Example_API_usage.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Example CREEDS API Usages Walk-Through\n",
    "### _Zichen Wang_\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "## required python packages\n",
    "import json\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import requests\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import warnings \n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "BASE_URL = 'http://amp.pharm.mssm.edu/CREEDS/'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Search signatures using text\n",
    "\n",
    "The `/search` endpoint handles the search of the metadata associated with gene expression signatures. You can use a query term of interest such as 'breast cancer' to perform a search using an HTTP GET request.\n",
    "\n",
    "The `requests.get()` function performs an HTTP GET request and returns a response object. JSON data associated with the response object can be parsed using the `.json()` method:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200\n"
     ]
    }
   ],
   "source": [
    "r = requests.get(BASE_URL + 'search', params={'q':'breast cancer'})\n",
    "response = r.json()\n",
    "print r.status_code"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We next convert the JSON response to a `pandas.DataFrame` and examine the search result:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>cell_type</th>\n",
       "      <th>ctrl_ids</th>\n",
       "      <th>curator</th>\n",
       "      <th>disease_name</th>\n",
       "      <th>do_id</th>\n",
       "      <th>geo_id</th>\n",
       "      <th>organism</th>\n",
       "      <th>pert_ids</th>\n",
       "      <th>platform</th>\n",
       "      <th>umls_cui</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>dz:11</th>\n",
       "      <td>Mammary Epithelium</td>\n",
       "      <td>[GSM1932, GSM1933, GSM1935, GSM1949, GSM1950, ...</td>\n",
       "      <td>Joel.Dudley</td>\n",
       "      <td>breast cancer</td>\n",
       "      <td>DOID:1612</td>\n",
       "      <td>GSE53</td>\n",
       "      <td>human</td>\n",
       "      <td>[GSM1928, GSM1929, GSM1930, GSM1931, GSM1934, ...</td>\n",
       "      <td>GPL170</td>\n",
       "      <td>C0006142</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dz:24</th>\n",
       "      <td>Mammary Gland Tissue</td>\n",
       "      <td>[GSM85513, GSM85514, GSM85515, GSM85516, GSM85...</td>\n",
       "      <td>Joel.Dudley</td>\n",
       "      <td>breast cancer</td>\n",
       "      <td>DOID:1612</td>\n",
       "      <td>GSE3744</td>\n",
       "      <td>human</td>\n",
       "      <td>[GSM85473, GSM85474, GSM85475, GSM85476, GSM85...</td>\n",
       "      <td>GPL570</td>\n",
       "      <td>C0006142</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dz:28</th>\n",
       "      <td>Mammary gland</td>\n",
       "      <td>[GSM48242, GSM48244, GSM48245]</td>\n",
       "      <td>Joel.Dudley</td>\n",
       "      <td>breast cancer</td>\n",
       "      <td>DOID:1612</td>\n",
       "      <td>GSE2528</td>\n",
       "      <td>mouse</td>\n",
       "      <td>[GSM48236, GSM48237, GSM48238, GSM48239, GSM48...</td>\n",
       "      <td>GPL81</td>\n",
       "      <td>C0006142</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dz:39</th>\n",
       "      <td>Epithelial Cell</td>\n",
       "      <td>[GSM38895, GSM38896]</td>\n",
       "      <td>Joel.Dudley</td>\n",
       "      <td>breast cancer</td>\n",
       "      <td>DOID:1612</td>\n",
       "      <td>GSE2155</td>\n",
       "      <td>human</td>\n",
       "      <td>[GSM38897, GSM38898, GSM38899, GSM38900, GSM38...</td>\n",
       "      <td>GPL1794</td>\n",
       "      <td>C0006142</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dz:52</th>\n",
       "      <td>Mammary Gland Tissue</td>\n",
       "      <td>[GSM22369, GSM22374, GSM22381, GSM22382, GSM22...</td>\n",
       "      <td>Joel.Dudley</td>\n",
       "      <td>breast cancer</td>\n",
       "      <td>DOID:1612</td>\n",
       "      <td>GSE1378</td>\n",
       "      <td>human</td>\n",
       "      <td>[GSM22365, GSM22366, GSM22367, GSM22368, GSM22...</td>\n",
       "      <td>GPL1223</td>\n",
       "      <td>C0006142</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  cell_type  \\\n",
       "id                            \n",
       "dz:11    Mammary Epithelium   \n",
       "dz:24  Mammary Gland Tissue   \n",
       "dz:28         Mammary gland   \n",
       "dz:39       Epithelial Cell   \n",
       "dz:52  Mammary Gland Tissue   \n",
       "\n",
       "                                                ctrl_ids      curator  \\\n",
       "id                                                                      \n",
