame(np.log2(exp.iloc[:,6:12].mean(axis=1) / exp.iloc[:,1:6].mean(axis=1)), columns=['logFC'])\n",
    "logFC['pval'] = exp.apply(lambda x: ttest_ind(x[1:6], x[6:12], equal_var=False)[1], axis=1)\n",
    "logFC.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>logFC</th>\n",
       "      <th>pval</th>\n",
       "      <th>RefSeq</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Probe</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>ILMN_3245919</td>\n",
       "      <td>5.437057</td>\n",
       "      <td>0.363198</td>\n",
       "      <td>XM_001714734</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1672148</td>\n",
       "      <td>5.095663</td>\n",
       "      <td>0.093406</td>\n",
       "      <td>NM_020299</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1685387</td>\n",
       "      <td>4.932240</td>\n",
       "      <td>0.357067</td>\n",
       "      <td>NM_002644</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1720998</td>\n",
       "      <td>4.773884</td>\n",
       "      <td>0.083028</td>\n",
       "      <td>NM_001218</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1666536</td>\n",
       "      <td>4.706519</td>\n",
       "      <td>0.221115</td>\n",
       "      <td>NM_014312</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 logFC      pval        RefSeq\n",
       "Probe                                         \n",
       "ILMN_3245919  5.437057  0.363198  XM_001714734\n",
       "ILMN_1672148  5.095663  0.093406     NM_020299\n",
       "ILMN_1685387  4.932240  0.357067     NM_002644\n",
       "ILMN_1720998  4.773884  0.083028     NM_001218\n",
       "ILMN_1666536  4.706519  0.221115     NM_014312"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "logFC['RefSeq'] = exp['RefSeq']\n",
    "logFC.head()"
   ]
  },
  {
   "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>logFC</th>\n",
       "      <th>pval</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>RefSeq</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>NM_000014</td>\n",
       "      <td>-0.145564</td>\n",
       "      <td>0.121230</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_000015</td>\n",
       "      <td>0.766022</td>\n",
       "      <td>0.323496</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_000016</td>\n",
       "      <td>-0.544629</td>\n",
       "      <td>0.281687</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_000017</td>\n",
       "      <td>1.211667</td>\n",
       "      <td>0.057775</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_000018</td>\n",
       "      <td>0.413733</td>\n",
       "      <td>0.294752</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              logFC      pval\n",
       "RefSeq                       \n",
       "NM_000014 -0.145564  0.121230\n",
       "NM_000015  0.766022  0.323496\n",
       "NM_000016 -0.544629  0.281687\n",
       "NM_000017  1.211667  0.057775\n",
       "NM_000018  0.413733  0.294752"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "logFC_r = logFC.groupby(['RefSeq']).agg(np.mean)\n",
    "logFC_r.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "logFC_r.to_csv('input/high-vs-low_metastatic_lines_GSE59857_logFC_refseq.txt', sep='\\t', index=True, index_label='RefSeq')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>RefSeq</th>\n",
       "      <th>CACO2</th>\n",
       "      <th>COLO201</th>\n",
       "      <th>LS123</th>\n",
       "      <th>SW480</th>\n",
       "      <th>SW1417</th>\n",
       "      <th>LS174T</th>\n",
       "      <th>COLO320</th>\n",
       "      <th>HCT116</th>\n",
       "      <th>HT29</th>\n",
       "      <th>WIDR</th>\n",
       "      <th>LOVO</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Probe</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>ILMN_3245919</td>\n",
       "      <td>XM_001714734</td>\n",
       "      <td>151.323411</td>\n",
       "      <td>148.142560</td>\n",
       "      <td>171.410158</td>\n",
       "      <td>167.456323</td>\n",
       "      <td>183.939424</td>\n",
       "      <td>41923.971670</td>\n",
       "      <td>146.239302</td>\n",
       "      <td>177.357721</td>\n",
       "      <td>176.733719</td>\n",
       "      <td>164.927448</td>\n",
       "      <td>158.597969</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1672148</td>\n",
       "      <td>NM_020299</td>\n",
       "      <td>146.968324</td>\n",
