run_graph.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "kwdF-W4_xZIR"
   },
   "source": [
    "# Preprocessing and run hSBM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Filippo_Valle \n",
      "last updated: Fri Oct 16 2020 \n",
      "\n",
      "CPython 3.8.5\n",
      "IPython 7.18.1\n",
      "\n",
      "pandas 1.1.2\n",
      "numpy 1.19.1\n",
      "scanpy 1.6.0\n",
      "\n",
      "compiler   : GCC 7.5.0\n",
      "system     : Linux\n",
      "release    : 5.4.0-48-generic\n",
      "machine    : x86_64\n",
      "processor  : x86_64\n",
      "CPU cores  : 12\n",
      "interpreter: 64bit\n",
      "Git hash   : 636ec321ff90a30300419083fcd3340bd93173b7\n",
      "Git repo   : git@github.com:fvalle1/cancers\n",
      "Git branch : develop\n",
      "watermark 2.0.2\n"
     ]
    }
   ],
   "source": [
    "%load_ext watermark\n",
    "%watermark -v -m  -u -n -p pandas,numpy,scanpy -a Filippo_Valle -g -r -b -w"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Hrsc2s1MLFNv"
   },
   "outputs": [],
   "source": [
    "# import some libraries\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import os, sys"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This notebook should be run using **Python 3.6** kernel"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import graph_tool as gt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "collapsed": true,
    "id": "ITa5RLqTxluy",
    "jupyter": {
     "outputs_hidden": true
    },
    "outputId": "8a1d2ad9-141b-443c-8def-161c18d0e68c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Executing: /tmp/apt-key-gpghome.tBSONqGzqb/gpg.1.sh --keyserver keys.openpgp.org --recv-key 612DEFB798507F25\n",
      "gpg: key 612DEFB798507F25: public key \"Tiago de Paula Peixoto <tiago@skewed.de>\" imported\n",
      "gpg: Total number processed: 1\n",
      "gpg:               imported: 1\n",
      "Get:1 http://security.ubuntu.com/ubuntu bionic-security InRelease [88.7 kB]\n",
      "Ign:2 https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64  InRelease\n",
      "Get:3 https://cloud.r-project.org/bin/linux/ubuntu bionic-cran35/ InRelease [3,626 B]\n",
      "Hit:4 http://ppa.launchpad.net/graphics-drivers/ppa/ubuntu bionic InRelease\n",
      "Hit:5 http://archive.ubuntu.com/ubuntu bionic InRelease\n",
      "Ign:6 https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64  InRelease\n",
      "Get:7 https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64  Release [697 B]\n",
      "Get:8 https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64  Release [564 B]\n",
      "Get:9 https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64  Release.gpg [801 B]\n",
      "Get:10 https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64  Release.gpg [833 B]\n",
      "Get:11 http://archive.ubuntu.com/ubuntu bionic-updates InRelease [88.7 kB]\n",
      "Get:12 http://security.ubuntu.com/ubuntu bionic-security/main amd64 Packages [960 kB]\n",
      "Get:13 http://ppa.launchpad.net/marutter/c2d4u3.5/ubuntu bionic InRelease [15.4 kB]\n",
      "Get:14 http://downloads.skewed.de/apt bionic InRelease [4,310 B]\n",
      "Get:15 http://security.ubuntu.com/ubuntu bionic-security/universe amd64 Packages [861 kB]\n",
      "Get:16 https://cloud.r-project.org/bin/linux/ubuntu bionic-cran35/ Packages [93.1 kB]\n",
      "Get:17 http://archive.ubuntu.com/ubuntu bionic-backports InRelease [74.6 kB]\n",
      "Ign:18 https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64  Packages\n",
      "Get:18 https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64  Packages [151 kB]\n",
      "Get:19 https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64  Packages [32.9 kB]\n",
      "Get:20 http://ppa.launchpad.net/marutter/c2d4u3.5/ubuntu bionic/main Sources [1,839 kB]\n",
      "Get:21 http://downloads.skewed.de/apt bionic/main amd64 Packages [1,054 B]\n",
      "Get:22 http://archive.ubuntu.com/ubuntu bionic-updates/universe amd64 Packages [1,397 kB]\n",
      "Get:23 http://archive.ubuntu.com/ubuntu bionic-updates/main amd64 Packages [1,254 kB]\n",
      "Get:24 http://ppa.launchpad.net/marutter/c2d4u3.5/ubuntu bionic/main amd64 Packages [887 kB]\n",
      "Fetched 7,755 kB in 3s (2,617 kB/s)\n",
      "Reading package lists... Done\n",
      "Reading package lists... Done\n",
      "Building dependency tree       \n",
      "Reading state information... Done\n",
      "The following package was automatically installed and is no longer required:\n",
      "  libnvidia-common-440\n",
      "Use 'apt autoremove' to remove it.\n",
      "The following additional packages will be installed:\n",
      "  fonts-lyx gir1.2-atk-1.0 gir1.2-freedesktop gir1.2-gdkpixbuf-2.0\n",
      "  gir1.2-gtk-3.0 gir1.2-pango-1.0 javascript-common libcairomm-1.0-1v5\n",
      "  libjs-jquery libjs-jquery-ui libpangoxft-1.0-0 libsigc++-2.0-0v5\n",
      "  python-backports.functools-lru-cache python-cairo python-cycler\n",
      "  python-dateutil python-gi python-gi-cairo python-gv python-matplotlib\n",
      "  python-matplotlib-data python-olefile python-pil python-pyparsing python-six\n",
      "  python-subprocess32 python-tz python3-decorator python3-olefile python3-pil\n",
      "  python3-scipy ttf-bitstream-vera\n",
      "Suggested packages:\n",
      "  apache2 | lighttpd | httpd libjs-jquery-ui-docs python-cycler-doc dvipng\n",
      "  ghostscript inkscape ipython python-cairocffi python-configobj\n",
      "  python-excelerator python-gobject python-matplotlib-doc python-nose\n",
      "  python-qt4 python-scipy python-sip python-tornado python-traits\n",
      "  python-wxgtk3.0 texlive-extra-utils texlive-latex-extra ttf-staypuft\n",
      "  python-pil-doc python-pil-dbg python-pyparsing-doc python3-pil-dbg\n",
      "  python-scipy-doc\n",
      "The following NEW packages will be installed:\n",