       "dz:11  [GSM1932, GSM1933, GSM1935, GSM1949, GSM1950, ...  Joel.Dudley   \n",
       "dz:24  [GSM85513, GSM85514, GSM85515, GSM85516, GSM85...  Joel.Dudley   \n",
       "dz:28                     [GSM48242, GSM48244, GSM48245]  Joel.Dudley   \n",
       "dz:39                               [GSM38895, GSM38896]  Joel.Dudley   \n",
       "dz:52  [GSM22369, GSM22374, GSM22381, GSM22382, GSM22...  Joel.Dudley   \n",
       "\n",
       "        disease_name      do_id   geo_id organism  \\\n",
       "id                                                  \n",
       "dz:11  breast cancer  DOID:1612    GSE53    human   \n",
       "dz:24  breast cancer  DOID:1612  GSE3744    human   \n",
       "dz:28  breast cancer  DOID:1612  GSE2528    mouse   \n",
       "dz:39  breast cancer  DOID:1612  GSE2155    human   \n",
       "dz:52  breast cancer  DOID:1612  GSE1378    human   \n",
       "\n",
       "                                                pert_ids platform  umls_cui  \n",
       "id                                                                           \n",
       "dz:11  [GSM1928, GSM1929, GSM1930, GSM1931, GSM1934, ...   GPL170  C0006142  \n",
       "dz:24  [GSM85473, GSM85474, GSM85475, GSM85476, GSM85...   GPL570  C0006142  \n",
       "dz:28  [GSM48236, GSM48237, GSM48238, GSM48239, GSM48...    GPL81  C0006142  \n",
       "dz:39  [GSM38897, GSM38898, GSM38899, GSM38900, GSM38...  GPL1794  C0006142  \n",
       "dz:52  [GSM22365, GSM22366, GSM22367, GSM22368, GSM22...  GPL1223  C0006142  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "response_df = pd.DataFrame.from_records(response).set_index('id')\n",
    "response_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Query the signature database using up/down gene sets to find similar/opposite signatures\n",
    "Signatures can also be queried against using a pair of up/down gene lists from the `/search` endpoint using HTTP POST request. Here we generate vectors of up and down genes to query against the CREEDS database:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "## Get some up and down gene lists \n",
    "up_genes = ['KIAA0907','KDM5A','CDC25A','EGR1','GADD45B','RELB','TERF2IP','SMNDC1','TICAM1','NFKB2',\n",
    "            'RGS2','NCOA3','ICAM1','TEX10','CNOT4','ARID4B','CLPX','CHIC2','CXCL2','FBXO11','MTF2',\n",
    "            'CDK2','DNTTIP2','GADD45A','GOLT1B','POLR2K','NFKBIE','GABPB1','ECD','PHKG2','RAD9A',\n",
    "            'NET1','KIAA0753','EZH2','NRAS','ATP6V0B','CDK7','CCNH','SENP6','TIPARP','FOS','ARPP19',\n",
    "            'TFAP2A','KDM5B','NPC1','TP53BP2','NUSAP1']\n",
    "dn_genes = ['SCCPDH','KIF20A','FZD7','USP22','PIP4K2B','CRYZ','GNB5','EIF4EBP1','PHGDH','RRAGA',\n",
    "            'SLC25A46','RPA1','HADH','DAG1','RPIA','P4HA2','MACF1','TMEM97','MPZL1','PSMG1','PLK1',\n",
    "            'SLC37A4','GLRX','CBR3','PRSS23','NUDCD3','CDC20','KIAA0528','NIPSNAP1','TRAM2','STUB1',\n",
    "            'DERA','MTHFD2','BLVRA','IARS2','LIPA','PGM1','CNDP2','BNIP3','CTSL1','CDC25B','HSPA8',\n",
    "            'EPRS','PAX8','SACM1L','HOXA5','TLE1','PYGL','TUBB6','LOXL1']\n",
    "payload = {\n",
    "    'up_genes': up_genes,\n",
    "    'dn_genes': dn_genes,\n",
    "    'direction': 'similar',\n",
    "    'db_version': 'v1.0'\n",
    "    }"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "POST the input gene sets to the API and load the response to `pandas.DataFrame`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200\n"
     ]
    }
   ],
   "source": [
    "r = requests.post(BASE_URL + 'search', json=payload)\n",
    "print r.status_code\n",
    "response = r.json()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>geo_id</th>\n",
       "      <th>name</th>\n",
       "      <th>signed_jaccard</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>gene:2008</th>\n",
       "      <td>GSE8747</td>\n",
       "      <td>[Met, http://www.ncbi.nlm.nih.gov/gene/17295]</td>\n",
       "      <td>0.00289</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gene:84</th>\n",
       "      <td>GSE10516</td>\n",
       "      <td>[Lmx1b, http://www.ncbi.nlm.nih.gov/gene/16917]</td>\n",
       "      <td>0.00251</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gene:2314</th>\n",
       "      <td>GSE53388</td>\n",
       "      <td>[Arhgap5, http://www.ncbi.nlm.nih.gov/gene/11855]</td>\n",
       "      <td>0.00207</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gene:2916</th>\n",
       "      <td>GSE40377</td>\n",
       "      <td>[Pde10a, http://www.ncbi.nlm.nih.gov/gene/23984]</td>\n",
       "      <td>0.00178</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gene:601</th>\n",
       "      <td>GSE13590</td>\n",