       "      <td>192.024693</td>\n",
       "      <td>123.071839</td>\n",
       "      <td>133.238433</td>\n",
       "      <td>140.839404</td>\n",
       "      <td>2105.803564</td>\n",
       "      <td>124.424634</td>\n",
       "      <td>302.047197</td>\n",
       "      <td>2930.576970</td>\n",
       "      <td>12003.061920</td>\n",
       "      <td>12739.901670</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1685387</td>\n",
       "      <td>NM_002644</td>\n",
       "      <td>161.659510</td>\n",
       "      <td>143.659146</td>\n",
       "      <td>167.381620</td>\n",
       "      <td>159.185124</td>\n",
       "      <td>129.949763</td>\n",
       "      <td>26836.516690</td>\n",
       "      <td>143.809830</td>\n",
       "      <td>144.222191</td>\n",
       "      <td>153.363798</td>\n",
       "      <td>207.071610</td>\n",
       "      <td>427.244406</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1720998</td>\n",
       "      <td>NM_001218</td>\n",
       "      <td>205.546207</td>\n",
       "      <td>173.576579</td>\n",
       "      <td>171.661433</td>\n",
       "      <td>126.627450</td>\n",
       "      <td>198.623380</td>\n",
       "      <td>10232.277200</td>\n",
       "      <td>281.566402</td>\n",
       "      <td>318.708726</td>\n",
       "      <td>3647.612966</td>\n",
       "      <td>12367.323190</td>\n",
       "      <td>1912.260020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ILMN_1666536</td>\n",
       "      <td>NM_014312</td>\n",
       "      <td>196.454316</td>\n",
       "      <td>169.608556</td>\n",
       "      <td>225.914542</td>\n",
       "      <td>173.089286</td>\n",
       "      <td>184.556284</td>\n",
       "      <td>387.418966</td>\n",
       "      <td>165.364108</td>\n",
       "      <td>213.687788</td>\n",
       "      <td>7966.161694</td>\n",
       "      <td>20813.031510</td>\n",
       "      <td>207.675049</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    RefSeq       CACO2     COLO201       LS123       SW480  \\\n",
       "Probe                                                                        \n",
       "ILMN_3245919  XM_001714734  151.323411  148.142560  171.410158  167.456323   \n",
       "ILMN_1672148     NM_020299  146.968324  192.024693  123.071839  133.238433   \n",
       "ILMN_1685387     NM_002644  161.659510  143.659146  167.381620  159.185124   \n",
       "ILMN_1720998     NM_001218  205.546207  173.576579  171.661433  126.627450   \n",
       "ILMN_1666536     NM_014312  196.454316  169.608556  225.914542  173.089286   \n",
       "\n",
       "                  SW1417        LS174T     COLO320      HCT116         HT29  \\\n",
       "Probe                                                                         \n",
       "ILMN_3245919  183.939424  41923.971670  146.239302  177.357721   176.733719   \n",
       "ILMN_1672148  140.839404   2105.803564  124.424634  302.047197  2930.576970   \n",
       "ILMN_1685387  129.949763  26836.516690  143.809830  144.222191   153.363798   \n",
       "ILMN_1720998  198.623380  10232.277200  281.566402  318.708726  3647.612966   \n",
       "ILMN_1666536  184.556284    387.418966  165.364108  213.687788  7966.161694   \n",
       "\n",
       "                      WIDR          LOVO  \n",
       "Probe                                     \n",
       "ILMN_3245919    164.927448    158.597969  \n",
       "ILMN_1672148  12003.061920  12739.901670  \n",
       "ILMN_1685387    207.071610    427.244406  \n",
       "ILMN_1720998  12367.323190   1912.260020  \n",
       "ILMN_1666536  20813.031510    207.675049  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "exp_r = exp\n",
    "exp_r.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. Generating an RBP-target matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this section, we will create a binary matrix where the rows are transcripts and the columns are RBPs. If the transcript(i) contains a putative binding site for RBP(j), the element(i,j) will be set to '1', otherwise, it will remain '0'."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#RNA Dynamics file: a 3-column tab-delimited file with RefSeq IDs and changes in RNA dynamics (in this case expression)\n",
    "#and their associated p-values\n",
    "RDfile = 'input/high-vs-low_metastatic_lines_GSE59857_logFC_refseq.txt'"
   ]
  },
  {
   "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>RBP</th>\n",