      "  fonts-lyx gir1.2-atk-1.0 gir1.2-freedesktop gir1.2-gdkpixbuf-2.0\n",
      "  gir1.2-gtk-3.0 gir1.2-pango-1.0 javascript-common libcairomm-1.0-1v5\n",
      "  libjs-jquery libjs-jquery-ui libpangoxft-1.0-0 libsigc++-2.0-0v5\n",
      "  python-backports.functools-lru-cache python-cairo python-cycler\n",
      "  python-dateutil python-gi python-gi-cairo python-gv python-matplotlib\n",
      "  python-matplotlib-data python-olefile python-pil python-pyparsing python-six\n",
      "  python-subprocess32 python-tz python3-decorator python3-graph-tool\n",
      "  python3-olefile python3-pil python3-scipy ttf-bitstream-vera\n",
      "0 upgraded, 33 newly installed, 0 to remove and 50 not upgraded.\n",
      "Need to get 58.5 MB of archives.\n",
      "After this operation, 360 MB of additional disk space will be used.\n",
      "Get:1 http://archive.ubuntu.com/ubuntu bionic-updates/universe amd64 fonts-lyx all 2.2.4-0ubuntu0.18.04.1 [155 kB]\n",
      "Get:2 http://downloads.skewed.de/apt bionic/main amd64 python3-graph-tool amd64 2.32 [38.8 MB]\n",
      "Get:3 http://archive.ubuntu.com/ubuntu bionic/main amd64 gir1.2-atk-1.0 amd64 2.28.1-1 [17.8 kB]\n",
      "Get:4 http://archive.ubuntu.com/ubuntu bionic/main amd64 gir1.2-freedesktop amd64 1.56.1-1 [9,080 B]\n",
      "Get:5 http://archive.ubuntu.com/ubuntu bionic/main amd64 gir1.2-gdkpixbuf-2.0 amd64 2.36.11-2 [7,748 B]\n",
      "Get:6 http://archive.ubuntu.com/ubuntu bionic-updates/main amd64 libpangoxft-1.0-0 amd64 1.40.14-1ubuntu0.1 [15.0 kB]\n",
      "Get:7 http://archive.ubuntu.com/ubuntu bionic-updates/main amd64 gir1.2-pango-1.0 amd64 1.40.14-1ubuntu0.1 [21.6 kB]\n",
      "Get:8 http://archive.ubuntu.com/ubuntu bionic-updates/main amd64 gir1.2-gtk-3.0 amd64 3.22.30-1ubuntu4 [194 kB]\n",
      "Get:9 http://archive.ubuntu.com/ubuntu bionic/main amd64 javascript-common all 11 [6,066 B]\n",
      "Get:10 http://archive.ubuntu.com/ubuntu bionic/main amd64 libsigc++-2.0-0v5 amd64 2.10.0-2 [10.9 kB]\n",
      "Get:11 http://archive.ubuntu.com/ubuntu bionic/main amd64 libcairomm-1.0-1v5 amd64 1.12.2-3 [37.1 kB]\n",
      "Get:12 http://archive.ubuntu.com/ubuntu bionic/main amd64 libjs-jquery all 3.2.1-1 [152 kB]\n",
      "Get:13 http://archive.ubuntu.com/ubuntu bionic/universe amd64 libjs-jquery-ui all 1.12.1+dfsg-5 [232 kB]\n",
      "Get:14 http://archive.ubuntu.com/ubuntu bionic/universe amd64 python-backports.functools-lru-cache all 1.4-2 [5,960 B]\n",
      "Get:15 http://archive.ubuntu.com/ubuntu bionic/main amd64 python-cairo amd64 1.16.2-1 [56.4 kB]\n",
      "Get:16 http://archive.ubuntu.com/ubuntu bionic/main amd64 python-six all 1.11.0-2 [11.3 kB]\n",
      "Get:17 http://archive.ubuntu.com/ubuntu bionic/universe amd64 python-cycler all 0.10.0-1 [7,520 B]\n",
      "Get:18 http://archive.ubuntu.com/ubuntu bionic/main amd64 python-dateutil all 2.6.1-1 [60.6 kB]\n",
      "Get:19 http://archive.ubuntu.com/ubuntu bionic-updates/main amd64 python-gi amd64 3.26.1-2ubuntu1 [197 kB]\n",
      "Get:20 http://archive.ubuntu.com/ubuntu bionic-updates/universe amd64 python-gi-cairo amd64 3.26.1-2ubuntu1 [6,632 B]\n",
      "Get:21 http://archive.ubuntu.com/ubuntu bionic/universe amd64 python-gv amd64 2.40.1-2 [32.3 kB]\n",
      "Get:22 http://archive.ubuntu.com/ubuntu bionic/universe amd64 ttf-bitstream-vera all 1.10-8 [352 kB]\n",
      "Get:23 http://archive.ubuntu.com/ubuntu bionic/universe amd64 python-matplotlib-data all 2.1.1-2ubuntu3 [3,774 kB]\n",
      "Get:24 http://archive.ubuntu.com/ubuntu bionic/main amd64 python-pyparsing all 2.2.0+dfsg1-2 [52.1 kB]\n",
      "Get:25 http://archive.ubuntu.com/ubuntu bionic/main amd64 python-tz all 2018.3-2 [31.6 kB]\n",
      "Get:26 http://archive.ubuntu.com/ubuntu bionic/universe amd64 python-subprocess32 amd64 3.2.7-3 [27.2 kB]\n",
      "Get:27 http://archive.ubuntu.com/ubuntu bionic/universe amd64 python-matplotlib amd64 2.1.1-2ubuntu3 [3,901 kB]\n",
      "Get:28 http://archive.ubuntu.com/ubuntu bionic/main amd64 python-olefile all 0.45.1-1 [33.2 kB]\n",
      "Get:29 http://archive.ubuntu.com/ubuntu bionic-updates/main amd64 python-pil amd64 5.1.0-1ubuntu0.2 [329 kB]\n",
      "Get:30 http://archive.ubuntu.com/ubuntu bionic/universe amd64 python3-decorator all 4.1.2-1 [9,364 B]\n",
      "Get:31 http://archive.ubuntu.com/ubuntu bionic/main amd64 python3-olefile all 0.45.1-1 [33.3 kB]\n",
      "Get:32 http://archive.ubuntu.com/ubuntu bionic-updates/main amd64 python3-pil amd64 5.1.0-1ubuntu0.2 [329 kB]\n",
      "Get:33 http://archive.ubuntu.com/ubuntu bionic/universe amd64 python3-scipy amd64 0.19.1-2ubuntu1 [9,619 kB]\n",
      "Fetched 58.5 MB in 3s (22.6 MB/s)\n",
      "Extracting templates from packages: 100%\n",
      "Selecting previously unselected package fonts-lyx.\n",
      "(Reading database ... 144328 files and directories currently installed.)\n",
      "Preparing to unpack .../00-fonts-lyx_2.2.4-0ubuntu0.18.04.1_all.deb ...\n",
      "Unpacking fonts-lyx (2.2.4-0ubuntu0.18.04.1) ...\n",
      "Selecting previously unselected package gir1.2-atk-1.0:amd64.\n",
      "Preparing to unpack .../01-gir1.2-atk-1.0_2.28.1-1_amd64.deb ...\n",
      "Unpacking gir1.2-atk-1.0:amd64 (2.28.1-1) ...\n",
      "Selecting previously unselected package gir1.2-freedesktop:amd64.\n",
      "Preparing to unpack .../02-gir1.2-freedesktop_1.56.1-1_amd64.deb ...\n",
      "Unpacking gir1.2-freedesktop:amd64 (1.56.1-1) ...\n",
      "Selecting previously unselected package gir1.2-gdkpixbuf-2.0:amd64.\n",
      "Preparing to unpack .../03-gir1.2-gdkpixbuf-2.0_2.36.11-2_amd64.deb ...\n",
      "Unpacking gir1.2-gdkpixbuf-2.0:amd64 (2.36.11-2) ...\n",
      "Selecting previously unselected package libpangoxft-1.0-0:amd64.\n",
      "Preparing to unpack .../04-libpangoxft-1.0-0_1.40.14-1ubuntu0.1_amd64.deb ...\n",
      "Unpacking libpangoxft-1.0-0:amd64 (1.40.14-1ubuntu0.1) ...\n",
