fig1_quick_viz_pv.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Quick vizualisation of the PV procedure\n",
    "This notebook support the figure 1. Figure 1 gives an overview on the method. It has been realised using solely breast cancer data and the package. Sparse PCA has been used instead of PCA to make the factors sparse and thus easier to understand visually."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Tissue to consider\n",
    "tumor_type = 'Breast'\n",
    "cell_line_type = 'BRCA'\n",
    "pdx_type = 'BRCA'\n",
    "\n",
    "# Normalization parameters\n",
    "normalization = 'TMM'\n",
    "transformation = 'log'\n",
    "mean_center = True\n",
    "std_unit = False\n",
    "protein_coding_only = True\n",
    "filter_mytochondrial = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os, sys\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import matplotlib\n",
    "from sklearn.decomposition import PCA, FastICA, SparsePCA\n",
    "import scipy\n",
    "from scipy.stats import ks_2samp\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "#Import src implementations\n",
    "os.environ['OMP_NUM_THREADS'] = '3'\n",
    "os.environ['KMP_DUPLICATE_LIB_OK']='True'\n",
    "from data_reader.read_data import read_data\n",
    "from normalization_methods.feature_engineering import feature_engineering\n",
    "from precise import PVComputation\n",
    "from precise import PVComputation, IntermediateFactors, ConsensusRepresentation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Import data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Count data imported\n"
     ]
    }
   ],
   "source": [
    "# Import tumor + cell line data (RNA-seq)\n",
    "X_target, X_source, gene_names, source_names, target_names = read_data('cell_line',\n",
    "                                                                       'tumor',\n",
    "                                                                       'count',\n",
    "                                                                        None if cell_line_type == 'All' else cell_line_type,\n",
    "                                                                        tumor_type,\n",
    "                                                                        filter_mytochondrial)\n",
    "print('Count data imported')\n",
    "\n",
    "X_target = feature_engineering(X_target, normalization, transformation, mean_center, std_unit)\n",
    "X_source = feature_engineering(X_source, normalization, transformation, mean_center, std_unit)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Filter on top target genes\n",
    "number_top_genes = 100\n",
    "\n",
    "top_genes = np.argsort(np.var(X_target, 0))[::-1][:number_top_genes]\n",
    "\n",
    "X_target_filtered = X_target[:,top_genes]\n",
    "X_source_filtered = X_source[:, top_genes]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Compute PCs and PVs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_factors = 5\n",
    "n_pv = 5\n",
    "dim_reduction = 'pca'\n",
    "dim_reduction_target = 'pca'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.7/site-packages/sklearn/decomposition/sparse_pca.py:170: DeprecationWarning: normalize_components=False is a backward-compatible setting that implements a non-standard definition of sparse PCA. This compatibility mode will be removed in 0.22.\n",
      "  DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration   0 (elapsed time:   0s,  0.0mn, current cost     nan)\n",
      "Iteration   1 (elapsed time:   0s,  0.0mn, current cost  12132.927)\n",