       "      <th>motif</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>RBP</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>MATR3</td>\n",
       "      <td>MATR3</td>\n",
       "      <td>[AC]ATCTT[AG]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>ENOX1</td>\n",
       "      <td>ENOX1</td>\n",
       "      <td>[ACT][AG][TG]ACAG</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>PTBP1</td>\n",
       "      <td>PTBP1</td>\n",
       "      <td>[ACT][TC]TTT[TC]T</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>RBMS3</td>\n",
       "      <td>RBMS3</td>\n",
       "      <td>[ACT]ATATA</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>RBM6</td>\n",
       "      <td>RBM6</td>\n",
       "      <td>[ACT]ATCCA[AG]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         RBP              motif\n",
       "RBP                            \n",
       "MATR3  MATR3      [AC]ATCTT[AG]\n",
       "ENOX1  ENOX1  [ACT][AG][TG]ACAG\n",
       "PTBP1  PTBP1  [ACT][TC]TTT[TC]T\n",
       "RBMS3  RBMS3         [ACT]ATATA\n",
       "RBM6    RBM6     [ACT]ATCCA[AG]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "motifs = pd.read_csv('data/motif_RBP_map.txt', sep='\\t', header=0)\n",
    "motifs.set_index('RBP', inplace=True, drop=False)\n",
    "motifs.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "#write motifs to file\n",
    "motifs['motif'].unique().tofile('outputs/motifs.txt', sep=\"\\n\", format=\"%s\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gunzip data/hg19_mrna.fa.gz\n",
      "perl programs/scan_fasta_for_regex_matches.pl outputs/motifs.txt data/hg19_mrna.fa 1 > outputs/motifs-mrna.out\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#match motifs to fasta file\n",
    "#needs genregexp\n",
    "import subprocess\n",
    "cmd = \"gunzip data/hg19_mrna.fa.gz\"\n",
    "print(cmd)\n",
    "subprocess.call(cmd,shell=True)\n",
    "cmd = \"perl programs/scan_fasta_for_regex_matches.pl outputs/motifs.txt data/hg19_mrna.fa 1 > outputs/motifs-mrna.out\"\n",
    "print(cmd)\n",
    "subprocess.call(cmd,shell=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "rmap = defaultdict(dict)\n",
    "motif=\"\"\n",
    "refs = {}\n",
    "with open(\"outputs/motifs-mrna.out\", \"rt\") as f:\n",
    "    for l in f:\n",
    "        if l.startswith('>'):\n",
    "            l=re.sub('\\s+$','',l)\n",
    "            motif = l[1:]\n",
    "            continue\n",
    "        a = l.split('\\t')\n",
    "        rmap[motif][a[0]]=1\n",
    "        refs[a[0]] = 1\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "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>logFC</th>\n",
       "      <th>pval</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>RefSeq</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>NM_000014</td>\n",
       "      <td>-0.145564</td>\n",
       "      <td>0.121230</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_000015</td>\n",
       "      <td>0.766022</td>\n",
       "      <td>0.323496</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_000016</td>\n",
       "      <td>-0.544629</td>\n",
       "      <td>0.281687</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_000017</td>\n",
       "      <td>1.211667</td>\n",
       "      <td>0.057775</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_000018</td>\n",
       "      <td>0.413733</td>\n",
       "      <td>0.294752</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              logFC      pval\n",
       "RefSeq                       \n",
       "NM_000014 -0.145564  0.121230\n",
       "NM_000015  0.766022  0.323496\n",
       "NM_000016 -0.544629  0.281687\n",
       "NM_000017  1.211667  0.057775\n",
       "NM_000018  0.413733  0.294752"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "exp = pd.read_csv(RDfile, sep='\\t', header=0, index_col=0)\n",
    "exp.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "mat = pd.DataFrame(0, index=list(set(exp.index) & set(refs.keys())), columns=motifs['RBP'])"
   ]
  },
  {
   "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>RBP</th>\n",
       "      <th>MATR3</th>\n",
       "      <th>ENOX1</th>\n",
       "      <th>PTBP1</th>\n",
       "      <th>RBMS3</th>\n",
       "      <th>RBM6</th>\n",
       "      <th>LIN28A</th>\n",
       "      <th>HNRNPC</th>\n",
       "      <th>HNRNPCL1</th>\n",
       "      <th>SNRNP70</th>\n",