      "Selecting previously unselected package gir1.2-pango-1.0:amd64.\n",
      "Preparing to unpack .../05-gir1.2-pango-1.0_1.40.14-1ubuntu0.1_amd64.deb ...\n",
      "Unpacking gir1.2-pango-1.0:amd64 (1.40.14-1ubuntu0.1) ...\n",
      "Selecting previously unselected package gir1.2-gtk-3.0:amd64.\n",
      "Preparing to unpack .../06-gir1.2-gtk-3.0_3.22.30-1ubuntu4_amd64.deb ...\n",
      "Unpacking gir1.2-gtk-3.0:amd64 (3.22.30-1ubuntu4) ...\n",
      "Selecting previously unselected package javascript-common.\n",
      "Preparing to unpack .../07-javascript-common_11_all.deb ...\n",
      "Unpacking javascript-common (11) ...\n",
      "Selecting previously unselected package libsigc++-2.0-0v5:amd64.\n",
      "Preparing to unpack .../08-libsigc++-2.0-0v5_2.10.0-2_amd64.deb ...\n",
      "Unpacking libsigc++-2.0-0v5:amd64 (2.10.0-2) ...\n",
      "Selecting previously unselected package libcairomm-1.0-1v5:amd64.\n",
      "Preparing to unpack .../09-libcairomm-1.0-1v5_1.12.2-3_amd64.deb ...\n",
      "Unpacking libcairomm-1.0-1v5:amd64 (1.12.2-3) ...\n",
      "Selecting previously unselected package libjs-jquery.\n",
      "Preparing to unpack .../10-libjs-jquery_3.2.1-1_all.deb ...\n",
      "Unpacking libjs-jquery (3.2.1-1) ...\n",
      "Selecting previously unselected package libjs-jquery-ui.\n",
      "Preparing to unpack .../11-libjs-jquery-ui_1.12.1+dfsg-5_all.deb ...\n",
      "Unpacking libjs-jquery-ui (1.12.1+dfsg-5) ...\n",
      "Selecting previously unselected package python-backports.functools-lru-cache.\n",
      "Preparing to unpack .../12-python-backports.functools-lru-cache_1.4-2_all.deb ...\n",
      "Unpacking python-backports.functools-lru-cache (1.4-2) ...\n",
      "Selecting previously unselected package python-cairo:amd64.\n",
      "Preparing to unpack .../13-python-cairo_1.16.2-1_amd64.deb ...\n",
      "Unpacking python-cairo:amd64 (1.16.2-1) ...\n",
      "Selecting previously unselected package python-six.\n",
      "Preparing to unpack .../14-python-six_1.11.0-2_all.deb ...\n",
      "Unpacking python-six (1.11.0-2) ...\n",
      "Selecting previously unselected package python-cycler.\n",
      "Preparing to unpack .../15-python-cycler_0.10.0-1_all.deb ...\n",
      "Unpacking python-cycler (0.10.0-1) ...\n",
      "Selecting previously unselected package python-dateutil.\n",
      "Preparing to unpack .../16-python-dateutil_2.6.1-1_all.deb ...\n",
      "Unpacking python-dateutil (2.6.1-1) ...\n",
      "Selecting previously unselected package python-gi.\n",
      "Preparing to unpack .../17-python-gi_3.26.1-2ubuntu1_amd64.deb ...\n",
      "Unpacking python-gi (3.26.1-2ubuntu1) ...\n",
      "Selecting previously unselected package python-gi-cairo.\n",
      "Preparing to unpack .../18-python-gi-cairo_3.26.1-2ubuntu1_amd64.deb ...\n",
      "Unpacking python-gi-cairo (3.26.1-2ubuntu1) ...\n",
      "Selecting previously unselected package python-gv.\n",
      "Preparing to unpack .../19-python-gv_2.40.1-2_amd64.deb ...\n",
      "Unpacking python-gv (2.40.1-2) ...\n",
      "Selecting previously unselected package ttf-bitstream-vera.\n",
      "Preparing to unpack .../20-ttf-bitstream-vera_1.10-8_all.deb ...\n",
      "Unpacking ttf-bitstream-vera (1.10-8) ...\n",
      "Selecting previously unselected package python-matplotlib-data.\n",
      "Preparing to unpack .../21-python-matplotlib-data_2.1.1-2ubuntu3_all.deb ...\n",
      "Unpacking python-matplotlib-data (2.1.1-2ubuntu3) ...\n",
      "Selecting previously unselected package python-pyparsing.\n",
      "Preparing to unpack .../22-python-pyparsing_2.2.0+dfsg1-2_all.deb ...\n",
      "Unpacking python-pyparsing (2.2.0+dfsg1-2) ...\n",
      "Selecting previously unselected package python-tz.\n",
      "Preparing to unpack .../23-python-tz_2018.3-2_all.deb ...\n",
      "Unpacking python-tz (2018.3-2) ...\n",
      "Selecting previously unselected package python-subprocess32.\n",
      "Preparing to unpack .../24-python-subprocess32_3.2.7-3_amd64.deb ...\n",
      "Unpacking python-subprocess32 (3.2.7-3) ...\n",
      "Selecting previously unselected package python-matplotlib.\n",
      "Preparing to unpack .../25-python-matplotlib_2.1.1-2ubuntu3_amd64.deb ...\n",
      "Unpacking python-matplotlib (2.1.1-2ubuntu3) ...\n",
      "Selecting previously unselected package python-olefile.\n",
      "Preparing to unpack .../26-python-olefile_0.45.1-1_all.deb ...\n",
      "Unpacking python-olefile (0.45.1-1) ...\n",
      "Selecting previously unselected package python-pil:amd64.\n",
      "Preparing to unpack .../27-python-pil_5.1.0-1ubuntu0.2_amd64.deb ...\n",
      "Unpacking python-pil:amd64 (5.1.0-1ubuntu0.2) ...\n",
      "Selecting previously unselected package python3-decorator.\n",
      "Preparing to unpack .../28-python3-decorator_4.1.2-1_all.deb ...\n",
      "Unpacking python3-decorator (4.1.2-1) ...\n",
      "Selecting previously unselected package python3-olefile.\n",
      "Preparing to unpack .../29-python3-olefile_0.45.1-1_all.deb ...\n",
      "Unpacking python3-olefile (0.45.1-1) ...\n",
      "Selecting previously unselected package python3-pil:amd64.\n",
      "Preparing to unpack .../30-python3-pil_5.1.0-1ubuntu0.2_amd64.deb ...\n",
      "Unpacking python3-pil:amd64 (5.1.0-1ubuntu0.2) ...\n",
      "Selecting previously unselected package python3-scipy.\n",
      "Preparing to unpack .../31-python3-scipy_0.19.1-2ubuntu1_amd64.deb ...\n",
      "Unpacking python3-scipy (0.19.1-2ubuntu1) ...\n",
      "Selecting previously unselected package python3-graph-tool.\n",
      "Preparing to unpack .../32-python3-graph-tool_2.32_amd64.deb ...\n",
      "Unpacking python3-graph-tool (2.32) ...\n",
      "Setting up libjs-jquery (3.2.1-1) ...\n",
      "Setting up gir1.2-atk-1.0:amd64 (2.28.1-1) ...\n",
      "Setting up python3-pil:amd64 (5.1.0-1ubuntu0.2) ...\n",
      "Setting up python3-olefile (0.45.1-1) ...\n",