      "Iteration   2 (elapsed time:   0s,  0.0mn, current cost  7472.137)\n",
      "Iteration   3 (elapsed time:   0s,  0.0mn, current cost  7461.421)\n",
      "Iteration   4 (elapsed time:   0s,  0.0mn, current cost  7453.444)\n",
      "Iteration   5 (elapsed time:   0s,  0.0mn, current cost  7446.284)\n",
      "Iteration   6 (elapsed time:   0s,  0.0mn, current cost  7439.600)\n",
      "Iteration   7 (elapsed time:   0s,  0.0mn, current cost  7432.972)\n",
      "Iteration   8 (elapsed time:   0s,  0.0mn, current cost  7426.197)\n",
      "Iteration   9 (elapsed time:   0s,  0.0mn, current cost  7419.219)\n",
      "Iteration  10 (elapsed time:   0s,  0.0mn, current cost  7411.950)\n",
      "Iteration  11 (elapsed time:   0s,  0.0mn, current cost  7404.993)\n",
      "Iteration  12 (elapsed time:   0s,  0.0mn, current cost  7399.133)\n",
      "Iteration  13 (elapsed time:   0s,  0.0mn, current cost  7394.347)\n",
      "Iteration  14 (elapsed time:   0s,  0.0mn, current cost  7390.406)\n",
      "Iteration  15 (elapsed time:   0s,  0.0mn, current cost  7387.270)\n",
      "Iteration  16 (elapsed time:   0s,  0.0mn, current cost  7384.752)\n",
      "Iteration  17 (elapsed time:   0s,  0.0mn, current cost  7382.620)\n",
      "Iteration  18 (elapsed time:   0s,  0.0mn, current cost  7380.805)\n",
      "Iteration  19 (elapsed time:   0s,  0.0mn, current cost  7379.237)\n",
      "Iteration  20 (elapsed time:   1s,  0.0mn, current cost  7377.843)\n",
      "Iteration  21 (elapsed time:   1s,  0.0mn, current cost  7376.564)\n",
      "Iteration  22 (elapsed time:   1s,  0.0mn, current cost  7375.367)\n",
      "Iteration  23 (elapsed time:   1s,  0.0mn, current cost  7374.235)\n",
      "Iteration  24 (elapsed time:   1s,  0.0mn, current cost  7373.162)\n",
      "Iteration  25 (elapsed time:   1s,  0.0mn, current cost  7372.156)\n",
      "Iteration  26 (elapsed time:   1s,  0.0mn, current cost  7371.204)\n",
      "Iteration  27 (elapsed time:   1s,  0.0mn, current cost  7370.273)\n",
      "Iteration  28 (elapsed time:   1s,  0.0mn, current cost  7369.373)\n",
      "Iteration  29 (elapsed time:   1s,  0.0mn, current cost  7368.518)\n",
      "Iteration  30 (elapsed time:   1s,  0.0mn, current cost  7367.693)\n",
      "Iteration  31 (elapsed time:   1s,  0.0mn, current cost  7366.898)\n",
      "Iteration  32 (elapsed time:   1s,  0.0mn, current cost  7366.108)\n",
      "Iteration  33 (elapsed time:   1s,  0.0mn, current cost  7365.280)\n",
      "Iteration  34 (elapsed time:   1s,  0.0mn, current cost  7364.393)\n",
      "Iteration  35 (elapsed time:   1s,  0.0mn, current cost  7363.442)\n",
      "Iteration  36 (elapsed time:   1s,  0.0mn, current cost  7362.471)\n",
      "Iteration  37 (elapsed time:   1s,  0.0mn, current cost  7361.549)\n",
      "Iteration  38 (elapsed time:   1s,  0.0mn, current cost  7360.708)\n",
      "Iteration  39 (elapsed time:   1s,  0.0mn, current cost  7359.964)\n",
      "Iteration  40 (elapsed time:   1s,  0.0mn, current cost  7359.304)\n",
      "Iteration  41 (elapsed time:   1s,  0.0mn, current cost  7358.707)\n",
      "Iteration  42 (elapsed time:   2s,  0.0mn, current cost  7358.157)\n",
      "Iteration  43 (elapsed time:   2s,  0.0mn, current cost  7357.635)\n",
      "Iteration  44 (elapsed time:   2s,  0.0mn, current cost  7357.131)\n",
      "Iteration  45 (elapsed time:   2s,  0.0mn, current cost  7356.638)\n",
      "Iteration  46 (elapsed time:   2s,  0.0mn, current cost  7356.156)\n",
      "Iteration  47 (elapsed time:   2s,  0.0mn, current cost  7355.674)\n",