       "      <th>RBM8A</th>\n",
       "      <th>...</th>\n",
       "      <th>SRSF2</th>\n",
       "      <th>HNRNPH2</th>\n",
       "      <th>DAZAP1</th>\n",
       "      <th>MSI1</th>\n",
       "      <th>ESRP2</th>\n",
       "      <th>ZC3H14</th>\n",
       "      <th>TIA1</th>\n",
       "      <th>U2AF2</th>\n",
       "      <th>CPEB4</th>\n",
       "      <th>RALY</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>NM_173362</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_006196</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_016579</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_001700</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_001920</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 75 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "RBP        MATR3  ENOX1  PTBP1  RBMS3  RBM6  LIN28A  HNRNPC  HNRNPCL1  \\\n",
       "NM_173362      1      1      1      1     0       1       1         1   \n",
       "NM_006196      0      0      1      0     0       0       1         1   \n",
       "NM_016579      0      1      0      0     0       0       0         0   \n",
       "NM_001700      0      0      0      0     0       0       0         0   \n",
       "NM_001920      1      1      1      1     1       1       1         1   \n",
       "\n",
       "RBP        SNRNP70  RBM8A  ...  SRSF2  HNRNPH2  DAZAP1  MSI1  ESRP2  ZC3H14  \\\n",
       "NM_173362        1      1  ...      1        0       1     1      0       1   \n",
       "NM_006196        1      1  ...      1        0       1     0      1       1   \n",
       "NM_016579        0      1  ...      1        0       1     0      1       0   \n",
       "NM_001700        1      0  ...      1        0       0     0      0       0   \n",
       "NM_001920        1      1  ...      1        0       1     1      0       0   \n",
       "\n",
       "RBP        TIA1  U2AF2  CPEB4  RALY  \n",
       "NM_173362     1      0      1     1  \n",
       "NM_006196     1      0      1     1  \n",
       "NM_016579     0      0      0     0  \n",
       "NM_001700     0      0      0     0  \n",
       "NM_001920     0      0      1     0  \n",
       "\n",
       "[5 rows x 75 columns]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "for rbp in mat.columns:\n",
    "    m = motifs.loc[rbp,'motif']\n",
    "    #print(m)\n",
    "    #sys.stdout.flush()\n",
    "    ref = rmap[m].keys()\n",
    "    ref = list(set(rmap[m].keys()) & set(mat.index))\n",
    "    mat.loc[ref,rbp] = 1\n",
    "mat.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "mat.to_csv('outputs/RBP-v-RefSeq_target_matrix.txt', sep='\\t', index=True, index_label='RefSeq')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3. Generating the proper matrices"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The general form of the model:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$\\Delta Exp(g)=\\sum_{i} \\alpha_{i} \\cdot t_{i,g} \\cdot \\Delta Exp(RBP_{i}) + c_{g}$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Contraints:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$min\\ \\frac{1}{2n} \\left [ \\left \\| \\alpha X - Exp \\right \\|+ \\lambda \\sum_{i}\\frac{\\left | \\alpha_{i} \\right |}{\\left | \\Delta Exp\\left ( RBP_{i} \\right )\\right |} \\right]$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "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>HGNC</th>\n",
       "      <th>RefSeq</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1/2-SBSRNA4</td>\n",
       "      <td>NR_039978</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>A1BG</td>\n",
       "      <td>NM_130786</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>A1BG-AS1</td>\n",
       "      <td>NR_015380</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>A1CF</td>\n",
       "      <td>NM_001198818</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>A1CF</td>\n",
       "      <td>NM_001198819</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          HGNC        RefSeq\n",
       "0  1/2-SBSRNA4     NR_039978\n",
       "1         A1BG     NM_130786\n",
       "2     A1BG-AS1     NR_015380\n",
       "3         A1CF  NM_001198818\n",
       "4         A1CF  NM_001198819"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hgnc_to_ref = pd.read_csv('data/hg19_genes_vs_refseq.txt', sep='\\t', header=0)\n",