      "Setting up python-backports.functools-lru-cache (1.4-2) ...\n",
      "Setting up gir1.2-freedesktop:amd64 (1.56.1-1) ...\n",
      "Setting up libpangoxft-1.0-0:amd64 (1.40.14-1ubuntu0.1) ...\n",
      "Setting up python-tz (2018.3-2) ...\n",
      "Setting up gir1.2-gdkpixbuf-2.0:amd64 (2.36.11-2) ...\n",
      "Setting up gir1.2-pango-1.0:amd64 (1.40.14-1ubuntu0.1) ...\n",
      "Setting up python-gi (3.26.1-2ubuntu1) ...\n",
      "Setting up python-six (1.11.0-2) ...\n",
      "Setting up python-gv (2.40.1-2) ...\n",
      "Setting up libjs-jquery-ui (1.12.1+dfsg-5) ...\n",
      "Setting up python-cairo:amd64 (1.16.2-1) ...\n",
      "Setting up python-olefile (0.45.1-1) ...\n",
      "Setting up ttf-bitstream-vera (1.10-8) ...\n",
      "Setting up javascript-common (11) ...\n",
      "Setting up python-subprocess32 (3.2.7-3) ...\n",
      "Setting up python3-decorator (4.1.2-1) ...\n",
      "Setting up python-pil:amd64 (5.1.0-1ubuntu0.2) ...\n",
      "Setting up libsigc++-2.0-0v5:amd64 (2.10.0-2) ...\n",
      "Setting up fonts-lyx (2.2.4-0ubuntu0.18.04.1) ...\n",
      "Setting up python-pyparsing (2.2.0+dfsg1-2) ...\n",
      "Setting up python-gi-cairo (3.26.1-2ubuntu1) ...\n",
      "Setting up libcairomm-1.0-1v5:amd64 (1.12.2-3) ...\n",
      "Setting up python3-scipy (0.19.1-2ubuntu1) ...\n",
      "Setting up gir1.2-gtk-3.0:amd64 (3.22.30-1ubuntu4) ...\n",
      "Setting up python-matplotlib-data (2.1.1-2ubuntu3) ...\n",
      "Setting up python3-graph-tool (2.32) ...\n",
      "Setting up python-cycler (0.10.0-1) ...\n",
      "Setting up python-dateutil (2.6.1-1) ...\n",
      "Setting up python-matplotlib (2.1.1-2ubuntu3) ...\n",
      "Processing triggers for fontconfig (2.12.6-0ubuntu2) ...\n",
      "Processing triggers for libc-bin (2.27-3ubuntu1) ...\n",
      "/sbin/ldconfig.real: /usr/local/lib/python3.6/dist-packages/ideep4py/lib/libmkldnn.so.0 is not a symbolic link\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#install graph-tool\n",
    "#!echo \"deb http://downloads.skewed.de/apt/ bionic main\" >> /etc/apt/sources.list\n",
    "#!apt-key adv --keyserver keys.openpgp.org --recv-key 612DEFB798507F25\n",
    "#!apt-get update\n",
    "#!apt-get install python3-graph-tool"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Rpy9eQ2WLFN1"
   },
   "source": [
    "# Download data or get the data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "yqSD5nuQx6tW"
   },
   "source": [
    "## Use already downloaded (example from TCGA) data..."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can use [TCGA_GetManifest.ipynb](TCGA_GetManifest.ipynb) and [Table_Creation.ipynb](Table_Creation.ipynb) to create a *mainTable.csv* file from TCGA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "pTTrYG9EEHR8"
   },
   "outputs": [],
   "source": [
    "df = pd.read_csv(\"mainTable_all.csv\", index_col=0)\n",
    "df.index = [g[:15] for g in df.index]\n",
    "df.drop_duplicates(subset=df.index.name, ignore_index=False, inplace=True)\n",
    "df_files = pd.read_csv(\"files.dat\",index_col=0)\n",
    "df_files = df_files[df_files.index.isin(df.columns)]\n",
    "print(df.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 765
    },
    "colab_type": "code",
    "id": "DA7nomq8tFTN",
    "outputId": "230a3e58-0e93-45d1-9cdd-93b72bf7dd51"
   },
   "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>SMATSSCR</th>\n",
       "      <th>SMCENTER</th>\n",
       "      <th>SMPTHNTS</th>\n",
       "      <th>SMRIN</th>\n",
       "      <th>SMTS</th>\n",
       "      <th>SMTSD</th>\n",
       "      <th>SMUBRID</th>\n",
       "      <th>SMTSISCH</th>\n",
       "      <th>SMTSPAX</th>\n",
       "      <th>SMNABTCH</th>\n",
       "      <th>SMNABTCHT</th>\n",
       "      <th>SMNABTCHD</th>\n",
       "      <th>SMGEBTCH</th>\n",
       "      <th>SMGEBTCHD</th>\n",
       "      <th>SMGEBTCHT</th>\n",
       "      <th>SMAFRZE</th>\n",
       "      <th>SMGTC</th>\n",
       "      <th>SME2MPRT</th>\n",
       "      <th>SMCHMPRS</th>\n",
       "      <th>SMNTRART</th>\n",
       "      <th>SMNUMGPS</th>\n",
       "      <th>SMMAPRT</th>\n",
       "      <th>SMEXNCRT</th>\n",
       "      <th>SM550NRM</th>\n",
       "      <th>SMGNSDTC</th>\n",
       "      <th>SMUNMPRT</th>\n",
       "      <th>SM350NRM</th>\n",
       "      <th>SMRDLGTH</th>\n",
       "      <th>SMMNCPB</th>\n",
       "      <th>SME1MMRT</th>\n",
       "      <th>SMSFLGTH</th>\n",
       "      <th>SMESTLBS</th>\n",
       "      <th>SMMPPD</th>\n",
       "      <th>SMNTERRT</th>\n",
       "      <th>SMRRNANM</th>\n",
       "      <th>SMRDTTL</th>\n",
       "      <th>SMVQCFL</th>\n",
       "      <th>SMMNCV</th>\n",
       "      <th>SMTRSCPT</th>\n",
       "      <th>SMMPPDPR</th>\n",
       "      <th>SMCGLGTH</th>\n",
       "      <th>SMGAPPCT</th>\n",
       "      <th>SMUNPDRD</th>\n",
       "      <th>SMNTRNRT</th>\n",
       "      <th>SMMPUNRT</th>\n",
       "      <th>SMEXPEFF</th>\n",
       "      <th>SMMPPDUN</th>\n",
       "      <th>SME2MMRT</th>\n",
       "      <th>SME2ANTI</th>\n",
       "      <th>SMALTALG</th>\n",
       "      <th>SME2SNSE</th>\n",
       "      <th>SMMFLGTH</th>\n",
       "      <th>SME1ANTI</th>\n",
       "      <th>SMSPLTRD</th>\n",
       "      <th>SMBSMMRT</th>\n",
       "      <th>SME1SNSE</th>\n",
       "      <th>SME1PCTS</th>\n",
       "      <th>SMRRNART</th>\n",
       "      <th>SME1MPRT</th>\n",
       "      <th>SMNUM5CD</th>\n",
       "      <th>SMDPMPRT</th>\n",
       "      <th>SME2PCTS</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SAMPID</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",
       "      <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",
       "      <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",
       "      <th>GTEX-1117F-0226-SM-5GZZ7</th>\n",