      "Iteration  48 (elapsed time:   2s,  0.0mn, current cost  7355.167)\n",
      "Iteration  49 (elapsed time:   2s,  0.0mn, current cost  7354.617)\n",
      "Iteration  50 (elapsed time:   2s,  0.0mn, current cost  7354.011)\n",
      "Iteration  51 (elapsed time:   2s,  0.0mn, current cost  7353.383)\n",
      "Iteration  52 (elapsed time:   2s,  0.0mn, current cost  7352.764)\n",
      "Iteration  53 (elapsed time:   2s,  0.0mn, current cost  7352.138)\n",
      "Iteration  54 (elapsed time:   2s,  0.0mn, current cost  7351.543)\n",
      "Iteration  55 (elapsed time:   2s,  0.0mn, current cost  7351.002)\n",
      "Iteration  56 (elapsed time:   2s,  0.0mn, current cost  7350.508)\n",
      "Iteration  57 (elapsed time:   2s,  0.0mn, current cost  7350.056)\n",
      "Iteration  58 (elapsed time:   2s,  0.0mn, current cost  7349.648)\n",
      "Iteration  59 (elapsed time:   2s,  0.0mn, current cost  7349.274)\n",
      "Iteration  60 (elapsed time:   2s,  0.0mn, current cost  7348.929)\n",
      "Iteration  61 (elapsed time:   2s,  0.0mn, current cost  7348.608)\n",
      "Iteration  62 (elapsed time:   2s,  0.0mn, current cost  7348.305)\n",
      "Iteration  63 (elapsed time:   2s,  0.0mn, current cost  7348.017)\n",
      "Iteration  64 (elapsed time:   2s,  0.0mn, current cost  7347.742)\n",
      "Iteration  65 (elapsed time:   2s,  0.0mn, current cost  7347.477)\n",
      "Iteration  66 (elapsed time:   3s,  0.1mn, current cost  7347.224)\n",
      "Iteration  67 (elapsed time:   3s,  0.1mn, current cost  7346.984)\n",
      "Iteration  68 (elapsed time:   3s,  0.1mn, current cost  7346.754)\n",
      "Iteration  69 (elapsed time:   3s,  0.1mn, current cost  7346.533)\n",
      "Iteration  70 (elapsed time:   3s,  0.1mn, current cost  7346.322)\n",
      "Iteration  71 (elapsed time:   3s,  0.1mn, current cost  7346.120)\n",
      "Iteration  72 (elapsed time:   3s,  0.1mn, current cost  7345.929)\n",
      "Iteration  73 (elapsed time:   3s,  0.1mn, current cost  7345.749)\n",
      "Iteration  74 (elapsed time:   3s,  0.1mn, current cost  7345.580)\n",
      "Iteration  75 (elapsed time:   3s,  0.1mn, current cost  7345.421)\n",
      "Iteration  76 (elapsed time:   3s,  0.1mn, current cost  7345.271)\n",
      "Iteration  77 (elapsed time:   3s,  0.1mn, current cost  7345.131)\n",
      "Iteration  78 (elapsed time:   3s,  0.1mn, current cost  7345.000)\n",
      "Iteration  79 (elapsed time:   3s,  0.1mn, current cost  7344.879)\n",
      "Iteration  80 (elapsed time:   3s,  0.1mn, current cost  7344.763)\n",
      "Iteration  81 (elapsed time:   3s,  0.1mn, current cost  7344.647)\n",
      "Iteration  82 (elapsed time:   3s,  0.1mn, current cost  7344.528)\n",
      "Iteration  83 (elapsed time:   3s,  0.1mn, current cost  7344.407)\n",
      "Iteration  84 (elapsed time:   3s,  0.1mn, current cost  7344.283)\n",
      "Iteration  85 (elapsed time:   3s,  0.1mn, current cost  7344.154)\n",
      "Iteration  86 (elapsed time:   3s,  0.1mn, current cost  7344.022)\n",
      "Iteration  87 (elapsed time:   3s,  0.1mn, current cost  7343.894)\n",
      "Iteration  88 (elapsed time:   3s,  0.1mn, current cost  7343.775)\n",
      "Iteration  89 (elapsed time:   3s,  0.1mn, current cost  7343.672)\n",
      "Iteration  90 (elapsed time:   4s,  0.1mn, current cost  7343.582)\n",
      "Iteration  91 (elapsed time:   4s,  0.1mn, current cost  7343.503)\n",
      "Iteration  92 (elapsed time:   4s,  0.1mn, current cost  7343.432)\n",
      "Iteration  93 (elapsed time:   4s,  0.1mn, current cost  7343.370)\n",