    "hgnc_to_ref.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "hgnc_to_ref = hgnc_to_ref.groupby('HGNC')['RefSeq'].apply(lambda x: \"%s\" % ','.join(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "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>MATR3</th>\n",
       "      <th>ENOX1</th>\n",
       "      <th>PTBP1</th>\n",
       "      <th>RBMS3</th>\n",
       "      <th>RBM6</th>\n",
       "      <th>LIN28A</th>\n",
       "      <th>HNRNPC</th>\n",
       "      <th>HNRNPCL1</th>\n",
       "      <th>SNRNP70</th>\n",
       "      <th>RBM8A</th>\n",
       "      <th>...</th>\n",
       "      <th>SRSF2</th>\n",
       "      <th>HNRNPH2</th>\n",
       "      <th>DAZAP1</th>\n",
       "      <th>MSI1</th>\n",
       "      <th>ESRP2</th>\n",
       "      <th>ZC3H14</th>\n",
       "      <th>TIA1</th>\n",
       "      <th>U2AF2</th>\n",
       "      <th>CPEB4</th>\n",
       "      <th>RALY</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>RefSeq</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>NM_173362</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_006196</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_016579</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_001700</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_001920</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 75 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           MATR3  ENOX1  PTBP1  RBMS3  RBM6  LIN28A  HNRNPC  HNRNPCL1  \\\n",
       "RefSeq                                                                  \n",
       "NM_173362      1      1      1      1     0       1       1         1   \n",
       "NM_006196      0      0      1      0     0       0       1         1   \n",
       "NM_016579      0      1      0      0     0       0       0         0   \n",
       "NM_001700      0      0      0      0     0       0       0         0   \n",
       "NM_001920      1      1      1      1     1       1       1         1   \n",
       "\n",
       "           SNRNP70  RBM8A  ...  SRSF2  HNRNPH2  DAZAP1  MSI1  ESRP2  ZC3H14  \\\n",
       "RefSeq                     ...                                                \n",
       "NM_173362        1      1  ...      1        0       1     1      0       1   \n",
       "NM_006196        1      1  ...      1        0       1     0      1       1   \n",
       "NM_016579        0      1  ...      1        0       1     0      1       0   \n",
       "NM_001700        1      0  ...      1        0       0     0      0       0   \n",
       "NM_001920        1      1  ...      1        0       1     1      0       0   \n",
       "\n",
       "           TIA1  U2AF2  CPEB4  RALY  \n",
       "RefSeq                               \n",
       "NM_173362     1      0      1     1  \n",
       "NM_006196     1      0      1     1  \n",
       "NM_016579     0      0      0     0  \n",
       "NM_001700     0      0      0     0  \n",
       "NM_001920     0      0      1     0  \n",
       "\n",
       "[5 rows x 75 columns]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mat = pd.read_csv('outputs/RBP-v-RefSeq_target_matrix.txt', sep='\\t', header=0, index_col=0)\n",
    "mat.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "for rbp in mat.columns:\n",
    "    #print(rbp)\n",
    "    rbp_refs = hgnc_to_ref[rbp].split(',')\n",
    "    rbp_sum = 0\n",
    "    rbp_cnt = 0\n",
    "    rbp_max = 0\n",
    "    for r in rbp_refs:\n",
    "        if r in exp.index:\n",
    "            if (abs(exp.loc[r,'logFC']) > rbp_max):\n",
    "                motifs.loc[rbp,'diff'] = exp.loc[r,'logFC']\n",
    "                motifs.loc[rbp,'pval'] = exp.loc[r,'pval']\n",
    "                rbp_max = abs(exp.loc[r,'logFC'])\n",
    "\n",
    "motifs.to_csv('outputs/RBP_motif_diff.txt', sep='\\t', index=True, index_label='RefSeq')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>MATR3</th>\n",
       "      <th>ENOX1</th>\n",
       "      <th>PTBP1</th>\n",
       "      <th>RBMS3</th>\n",
       "      <th>RBM6</th>\n",
       "      <th>LIN28A</th>\n",
       "      <th>HNRNPC</th>\n",
       "      <th>HNRNPCL1</th>\n",
       "      <th>SNRNP70</th>\n",
       "      <th>RBM8A</th>\n",
       "      <th>...</th>\n",
       "      <th>SRSF2</th>\n",
       "      <th>HNRNPH2</th>\n",
       "      <th>DAZAP1</th>\n",
       "      <th>MSI1</th>\n",
       "      <th>ESRP2</th>\n",
       "      <th>ZC3H14</th>\n",