       "      <td>0.0</td>\n",
       "      <td>B1</td>\n",
       "      <td>2 pieces, ~15% vessel stroma, rep delineated</td>\n",
       "      <td>6.8</td>\n",
       "      <td>Adipose Tissue</td>\n",
       "      <td>Adipose - Subcutaneous</td>\n",
       "      <td>0002190</td>\n",
       "      <td>1214.0</td>\n",
       "      <td>1125.0</td>\n",
       "      <td>BP-43693</td>\n",
       "      <td>RNA Extraction from Paxgene-derived Lysate Pla...</td>\n",
       "      <td>09/17/2013</td>\n",
       "      <td>LCSET-4804</td>\n",
       "      <td>03/05/2014</td>\n",
       "      <td>TruSeq.v1</td>\n",
       "      <td>RNASEQ</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.986026</td>\n",
       "      <td>345562.0</td>\n",
       "      <td>0.966793</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.990383</td>\n",
       "      <td>0.756726</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23548.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>76.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.002403</td>\n",
       "      <td>136.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>66833200.0</td>\n",
       "      <td>0.032919</td>\n",
       "      <td>209558.0</td>\n",
       "      <td>67482200.0</td>\n",
       "      <td>8797660.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23575.0</td>\n",
       "      <td>33158000.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.210067</td>\n",
       "      <td>0.990383</td>\n",
       "      <td>0.749449</td>\n",
       "      <td>66833200.0</td>\n",
       "      <td>0.003919</td>\n",
       "      <td>14462700.0</td>\n",
       "      <td>8914900.0</td>\n",
       "      <td>14575600.0</td>\n",
       "      <td>144.0</td>\n",
       "      <td>14648800.0</td>\n",
       "      <td>11999300.0</td>\n",
       "      <td>0.003158</td>\n",
       "      <td>14669500.0</td>\n",
       "      <td>50.0354</td>\n",
       "      <td>0.003105</td>\n",
       "      <td>0.994740</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>50.1944</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GTEX-1117F-0426-SM-5EGHI</th>\n",
       "      <td>0.0</td>\n",
       "      <td>B1</td>\n",
       "      <td>2 pieces, !5% fibrous connective tissue, delin...</td>\n",
       "      <td>7.1</td>\n",
       "      <td>Muscle</td>\n",
       "      <td>Muscle - Skeletal</td>\n",
       "      <td>0011907</td>\n",
       "      <td>1220.0</td>\n",
       "      <td>1119.0</td>\n",
       "      <td>BP-43495</td>\n",
       "      <td>RNA Extraction from Paxgene-derived Lysate Pla...</td>\n",
       "      <td>09/12/2013</td>\n",
       "      <td>LCSET-4764</td>\n",
       "      <td>02/09/2014</td>\n",
       "      <td>TruSeq.v1</td>\n",
       "      <td>RNASEQ</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.985786</td>\n",
       "      <td>145441.0</td>\n",
       "      <td>0.978816</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.990413</td>\n",
       "      <td>0.838961</td>\n",
       "      <td>NaN</td>\n",
       "      <td>20246.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>76.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.002413</td>\n",
       "      <td>137.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>59963600.0</td>\n",
       "      <td>0.020952</td>\n",
       "      <td>423484.0</td>\n",
       "      <td>60544100.0</td>\n",
       "      <td>9601460.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>20259.0</td>\n",
       "      <td>29785800.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.139855</td>\n",
       "      <td>0.990413</td>\n",
       "      <td>0.830918</td>\n",
       "      <td>59963600.0</td>\n",
       "      <td>0.005538</td>\n",
       "      <td>13154500.0</td>\n",
       "      <td>8383440.0</td>\n",
       "      <td>13125800.0</td>\n",
       "      <td>139.0</td>\n",
       "      <td>13191500.0</td>\n",
       "      <td>11550200.0</td>\n",
       "      <td>0.003968</td>\n",
       "      <td>13340500.0</td>\n",
       "      <td>50.2809</td>\n",
       "      <td>0.006995</td>\n",
       "      <td>0.995041</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>49.9455</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GTEX-1117F-0526-SM-5EGHJ</th>\n",
       "      <td>0.0</td>\n",
       "      <td>B1</td>\n",
       "      <td>2 pieces, clean, Monckebeg medial sclerosis, r...</td>\n",
       "      <td>8.0</td>\n",
       "      <td>Blood Vessel</td>\n",
       "      <td>Artery - Tibial</td>\n",
       "      <td>0007610</td>\n",
       "      <td>1221.0</td>\n",
       "      <td>1120.0</td>\n",
       "      <td>BP-43495</td>\n",
       "      <td>RNA Extraction from Paxgene-derived Lysate Pla...</td>\n",
       "      <td>09/12/2013</td>\n",
       "      <td>LCSET-4764</td>\n",
       "      <td>02/09/2014</td>\n",
       "      <td>TruSeq.v1</td>\n",
       "      <td>RNASEQ</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.983859</td>\n",
       "      <td>223040.0</td>\n",
       "      <td>0.971360</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.988930</td>\n",
       "      <td>0.767837</td>\n",
       "      <td>NaN</td>\n",
       "      <td>22193.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>76.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.002631</td>\n",
       "      <td>126.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>59839400.0</td>\n",
       "      <td>0.028310</td>\n",
       "      <td>173556.0</td>\n",
       "      <td>60509200.0</td>\n",
       "      <td>9407590.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>22218.0</td>\n",
       "      <td>29681000.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.203524</td>\n",
       "      <td>0.988930</td>\n",
       "      <td>0.759337</td>\n",
       "      <td>59839400.0</td>\n",
       "      <td>0.005929</td>\n",