      "Iteration  94 (elapsed time:   4s,  0.1mn, current cost  7343.314)\n",
      "Iteration  95 (elapsed time:   4s,  0.1mn, current cost  7343.263)\n",
      "Iteration  96 (elapsed time:   4s,  0.1mn, current cost  7343.218)\n",
      "Iteration  97 (elapsed time:   4s,  0.1mn, current cost  7343.176)\n",
      "Iteration  98 (elapsed time:   4s,  0.1mn, current cost  7343.138)\n",
      "Iteration  99 (elapsed time:   4s,  0.1mn, current cost  7343.104)\n",
      "Iteration  100 (elapsed time:   4s,  0.1mn, current cost  7343.073)\n",
      "Iteration  101 (elapsed time:   4s,  0.1mn, current cost  7343.047)\n",
      "Iteration  102 (elapsed time:   4s,  0.1mn, current cost  7343.025)\n",
      "Iteration  103 (elapsed time:   4s,  0.1mn, current cost  7343.006)\n",
      "Iteration  104 (elapsed time:   4s,  0.1mn, current cost  7342.989)\n",
      "Iteration  105 (elapsed time:   4s,  0.1mn, current cost  7342.975)\n",
      "Iteration  106 (elapsed time:   4s,  0.1mn, current cost  7342.962)\n",
      "Iteration  107 (elapsed time:   4s,  0.1mn, current cost  7342.951)\n",
      "Iteration  108 (elapsed time:   4s,  0.1mn, current cost  7342.941)\n",
      "Iteration  109 (elapsed time:   4s,  0.1mn, current cost  7342.932)\n",
      "Iteration  110 (elapsed time:   4s,  0.1mn, current cost  7342.923)\n",
      "Iteration  111 (elapsed time:   4s,  0.1mn, current cost  7342.916)\n",
      "Iteration  112 (elapsed time:   5s,  0.1mn, current cost  7342.908)\n",
      "Iteration  113 (elapsed time:   5s,  0.1mn, current cost  7342.901)\n",
      "Iteration  114 (elapsed time:   5s,  0.1mn, current cost  7342.895)\n",
      "Iteration  115 (elapsed time:   5s,  0.1mn, current cost  7342.889)\n",
      "Iteration  116 (elapsed time:   5s,  0.1mn, current cost  7342.883)\n",
      "Iteration  117 (elapsed time:   5s,  0.1mn, current cost  7342.877)\n",
      "Iteration  118 (elapsed time:   5s,  0.1mn, current cost  7342.871)\n",
      "Iteration  119 (elapsed time:   5s,  0.1mn, current cost  7342.865)\n",
      "Iteration  120 (elapsed time:   5s,  0.1mn, current cost  7342.860)\n",
      "Iteration  121 (elapsed time:   5s,  0.1mn, current cost  7342.854)\n",
      "Iteration  122 (elapsed time:   5s,  0.1mn, current cost  7342.849)\n",
      "Iteration  123 (elapsed time:   5s,  0.1mn, current cost  7342.843)\n",
      "Iteration  124 (elapsed time:   5s,  0.1mn, current cost  7342.837)\n",
      "Iteration  125 (elapsed time:   5s,  0.1mn, current cost  7342.832)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration  126 (elapsed time:   5s,  0.1mn, current cost  7342.827)\n",
      "Iteration  127 (elapsed time:   5s,  0.1mn, current cost  7342.822)\n",
      "Iteration  128 (elapsed time:   5s,  0.1mn, current cost  7342.817)\n",
      "Iteration  129 (elapsed time:   5s,  0.1mn, current cost  7342.812)\n",
      "Iteration  130 (elapsed time:   5s,  0.1mn, current cost  7342.807)\n",
      "Iteration  131 (elapsed time:   5s,  0.1mn, current cost  7342.803)\n",
      "Iteration  132 (elapsed time:   5s,  0.1mn, current cost  7342.798)\n",
      "Iteration  133 (elapsed time:   5s,  0.1mn, current cost  7342.794)\n",
      "Iteration  134 (elapsed time:   5s,  0.1mn, current cost  7342.790)\n",
      "Iteration  135 (elapsed time:   5s,  0.1mn, current cost  7342.785)\n",
      "Iteration  136 (elapsed time:   6s,  0.1mn, current cost  7342.781)\n",
      "Iteration  137 (elapsed time:   6s,  0.1mn, current cost  7342.777)\n",
      "Iteration  138 (elapsed time:   6s,  0.1mn, current cost  7342.773)\n",