       "      <td>[MIR140, http://www.ncbi.nlm.nih.gov/gene/406932]</td>\n",
       "      <td>0.00162</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             geo_id                                               name  \\\n",
       "id                                                                       \n",
       "gene:2008   GSE8747      [Met, http://www.ncbi.nlm.nih.gov/gene/17295]   \n",
       "gene:84    GSE10516    [Lmx1b, http://www.ncbi.nlm.nih.gov/gene/16917]   \n",
       "gene:2314  GSE53388  [Arhgap5, http://www.ncbi.nlm.nih.gov/gene/11855]   \n",
       "gene:2916  GSE40377   [Pde10a, http://www.ncbi.nlm.nih.gov/gene/23984]   \n",
       "gene:601   GSE13590  [MIR140, http://www.ncbi.nlm.nih.gov/gene/406932]   \n",
       "\n",
       "           signed_jaccard  \n",
       "id                         \n",
       "gene:2008         0.00289  \n",
       "gene:84           0.00251  \n",
       "gene:2314         0.00207  \n",
       "gene:2916         0.00178  \n",
       "gene:601          0.00162  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "response_df = pd.DataFrame.from_records(response).set_index('id')\n",
    "response_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Retrieve a signature using `id`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "r = requests.get(BASE_URL + 'api', params={'id': response_df.index[0]})\n",
    "sig = r.json()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`sig` is a python `dictionary` with the following keys"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[u'cell_type', u'pert_ids', u'hs_gene_symbol', u'curator', u'id', u'geo_id', u'platform', u'ctrl_ids', u'up_genes', u'mm_gene_symbol', u'organism', u'down_genes', u'pert_type']\n"
     ]
    }
   ],
   "source": [
    "print sig.keys()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Retrieve a gene expression dataset from GEO and perform selected analyses\n",
    "### 4.0. Prepare data\n",
    "Examine the metadata required to retrieve the dataset from GEO:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GEO id: GSE8747\n",
      "Control GSMs: [u'GSM217200', u'GSM217202', u'GSM217204', u'GSM217206', u'GSM217208']\n",
      "Perturbation GSMs: [u'GSM217210', u'GSM217214', u'GSM217215', u'GSM217217']\n",
      "GEO platform id: GPL4200\n"
     ]
    }
   ],
   "source": [
    "print 'GEO id: %s' % sig['geo_id']\n",
    "print 'Control GSMs: %s' % sig['ctrl_ids']\n",
    "print 'Perturbation GSMs: %s' % sig['pert_ids']\n",
    "print 'GEO platform id: %s' % sig['platform']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": false
   },
   "source": [
    "In order to retrieve gene expression data from GEO, we need to use certain custom utility functions that can be found at: https://github.com/MaayanLab/creeds/tree/master/externals. \n",
    "\n",
    "To identify Differentially Expressed Genes (DEGs), we use the [Charateristic Direction (CD)](http://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-15-79) algorithm implemented in the [geode package](https://github.com/wangz10/geode)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from externals import (geodownloader, softparser, RNAseq)\n",
    "import geode"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "gse_file = geodownloader.download(sig['geo_id'])\n",
    "annot_file = geodownloader.download(sig['platform'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(36607, 9)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>GSM217200</th>\n",
       "      <th>GSM217202</th>\n",
       "      <th>GSM217204</th>\n",
       "      <th>GSM217206</th>\n",
       "      <th>GSM217208</th>\n",
       "      <th>GSM217210</th>\n",
       "      <th>GSM217214</th>\n",
       "      <th>GSM217215</th>\n",
       "      <th>GSM217217</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ID_REF</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1762437_1</th>\n",
       "      <td>0.789</td>\n",
       "      <td>0.353</td>\n",
       "      <td>0.562</td>\n",
       "      <td>-0.539</td>\n",
       "      <td>0.376</td>\n",
       "      <td>0.555</td>\n",
       "      <td>1.032</td>\n",
       "      <td>0.712</td>\n",
       "      <td>0.929</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1762438_1</th>\n",
       "      <td>0.694</td>\n",
       "      <td>0.445</td>\n",
       "      <td>0.681</td>\n",
       "      <td>-0.648</td>\n",
       "      <td>0.296</td>\n",
       "      <td>0.373</td>\n",
       "      <td>0.775</td>\n",
       "      <td>0.783</td>\n",
       "      <td>0.750</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1762439_1</th>\n",
       "      <td>0.741</td>\n",
       "      <td>0.240</td>\n",
       "      <td>0.569</td>\n",