       "      <td>12951200.0</td>\n",
       "      <td>6139790.0</td>\n",
       "      <td>13090100.0</td>\n",
       "      <td>125.0</td>\n",
       "      <td>13175400.0</td>\n",
       "      <td>11000300.0</td>\n",
       "      <td>0.004271</td>\n",
       "      <td>13150900.0</td>\n",
       "      <td>49.9535</td>\n",
       "      <td>0.002868</td>\n",
       "      <td>0.994001</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>50.2667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GTEX-1117F-0626-SM-5N9CS</th>\n",
       "      <td>1.0</td>\n",
       "      <td>B1</td>\n",
       "      <td>2 pieces, up to 4mm aderent fat/nerve/vessel, ...</td>\n",
       "      <td>6.9</td>\n",
       "      <td>Blood Vessel</td>\n",
       "      <td>Artery - Coronary</td>\n",
       "      <td>0001621</td>\n",
       "      <td>1243.0</td>\n",
       "      <td>1098.0</td>\n",
       "      <td>BP-43956</td>\n",
       "      <td>RNA Extraction from Paxgene-derived Lysate Pla...</td>\n",
       "      <td>09/25/2013</td>\n",
       "      <td>LCSET-4904</td>\n",
       "      <td>03/22/2014</td>\n",
       "      <td>TruSeq.v1</td>\n",
       "      <td>RNASEQ</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.990294</td>\n",
       "      <td>174501.0</td>\n",
       "      <td>0.970076</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.991275</td>\n",
       "      <td>0.769820</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23331.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>76.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.003053</td>\n",
       "      <td>136.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>69859200.0</td>\n",
       "      <td>0.029643</td>\n",
       "      <td>376087.0</td>\n",
       "      <td>70474000.0</td>\n",
       "      <td>3633120.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23354.0</td>\n",
       "      <td>34766500.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.200257</td>\n",
       "      <td>0.991275</td>\n",
       "      <td>0.763103</td>\n",
       "      <td>69859200.0</td>\n",
       "      <td>0.003671</td>\n",
       "      <td>15167600.0</td>\n",
       "      <td>6124860.0</td>\n",
       "      <td>15220100.0</td>\n",
       "      <td>151.0</td>\n",
       "      <td>15164200.0</td>\n",
       "      <td>11617900.0</td>\n",
       "      <td>0.003362</td>\n",
       "      <td>15291900.0</td>\n",
       "      <td>50.2096</td>\n",
       "      <td>0.005337</td>\n",
       "      <td>0.992257</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>50.0865</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GTEX-1117F-0726-SM-5GIEN</th>\n",
       "      <td>1.0</td>\n",
       "      <td>B1</td>\n",
       "      <td>2 pieces, no abnormalities</td>\n",
       "      <td>6.3</td>\n",
       "      <td>Heart</td>\n",
       "      <td>Heart - Atrial Appendage</td>\n",
       "      <td>0006631</td>\n",
       "      <td>1244.0</td>\n",
       "      <td>1097.0</td>\n",
       "      <td>BP-44261</td>\n",
       "      <td>RNA Extraction from Paxgene-derived Lysate Pla...</td>\n",
       "      <td>10/03/2013</td>\n",
       "      <td>LCSET-4796</td>\n",
       "      <td>02/27/2014</td>\n",
       "      <td>TruSeq.v1</td>\n",
       "      <td>RNASEQ</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.990734</td>\n",
       "      <td>184394.0</td>\n",
       "      <td>0.978830</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.993223</td>\n",
       "      <td>0.803478</td>\n",
       "      <td>NaN</td>\n",
       "      <td>22000.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>76.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.003114</td>\n",
       "      <td>131.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>62885700.0</td>\n",
       "      <td>0.020891</td>\n",
       "      <td>1936430.0</td>\n",
       "      <td>63314800.0</td>\n",
       "      <td>8252970.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>22015.0</td>\n",
       "      <td>31291500.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.175352</td>\n",
       "      <td>0.993223</td>\n",
       "      <td>0.798033</td>\n",
       "      <td>62885700.0</td>\n",
       "      <td>0.003772</td>\n",
       "      <td>13636600.0</td>\n",
       "      <td>9075650.0</td>\n",
       "      <td>13612800.0</td>\n",
       "      <td>146.0</td>\n",
       "      <td>13627100.0</td>\n",
       "      <td>9266070.0</td>\n",
       "      <td>0.003442</td>\n",
       "      <td>13756700.0</td>\n",
       "      <td>50.2367</td>\n",
       "      <td>0.030584</td>\n",
       "      <td>0.995711</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>49.9563</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          SMATSSCR SMCENTER  ... SMDPMPRT  SME2PCTS\n",
       "SAMPID                                       ...                   \n",
       "GTEX-1117F-0226-SM-5GZZ7       0.0       B1  ...      0.0   50.1944\n",
       "GTEX-1117F-0426-SM-5EGHI       0.0       B1  ...      0.0   49.9455\n",
       "GTEX-1117F-0526-SM-5EGHJ       0.0       B1  ...      0.0   50.2667\n",
       "GTEX-1117F-0626-SM-5N9CS       1.0       B1  ...      0.0   50.0865\n",
       "GTEX-1117F-0726-SM-5GIEN       1.0       B1  ...      0.0   49.9563\n",
       "\n",
       "[5 rows x 62 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('https://storage.googleapis.com/gtex_analysis_v8/rna_seq_data/GTEx_Analysis_2017-06-05_v8_RNASeQCv1.1.9_gene_tpm.gct.gz', skiprows=2, sep='\\t', index_col=0)\n",
    "df.index=[g[:15] for g in df.index]\n",
    "df=df[df.index.isin(pd.read_csv(\"http://stephenslab.github.io/count-clustering/project/utilities/gene_names_all_gtex.txt\", header=None).values.ravel())]\n",
    "df_files=pd.read_csv(\"https://storage.googleapis.com/gtex_analysis_v8/annotations/GTEx_Analysis_v8_Annotations_SampleAttributesDS.txt\", sep='\\t', index_col=0)\n",