      "Iteration  139 (elapsed time:   6s,  0.1mn, current cost  7342.769)\n",
      "Iteration  140 (elapsed time:   6s,  0.1mn, current cost  7342.766)\n",
      "Iteration  141 (elapsed time:   6s,  0.1mn, current cost  7342.762)\n",
      "Iteration  142 (elapsed time:   6s,  0.1mn, current cost  7342.758)\n",
      "Iteration  143 (elapsed time:   6s,  0.1mn, current cost  7342.754)\n",
      "Iteration  144 (elapsed time:   6s,  0.1mn, current cost  7342.751)\n",
      "Iteration  145 (elapsed time:   6s,  0.1mn, current cost  7342.747)\n",
      "Iteration  146 (elapsed time:   6s,  0.1mn, current cost  7342.744)\n",
      "Iteration  147 (elapsed time:   6s,  0.1mn, current cost  7342.740)\n",
      "Iteration  148 (elapsed time:   6s,  0.1mn, current cost  7342.737)\n",
      "Iteration  149 (elapsed time:   6s,  0.1mn, current cost  7342.733)\n",
      "Iteration  150 (elapsed time:   6s,  0.1mn, current cost  7342.730)\n",
      "Iteration  151 (elapsed time:   6s,  0.1mn, current cost  7342.726)\n",
      "Iteration  152 (elapsed time:   6s,  0.1mn, current cost  7342.723)\n",
      "Iteration  153 (elapsed time:   6s,  0.1mn, current cost  7342.720)\n",
      "Iteration  154 (elapsed time:   6s,  0.1mn, current cost  7342.716)\n",
      "Iteration  155 (elapsed time:   6s,  0.1mn, current cost  7342.713)\n",
      "Iteration  156 (elapsed time:   6s,  0.1mn, current cost  7342.710)\n",
      "Iteration  157 (elapsed time:   6s,  0.1mn, current cost  7342.707)\n",
      "Iteration  158 (elapsed time:   6s,  0.1mn, current cost  7342.704)\n",
      "Iteration  159 (elapsed time:   6s,  0.1mn, current cost  7342.701)\n",
      "Iteration  160 (elapsed time:   7s,  0.1mn, current cost  7342.697)\n",
      "Iteration  161 (elapsed time:   7s,  0.1mn, current cost  7342.694)\n",
      "Iteration  162 (elapsed time:   7s,  0.1mn, current cost  7342.691)\n",
      "Iteration  163 (elapsed time:   7s,  0.1mn, current cost  7342.688)\n",
      "Iteration  164 (elapsed time:   7s,  0.1mn, current cost  7342.685)\n",
      "Iteration  165 (elapsed time:   7s,  0.1mn, current cost  7342.682)\n",
      "Iteration  166 (elapsed time:   7s,  0.1mn, current cost  7342.679)\n",
      "Iteration  167 (elapsed time:   7s,  0.1mn, current cost  7342.676)\n",
      "Iteration  168 (elapsed time:   7s,  0.1mn, current cost  7342.673)\n",
      "Iteration  169 (elapsed time:   7s,  0.1mn, current cost  7342.670)\n",
      "Iteration  170 (elapsed time:   7s,  0.1mn, current cost  7342.666)\n",
      "Iteration  171 (elapsed time:   7s,  0.1mn, current cost  7342.663)\n",
      "Iteration  172 (elapsed time:   7s,  0.1mn, current cost  7342.660)\n",
      "Iteration  173 (elapsed time:   7s,  0.1mn, current cost  7342.657)\n",
      "Iteration  174 (elapsed time:   7s,  0.1mn, current cost  7342.654)\n",
      "Iteration  175 (elapsed time:   7s,  0.1mn, current cost  7342.651)\n",
      "Iteration  176 (elapsed time:   7s,  0.1mn, current cost  7342.647)\n",
      "Iteration  177 (elapsed time:   7s,  0.1mn, current cost  7342.644)\n",
      "Iteration  178 (elapsed time:   7s,  0.1mn, current cost  7342.641)\n",
      "Iteration  179 (elapsed time:   7s,  0.1mn, current cost  7342.637)\n",
      "Iteration  180 (elapsed time:   7s,  0.1mn, current cost  7342.634)\n",
      "Iteration  181 (elapsed time:   7s,  0.1mn, current cost  7342.631)\n",
      "Iteration  182 (elapsed time:   7s,  0.1mn, current cost  7342.627)\n",
      "Iteration  183 (elapsed time:   8s,  0.1mn, current cost  7342.624)\n",