       "      <td>-0.704</td>\n",
       "      <td>0.281</td>\n",
       "      <td>0.489</td>\n",
       "      <td>0.970</td>\n",
       "      <td>0.953</td>\n",
       "      <td>0.458</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1762440_1</th>\n",
       "      <td>0.728</td>\n",
       "      <td>0.322</td>\n",
       "      <td>0.633</td>\n",
       "      <td>-0.623</td>\n",
       "      <td>0.170</td>\n",
       "      <td>0.392</td>\n",
       "      <td>0.706</td>\n",
       "      <td>0.530</td>\n",
       "      <td>0.543</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1762441_1</th>\n",
       "      <td>0.613</td>\n",
       "      <td>0.380</td>\n",
       "      <td>0.703</td>\n",
       "      <td>-0.730</td>\n",
       "      <td>0.334</td>\n",
       "      <td>0.494</td>\n",
       "      <td>0.801</td>\n",
       "      <td>0.648</td>\n",
       "      <td>0.583</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           GSM217200  GSM217202  GSM217204  GSM217206  GSM217208  GSM217210  \\\n",
       "ID_REF                                                                        \n",
       "1762437_1      0.789      0.353      0.562     -0.539      0.376      0.555   \n",
       "1762438_1      0.694      0.445      0.681     -0.648      0.296      0.373   \n",
       "1762439_1      0.741      0.240      0.569     -0.704      0.281      0.489   \n",
       "1762440_1      0.728      0.322      0.633     -0.623      0.170      0.392   \n",
       "1762441_1      0.613      0.380      0.703     -0.730      0.334      0.494   \n",
       "\n",
       "           GSM217214  GSM217215  GSM217217  \n",
       "ID_REF                                      \n",
       "1762437_1      1.032      0.712      0.929  \n",
       "1762438_1      0.775      0.783      0.750  \n",
       "1762439_1      0.970      0.953      0.458  \n",
       "1762440_1      0.706      0.530      0.543  \n",
       "1762441_1      0.801      0.648      0.583  "
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "expr_df = softparser.parse(gse_file.path(), sig['ctrl_ids'], sig['pert_ids'])\n",
    "print expr_df.shape\n",
    "expr_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": false
   },
   "source": [
    "### 4.1. Explanatory analyses\n",
    "Visualize the gene expression values across samples using a density plot:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x111623590>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYEAAAENCAYAAADpK9mHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n7XUIOOUdBAAAAAAAAAAAAAAA/1gWUsiG+7BFPgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10e30de50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = expr_df.plot(kind='kde')\n",
    "ax.set_xlabel('Gene expression')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Visualize using a PCA plot:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjkAAAI6CAYAAADFf1nDAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nx3Ecx3Ecx3Ecx3Ecx3Ecp1P5H8ayI+arnEP0AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10c1d3ed0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sample_classes = ['CTRL'] * len(sig['ctrl_ids']) + ['PERT'] * len(sig['pert_ids'])\n",
    "RNAseq.PCA_plot(expr_df.values, sample_classes, log=True, standardize=2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.2. Perform differential expression analysis\n",
    "### 4.2.1. Applying the Characteristic Direcion (CD) algorithm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CD_coefficient</th>\n",
       "      <th>probes</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.004705</td>\n",
       "      <td>1762437_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.002536</td>\n",
       "      <td>1762438_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.004283</td>\n",
       "      <td>1762439_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.002002</td>\n",
       "      <td>1762440_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.002548</td>\n",
       "      <td>1762441_1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   CD_coefficient     probes\n",
       "0        0.004705  1762437_1\n",
       "1        0.002536  1762438_1\n",
       "2        0.004283  1762439_1\n",
       "3        0.002002  1762440_1\n",
       "4        0.002548  1762441_1"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_classes = [1] * len(sig['ctrl_ids']) + [2] * len(sig['pert_ids'])\n",
    "result = geode.chdir(expr_df.values, sample_classes, expr_df.index,\n",
    "                    gamma=.5, sort=False)\n",
    "result = pd.DataFrame(result, columns=['CD_coefficient', 'probes'])\n",
    "result.head()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}