    "df_files = df_files[df_files.index.isin(df.columns)]\n",
    "df_files.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "FGZXp-gSn22T"
   },
   "outputs": [],
   "source": [
    "df_files.to_csv(\"files.dat\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We ensure to have metadata for all files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 153
    },
    "colab_type": "code",
    "id": "nRdqQNR8tLbI",
    "outputId": "09f06f9e-c547-4ef1-b903-67bf3d8db81e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                 GTEX-1117F-0226-SM-5GZZ7  ...  GTEX-ZZPU-2726-SM-5NQ8O\n",
      "ENSG00000227232                    8.7640  ...                   2.1670\n",
      "ENSG00000238009                    0.0000  ...                   0.0212\n",
      "ENSG00000268903                    3.2150  ...                   6.2960\n",
      "ENSG00000269981                    7.0120  ...                  12.5900\n",
      "ENSG00000228463                    0.7821  ...                   2.7890\n",
      "\n",
      "[5 rows x 17382 columns]\n"
     ]
    }
   ],
   "source": [
    "df = df[df.columns[df.columns.isin(df_files.index)]]\n",
    "df.head(2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "i8M1VZiryCwS"
   },
   "source": [
    "### Sample columns\n",
    "We pick some samples at random. We get $100$ samples from 10 most represented tissues"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "-tsIh9S9LFN7"
   },
   "outputs": [],
   "source": [
    "rs = np.random.RandomState(seed=314)\n",
    "samples = pd.DataFrame(columns=df_files.columns)\n",
    "for site in df_files.groupby('SMTS').count().sort_values('SMTSD',ascending=False).index[:10]:\n",
    "    samples = samples.append(df_files[df_files['SMTS']==site].sample(100, random_state=rs))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 183
    },
    "colab_type": "code",
    "id": "m2WtMRH5LFN_",
    "outputId": "b4fe6ded-f767-4f1e-f4a5-ec739818450f"
   },
   "outputs": [],
   "source": [
    "# chack and verify the sampling process\n",
    "samples.groupby('SMTS').count().sort_values('SMTSD',ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "FYOnZVbZL2ZC"
   },
   "outputs": [],
   "source": [
    "df[samples.index].to_csv(\"mainTable_all.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "cSKLkFcaER9P"
   },
   "outputs": [],
   "source": [
    "df_files.to_csv(\"files.dat\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "dAmO1tHQLFOE"
   },
   "source": [
    "# Filter genes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Highly Variable Genes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We use scanpy to select $3000$ hvg"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "Uc32xZSlPCkc",
    "outputId": "6c6c9274-3ca4-42c3-8ebe-ae4b939a36b3"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1000 15177\n"
     ]
    }
   ],
   "source": [
    "samples = samples\n",
    "genes = df.index\n",
    "print(len(samples), len(genes))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "jVjvDFzBLFOF"
   },
   "outputs": [],
   "source": [
    "adata = sc.AnnData(X=df.reindex(index=genes, columns=samples.index).transpose(), obs=samples)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "R7wsXyxZLFOJ"
   },
   "outputs": [],
   "source": [
    "sc.pp.log1p(adata, copy=False)\n",
    "sc.pp.highly_variable_genes(adata, n_top_genes=3000, n_bins=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 296
    },
    "colab_type": "code",
    "id": "gdQauaSKLFOQ",
    "outputId": "8053d210-05ca-400b-aa64-9f20fa51021a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING: saving figure to file figures/filter_genes_dispersionhvg_tpm314.pdf\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sc.pl.highly_variable_genes(adata, log=False, save='hvg.pdf')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "sEo3X6pSLFOU"
   },
   "outputs": [],
   "source": [
    "hvg = adata.var[adata.var['highly_variable']==True].index\n",
    "samples = adata.obs.index"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "6Fxpu6Y6ylyI"
   },
   "source": [
    "Save a new table with only hvg"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "R4wI9T8PSq6i"
   },
   "outputs": [],
   "source": [
    "df.reindex(index=hvg, columns=samples).to_csv(\"mainTable_hv.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "s1GcJNV4yqKZ"
   },
   "source": [
    "We download [Human House Keeping genes](https://www.cell.com/trends/genetics/fulltext/S0168-9525(13)00089-9?_returnURL=https%3A%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS0168952513000899%3Fshowall%3Dtrue) from [https://www.tau.ac.il/~elieis/HKG/](https://www.tau.ac.il/~elieis/HKG/HK_exons.xlsx)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "GT-3FIkg22qo"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--2020-06-30 10:19:18--  https://www.tau.ac.il/~elieis/HKG/HK_exons.xlsx\n",
      "Resolving www.tau.ac.il (www.tau.ac.il)... 132.66.11.168\n",
      "Connecting to www.tau.ac.il (www.tau.ac.il)|132.66.11.168|:443... connected.\n",
      "HTTP request sent, awaiting response... 200 OK\n",
      "Length: 5709370 (5.4M) [application/vnd.openxmlformats-officedocument.spreadsheetml.sheet]\n",
      "Saving to: ‘HK_exons.xlsx.1’\n",
      "\n",
      "HK_exons.xlsx.1     100%[===================>]   5.44M   598KB/s    in 14s     \n",
      "\n",
      "2020-06-30 10:19:33 (404 KB/s) - ‘HK_exons.xlsx.1’ saved [5709370/5709370]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "!wget https://www.tau.ac.il/~elieis/HKG/HK_exons.xlsx\n",
    "hk = pd.read_excel(\"HK_exons.xlsx\")[\"Gene Name\"].unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "vgoikocwzU3z"
   },
   "source": [
    "Here we use [https://www.genenames.org](https://www.genenames.org) to convert  genes' names to Ensamble' ids"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "IyttG6Sr22pU"