      "Iteration  184 (elapsed time:   8s,  0.1mn, current cost  7342.620)\n",
      "Iteration  185 (elapsed time:   8s,  0.1mn, current cost  7342.617)\n",
      "Iteration  186 (elapsed time:   8s,  0.1mn, current cost  7342.614)\n",
      "Iteration  187 (elapsed time:   8s,  0.1mn, current cost  7342.610)\n",
      "Iteration  188 (elapsed time:   8s,  0.1mn, current cost  7342.607)\n",
      "Iteration  189 (elapsed time:   8s,  0.1mn, current cost  7342.604)\n",
      "Iteration  190 (elapsed time:   8s,  0.1mn, current cost  7342.600)\n",
      "Iteration  191 (elapsed time:   8s,  0.1mn, current cost  7342.597)\n",
      "Iteration  192 (elapsed time:   8s,  0.1mn, current cost  7342.594)\n",
      "Iteration  193 (elapsed time:   8s,  0.1mn, current cost  7342.590)\n",
      "Iteration  194 (elapsed time:   8s,  0.1mn, current cost  7342.587)\n",
      "Iteration  195 (elapsed time:   8s,  0.1mn, current cost  7342.584)\n",
      "Iteration  196 (elapsed time:   8s,  0.1mn, current cost  7342.581)\n",
      "Iteration  197 (elapsed time:   8s,  0.1mn, current cost  7342.578)\n",
      "Iteration  198 (elapsed time:   8s,  0.1mn, current cost  7342.574)\n",
      "Iteration  199 (elapsed time:   8s,  0.1mn, current cost  7342.571)\n",
      "Iteration  200 (elapsed time:   8s,  0.1mn, current cost  7342.568)\n",
      "Iteration  201 (elapsed time:   8s,  0.1mn, current cost  7342.565)\n",
      "Iteration  202 (elapsed time:   8s,  0.1mn, current cost  7342.562)\n",
      "Iteration  203 (elapsed time:   8s,  0.1mn, current cost  7342.559)\n",
      "Iteration  204 (elapsed time:   8s,  0.1mn, current cost  7342.556)\n",
      "Iteration  205 (elapsed time:   8s,  0.1mn, current cost  7342.553)\n",
      "Iteration  206 (elapsed time:   9s,  0.2mn, current cost  7342.550)\n",
      "Iteration  207 (elapsed time:   9s,  0.2mn, current cost  7342.547)\n",
      "Iteration  208 (elapsed time:   9s,  0.2mn, current cost  7342.544)\n",
      "Iteration  209 (elapsed time:   9s,  0.2mn, current cost  7342.541)\n",
      "Iteration  210 (elapsed time:   9s,  0.2mn, current cost  7342.538)\n",
      "Iteration  211 (elapsed time:   9s,  0.2mn, current cost  7342.535)\n",
      "Iteration  212 (elapsed time:   9s,  0.2mn, current cost  7342.532)\n",
      "Iteration  213 (elapsed time:   9s,  0.2mn, current cost  7342.529)\n",
      "Iteration  214 (elapsed time:   9s,  0.2mn, current cost  7342.525)\n",
      "Iteration  215 (elapsed time:   9s,  0.2mn, current cost  7342.522)\n",
      "Iteration  216 (elapsed time:   9s,  0.2mn, current cost  7342.519)\n",
      "Iteration  217 (elapsed time:   9s,  0.2mn, current cost  7342.516)\n",
      "Iteration  218 (elapsed time:   9s,  0.2mn, current cost  7342.513)\n",
      "Iteration  219 (elapsed time:   9s,  0.2mn, current cost  7342.510)\n",
      "Iteration  220 (elapsed time:   9s,  0.2mn, current cost  7342.506)\n",
      "Iteration  221 (elapsed time:   9s,  0.2mn, current cost  7342.503)\n",
      "Iteration  222 (elapsed time:   9s,  0.2mn, current cost  7342.500)\n",
      "Iteration  223 (elapsed time:   9s,  0.2mn, current cost  7342.496)\n",
      "Iteration  224 (elapsed time:   9s,  0.2mn, current cost  7342.493)\n",
      "Iteration  225 (elapsed time:   9s,  0.2mn, current cost  7342.490)\n",
      "Iteration  226 (elapsed time:   9s,  0.2mn, current cost  7342.486)\n",
      "Iteration  227 (elapsed time:  10s,  0.2mn, current cost  7342.483)\n",
      "Iteration  228 (elapsed time:  10s,  0.2mn, current cost  7342.480)\n",