       "      <th>TIA1</th>\n",
       "      <th>U2AF2</th>\n",
       "      <th>CPEB4</th>\n",
       "      <th>RALY</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>RefSeq</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>NM_173362</td>\n",
       "      <td>-1.129317</td>\n",
       "      <td>0.05266</td>\n",
       "      <td>0.229692</td>\n",
       "      <td>0.079549</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>-0.055535</td>\n",
       "      <td>0.79533</td>\n",
       "      <td>0.161332</td>\n",
       "      <td>0.462524</td>\n",
       "      <td>-0.017537</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.638921</td>\n",
       "      <td>-0.0</td>\n",
       "      <td>0.368176</td>\n",
       "      <td>0.623895</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.120822</td>\n",
       "      <td>-1.240809</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-1.366578</td>\n",
       "      <td>-0.146491</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_006196</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.229692</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>0.79533</td>\n",
       "      <td>0.161332</td>\n",
       "      <td>0.462524</td>\n",
       "      <td>-0.017537</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.638921</td>\n",
       "      <td>-0.0</td>\n",
       "      <td>0.368176</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.027613</td>\n",
       "      <td>0.120822</td>\n",
       "      <td>-1.240809</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-1.366578</td>\n",
       "      <td>-0.146491</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_016579</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>0.05266</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.017537</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.638921</td>\n",
       "      <td>-0.0</td>\n",
       "      <td>0.368176</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.027613</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_001700</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.462524</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.638921</td>\n",
       "      <td>-0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>NM_001920</td>\n",
       "      <td>-1.129317</td>\n",
       "      <td>0.05266</td>\n",
       "      <td>0.229692</td>\n",
       "      <td>0.079549</td>\n",
       "      <td>-0.950022</td>\n",
       "      <td>-0.055535</td>\n",
       "      <td>0.79533</td>\n",
       "      <td>0.161332</td>\n",
       "      <td>0.462524</td>\n",
       "      <td>-0.017537</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.638921</td>\n",
       "      <td>-0.0</td>\n",
       "      <td>0.368176</td>\n",
       "      <td>0.623895</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-1.366578</td>\n",
       "      <td>-0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 75 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              MATR3    ENOX1     PTBP1     RBMS3      RBM6    LIN28A   HNRNPC  \\\n",
       "RefSeq                                                                          \n",
       "NM_173362 -1.129317  0.05266  0.229692  0.079549 -0.000000 -0.055535  0.79533   \n",
       "NM_006196 -0.000000  0.00000  0.229692  0.000000 -0.000000 -0.000000  0.79533   \n",
       "NM_016579 -0.000000  0.05266  0.000000  0.000000 -0.000000 -0.000000  0.00000   \n",
       "NM_001700 -0.000000  0.00000  0.000000  0.000000 -0.000000 -0.000000  0.00000   \n",
       "NM_001920 -1.129317  0.05266  0.229692  0.079549 -0.950022 -0.055535  0.79533   \n",
       "\n",
       "           HNRNPCL1   SNRNP70     RBM8A  ...     SRSF2  HNRNPH2    DAZAP1  \\\n",
       "RefSeq                                   ...                                \n",
       "NM_173362  0.161332  0.462524 -0.017537  ... -0.638921     -0.0  0.368176   \n",
       "NM_006196  0.161332  0.462524 -0.017537  ... -0.638921     -0.0  0.368176   \n",
       "NM_016579  0.000000  0.000000 -0.017537  ... -0.638921     -0.0  0.368176   \n",
       "NM_001700  0.000000  0.462524 -0.000000  ... -0.638921     -0.0  0.000000   \n",
       "NM_001920  0.161332  0.462524 -0.017537  ... -0.638921     -0.0  0.368176   \n",
       "\n",
       "               MSI1     ESRP2    ZC3H14      TIA1  U2AF2     CPEB4      RALY  \n",
       "RefSeq                                                                        \n",