   },
   "outputs": [],
   "source": [
    "hgcn_url = \"https://www.genenames.org/cgi-bin/download/custom?col=gd_hgnc_id&col=gd_app_sym&col=gd_app_name&col=md_ensembl_id&status=Approved&status=Entry%20Withdrawn&hgnc_dbtag=on&order_by=gd_app_name&format=text&submit=submit\"\n",
    "df_conversion=pd.read_csv(hgcn_url, sep=\"\\t\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "NJQnZfInEREP"
   },
   "outputs": [],
   "source": [
    "df_hk_ensg = df_conversion[(df_conversion[\"Approved symbol\"].isin(hk))&(df_conversion[\"Ensembl ID(supplied by Ensembl)\"].isin(df.index))][\"Ensembl ID(supplied by Ensembl)\"].drop_duplicates().values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "xjQm3_p3IQo1"
   },
   "outputs": [],
   "source": [
    "df_hk = df.reindex(index=df_hk_ensg)\n",
    "df_hk.to_csv(\"mainTable_hk.csv\", index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "WJw8LIAuLFOa"
   },
   "source": [
    "# Run hierarchical Stochastic Block Model\n",
    "We run [stochastic block model](https://github.com/martingerlach/hSBM_Topicmodel/tree/develop).\n",
    "\n",
    "Clone the repository with the code and add it to the system path\n",
    "\n",
    "```bash\n",
    "git clone --branch develop https://github.com/martingerlach/hSBM_Topicmodel/\n",
    "```\n",
    "\n",
    "Note: [graph-tool](https://graph-tool.skewed.de/) should be installed and correctly configured"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cloning into 'hSBM_Topicmodel'...\n",
      "remote: Enumerating objects: 12, done.\u001b[K\n",
      "remote: Counting objects: 100% (12/12), done.\u001b[K\n",
      "remote: Compressing objects: 100% (12/12), done.\u001b[K\n",
      "remote: Total 260 (delta 4), reused 5 (delta 0), pack-reused 248\u001b[K\n",
      "Receiving objects: 100% (260/260), 3.61 MiB | 3.04 MiB/s, done.\n",
      "Resolving deltas: 100% (60/60), done.\n"
     ]
    }
   ],
   "source": [
    "!git clone --branch develop https://github.com/martingerlach/hSBM_Topicmodel/\n",
    "sys.path.append(\"hSBM_Topicmodel\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 153
    },
    "colab_type": "code",
    "id": "wc8eyP_LLFOb",
    "outputId": "21ec4c0a-5d64-48a9-a7ca-0f8930553d12"
   },
   "outputs": [],
   "source": [
    "from sbmtm import sbmtm\n",
    "\n",
    "# create the model\n",
    "model = sbmtm()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "9IWe7zvuLFOg",
    "outputId": "adc3e5ce-089d-43fe-d2a7-15099c7d261b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1000 3000\n"
     ]
    }
   ],
   "source": [
    "samples = samples\n",
    "genes = hvg\n",
    "print(len(samples), len(genes))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Put the data into the model\n",
    "\n",
    "You can choose to:\n",
    "- **create** a model from raw data\n",
    "- **load** a network provided in the form of a *graph.xml.gz* file\n",
    "- load a **pretrained** model stored in a *topsbm.pkl* file"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "viYv-ehPtIzq"
   },
   "source": [
    "### make a graph with the data\n",
    "\n",
    "Skip this if you want to use a *graph.xml.gz* file already provided"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "IqIVBZIJDIY8"
   },
   "outputs": [],
   "source": [
    "df = pd.read_csv(\"mainTable.csv\", index_col=0)\n",
    "genes = df.index\n",
    "samples = pd.Series(index=df.columns)\n",
    "print(len(samples), len(genes))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "awX5gbUoLFOr",
    "outputId": "d9d59938-1a93-450a-d796-303b0ae3cca1"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Graph object, undirected, with 4000 vertices and 2032164 edges, 2 internal vertex properties, 1 internal edge property, at 0x7fa2a88a5f98>"
      ]
     },
     "execution_count": 36,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.make_graph_from_BoW_df(df.reindex(index=genes, columns=samples.index).dropna())\n",
    "\n",
    "model.save_graph(\"graph.xml.gz\")\n",
    "model.g"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "ULe4g3qTtLFY"
   },
   "source": [
    "### load data\n",
    "Use this only if you **already have a *graph.xml.gz*** file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "VPMxekzvLFO1"
   },
   "outputs": [],
   "source": [
    "# load graph\n",
    "model.load_graph(\"graph.xml.gz\")\n",
    "model.g"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Ldd2ZxostNve"
   },
   "source": [
    "### load pretrained\n",
    "Use this only if you **already have a *topsbm.pkl*** file with a trained model "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "blowKA-nzRal"
   },
   "outputs": [],
   "source": [
    "# load pretrainded\n",
    "import graph_tool as gt\n",
    "import pickle\n",
    "\n",
    "with open(\"topsbm/topsbm.pkl\", \"rb\") as file:\n",
    "    model = pickle.load(file)\n",
    "\n",
    "model.g"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "QcdDMsOdtTPm"
   },
   "source": [
    "## Run the model\n",
    "\n",
    "This may take a long time and a lot of memory, think twice before running it in the local machine"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "9Phpa4YLWzuv"
   },
   "outputs": [],
   "source": [
    "config = \"breast\"\n",
    "os.system(f\"mkdir -p {config} && mkdir -p {config}/topsbm\")\n",
    "os.chdir(f\"{config}/topsbm\")\n",
    "\n",
    "model.fit(n_init=10, parallel=True, verbose=True)\n",
    "model.save_data()"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [
    "EjZgH93w2wRm",
    "ULe4g3qTtLFY",
    "Ldd2ZxostNve"
   ],
   "machine_shape": "hm",
   "name": "run_graph.ipynb",
   "provenance": [],
   "toc_visible": true
  },
  "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": "3.8.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}