      "Iteration  229 (elapsed time:  10s,  0.2mn, current cost  7342.476)\n",
      "Iteration  230 (elapsed time:  10s,  0.2mn, current cost  7342.473)\n",
      "Iteration  231 (elapsed time:  10s,  0.2mn, current cost  7342.469)\n",
      "Iteration  232 (elapsed time:  10s,  0.2mn, current cost  7342.466)\n",
      "Iteration  233 (elapsed time:  10s,  0.2mn, current cost  7342.462)\n",
      "Iteration  234 (elapsed time:  10s,  0.2mn, current cost  7342.459)\n",
      "Iteration  235 (elapsed time:  10s,  0.2mn, current cost  7342.455)\n",
      "Iteration  236 (elapsed time:  10s,  0.2mn, current cost  7342.451)\n",
      "Iteration  237 (elapsed time:  10s,  0.2mn, current cost  7342.448)\n",
      "Iteration  238 (elapsed time:  10s,  0.2mn, current cost  7342.444)\n",
      "Iteration  239 (elapsed time:  10s,  0.2mn, current cost  7342.441)\n",
      "Iteration  240 (elapsed time:  10s,  0.2mn, current cost  7342.437)\n",
      "Iteration  241 (elapsed time:  10s,  0.2mn, current cost  7342.433)\n",
      "Iteration  242 (elapsed time:  10s,  0.2mn, current cost  7342.429)\n",
      "Iteration  243 (elapsed time:  10s,  0.2mn, current cost  7342.426)\n",
      "Iteration  244 (elapsed time:  11s,  0.2mn, current cost  7342.422)\n",
      "Iteration  245 (elapsed time:  11s,  0.2mn, current cost  7342.418)\n",
      "Iteration  246 (elapsed time:  11s,  0.2mn, current cost  7342.414)\n",
      "Iteration  247 (elapsed time:  11s,  0.2mn, current cost  7342.411)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration  248 (elapsed time:  11s,  0.2mn, current cost  7342.407)\n",
      "Iteration  249 (elapsed time:  11s,  0.2mn, current cost  7342.403)\n",
      "Iteration  250 (elapsed time:  11s,  0.2mn, current cost  7342.399)\n",
      "Iteration  251 (elapsed time:  11s,  0.2mn, current cost  7342.396)\n",
      "Iteration  252 (elapsed time:  11s,  0.2mn, current cost  7342.392)\n",
      "Iteration  253 (elapsed time:  11s,  0.2mn, current cost  7342.388)\n",
      "Iteration  254 (elapsed time:  11s,  0.2mn, current cost  7342.384)\n",
      "Iteration  255 (elapsed time:  11s,  0.2mn, current cost  7342.381)\n",
      "Iteration  256 (elapsed time:  11s,  0.2mn, current cost  7342.377)\n",
      "Iteration  257 (elapsed time:  11s,  0.2mn, current cost  7342.373)\n",
      "Iteration  258 (elapsed time:  11s,  0.2mn, current cost  7342.369)\n",
      "Iteration  259 (elapsed time:  12s,  0.2mn, current cost  7342.366)\n",
      "Iteration  260 (elapsed time:  12s,  0.2mn, current cost  7342.362)\n",
      "Iteration  261 (elapsed time:  12s,  0.2mn, current cost  7342.358)\n",
      "Iteration  262 (elapsed time:  12s,  0.2mn, current cost  7342.354)\n",
      "Iteration  263 (elapsed time:  12s,  0.2mn, current cost  7342.350)\n",
      "Iteration  264 (elapsed time:  12s,  0.2mn, current cost  7342.347)\n",
      "Iteration  265 (elapsed time:  12s,  0.2mn, current cost  7342.343)\n",
      "Iteration  266 (elapsed time:  12s,  0.2mn, current cost  7342.339)\n",
      "Iteration  267 (elapsed time:  12s,  0.2mn, current cost  7342.335)\n",
      "Iteration  268 (elapsed time:  12s,  0.2mn, current cost  7342.331)\n",
      "Iteration  269 (elapsed time:  12s,  0.2mn, current cost  7342.328)\n",
      "Iteration  270 (elapsed time:  12s,  0.2mn, current cost  7342.324)\n",
      "Iteration  271 (elapsed time:  12s,  0.2mn, current cost  7342.321)\n",
      "Iteration  272 (elapsed time:  12s,  0.2mn, current cost  7342.318)\n",
      "Iteration  273 (elapsed time:  12s,  0.2mn, current cost  7342.315)\n",