       "NM_173362  0.623895  0.000000  0.120822 -1.240809    0.0 -1.366578 -0.146491  \n",
       "NM_006196  0.000000  0.027613  0.120822 -1.240809    0.0 -1.366578 -0.146491  \n",
       "NM_016579  0.000000  0.027613  0.000000 -0.000000    0.0 -0.000000 -0.000000  \n",
       "NM_001700  0.000000  0.000000  0.000000 -0.000000    0.0 -0.000000 -0.000000  \n",
       "NM_001920  0.623895  0.000000  0.000000 -0.000000    0.0 -1.366578 -0.000000  \n",
       "\n",
       "[5 rows x 75 columns]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "for rbp in mat.columns:\n",
    "    mat[rbp] = mat[rbp]*motifs.loc[rbp,'diff']\n",
    "mat.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "mat.to_csv('outputs/RBP-v-RefSeq_target_matrix_dExp.txt', sep='\\t', index=True, index_label='RefSeq')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "#penalties are defined as 1/|dExp|\n",
    "penalties = pd.DataFrame(index=motifs.index)\n",
    "penalties['penalties'] = motifs['diff'].apply(lambda x: 1/abs(x))\n",
    "penalties.to_csv('outputs/penalties.txt', sep='\\t', index=True, index_label='RBP')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "exp_fil = pd.DataFrame(index=mat.index)\n",
    "exp_fil['logFC'] = exp.loc[mat.index,'logFC']\n",
    "exp_fil.to_csv('outputs/high-vs-low_metastatic_lines_GSE59857_logFC_refseq_fil.txt', sep='\\t', index=True, index_label='RefSeq')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4. Deriving coefficients"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here, we use gneralized linear models (lasso regression with custom penalty terms) to identify RBPs whose expression is informative for predicting the expression of their putative regulon. This custom penalty term ensures that RBPs whose activity does not change are not selected by the model. This also stabilizes the resulting model, which would otherwise be a major issue as RBPs that belong to the same family often have very similar binding preferences resulting in correlated features in the interaction matrix. After running this regression analysis, the RBPs with the largest assigned coefficients (absolute value) are prioritized."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "import rpy2\n",
    "%load_ext rpy2.ipython"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "from rpy2.rinterface import RRuntimeWarning\n",
    "warnings.filterwarnings(\"ignore\", category=RRuntimeWarning)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%R -o coef,fit,x,y,y.t\n",
    "library(Matrix)\n",
    "library(glmnet)\n",
    "library(tidyverse)\n",
    "x <- read.table('outputs/RBP-v-RefSeq_target_matrix_dExp.txt', row.names=1, header=TRUE, sep=\"\\t\")\n",
    "y <- read.table('outputs/high-vs-low_metastatic_lines_GSE59857_logFC_refseq_fil.txt', row.names=1, header=TRUE, sep=\"\\t\")\n",
    "p.fac <- read.table('outputs/penalties.txt', row.names=1, header=TRUE, sep=\"\\t\")\n",
    "\n",
    "library(bestNormalize)\n",
    "(BNobject <- bestNormalize(as.matrix(y), quiet=T))\n",
    "\n",
    "fit <- glmnet(as.matrix(x), BNobject$x.t, penalty.factor=as.numeric(unlist(p.fac)), family=\"gaussian\", alpha=1)\n",
    "y.t <- BNobject$x.t\n",
    "\n",
    "coef <- as.matrix(coef(fit))\n",
    "colnames(coef) <- apply(abs(fit$beta), 2, sum) #set L1 norm as the header\n",
    "coef <- coef[,round(as.numeric(colnames(coef)), digits=2)<=0.75]\n",
    "coef <- coef[rowSums(coef)!=0,]\n",
    "coef <- coef[order(coef[,dim(coef)[2]], decreasing=T),]\n",
    "coef <- coef[-1,]\n",
    "x <- as_tibble(x, rownames=\"RefSeq\")\n",
    "y <- as_tibble(y, rownames=\"RefSeq\")\n",
    "fit$dev\n",
    "write.table(coef, file='outputs/high-vs-low_metastatic_lines_GSE59857_coef.txt',sep=\"\\t\", row.names=TRUE, col.names=NA, quote=F)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%%R -i coef\n",
    "library(tidyverse)\n",
    "library(reshape2)\n",
    "df <- as_tibble(melt(coef))\n",
    "colnames(df) <- c('RBP', 'norm', 'coefficient')\n",
    "df %>% ggplot(aes(x=norm, y=coefficient, group=RBP, color=RBP)) + geom_line(size=1.5) + theme_bw(12) + \n",
    "xlab(\"L1 Norm\") + ylab(\"Coefficient\") +\n",
    "theme(text = element_text(size=20), panel.grid.minor = element_blank(), panel.grid.major = element_blank())\n",
    "#ggsave(\"outputs/PRADA_coeffcients_vs_L1norm.pdf\", width=4.5, height=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}