      "Iteration  274 (elapsed time:  12s,  0.2mn, current cost  7342.312)\n",
      "Iteration  275 (elapsed time:  13s,  0.2mn, current cost  7342.309)\n",
      "Iteration  276 (elapsed time:  13s,  0.2mn, current cost  7342.306)\n",
      "Iteration  277 (elapsed time:  13s,  0.2mn, current cost  7342.304)\n",
      "Iteration  278 (elapsed time:  13s,  0.2mn, current cost  7342.302)\n",
      "Iteration  279 (elapsed time:  13s,  0.2mn, current cost  7342.300)\n",
      "Iteration  280 (elapsed time:  13s,  0.2mn, current cost  7342.298)\n",
      "Iteration  281 (elapsed time:  13s,  0.2mn, current cost  7342.296)\n",
      "Iteration  282 (elapsed time:  13s,  0.2mn, current cost  7342.295)\n",
      "Iteration  283 (elapsed time:  13s,  0.2mn, current cost  7342.293)\n",
      "Iteration  284 (elapsed time:  13s,  0.2mn, current cost  7342.292)\n",
      "Iteration  285 (elapsed time:  13s,  0.2mn, current cost  7342.291)\n",
      "Iteration  286 (elapsed time:  13s,  0.2mn, current cost  7342.289)\n",
      "Iteration  287 (elapsed time:  13s,  0.2mn, current cost  7342.288)\n",
      "Iteration  288 (elapsed time:  13s,  0.2mn, current cost  7342.287)\n",
      "Iteration  289 (elapsed time:  13s,  0.2mn, current cost  7342.286)\n",
      "Iteration  290 (elapsed time:  14s,  0.2mn, current cost  7342.286)\n",
      "Iteration  291 (elapsed time:  14s,  0.2mn, current cost  7342.285)\n",
      "Iteration  292 (elapsed time:  14s,  0.2mn, current cost  7342.284)\n",
      "Iteration  293 (elapsed time:  14s,  0.2mn, current cost  7342.283)\n",
      "Iteration  294 (elapsed time:  14s,  0.2mn, current cost  7342.283)\n",
      "Iteration  295 (elapsed time:  14s,  0.2mn, current cost  7342.282)\n",
      "Iteration  296 (elapsed time:  14s,  0.2mn, current cost  7342.282)\n",
      "Iteration  297 (elapsed time:  14s,  0.2mn, current cost  7342.281)\n",
      "Iteration  298 (elapsed time:  14s,  0.2mn, current cost  7342.281)\n",
      "Iteration  299 (elapsed time:  14s,  0.2mn, current cost  7342.280)\n",
      "Iteration  300 (elapsed time:  14s,  0.2mn, current cost  7342.280)\n",
      "Iteration  301 (elapsed time:  14s,  0.2mn, current cost  7342.280)\n",
      "Iteration  302 (elapsed time:  14s,  0.2mn, current cost  7342.279)\n",
      "Iteration  303 (elapsed time:  14s,  0.2mn, current cost  7342.279)\n",
      "Iteration  304 (elapsed time:  14s,  0.2mn, current cost  7342.279)\n",
      "Iteration  305 (elapsed time:  15s,  0.3mn, current cost  7342.278)\n",
      "Iteration  306 (elapsed time:  15s,  0.3mn, current cost  7342.278)\n",
      "Iteration  307 (elapsed time:  15s,  0.3mn, current cost  7342.278)\n",
      "Iteration  308 (elapsed time:  15s,  0.3mn, current cost  7342.278)\n",
      "Iteration  309 (elapsed time:  15s,  0.3mn, current cost  7342.277)\n",
      "Iteration  310 (elapsed time:  15s,  0.3mn, current cost  7342.277)\n",
      "Iteration  311 (elapsed time:  15s,  0.3mn, current cost  7342.277)\n",
      "Iteration  312 (elapsed time:  15s,  0.3mn, current cost  7342.277)\n",
      "Iteration  313 (elapsed time:  15s,  0.3mn, current cost  7342.277)\n",
      "Iteration  314 (elapsed time:  15s,  0.3mn, current cost  7342.277)\n",
      "Iteration  315 (elapsed time:  15s,  0.3mn, current cost  7342.276)\n",
      "Iteration  316 (elapsed time:  15s,  0.3mn, current cost  7342.276)\n",
      "Iteration  317 (elapsed time:  15s,  0.3mn, current cost  7342.276)\n",
      "Iteration  318 (elapsed time:  16s,  0.3mn, current cost  7342.276)\n",
      "Iteration  319 (elapsed time:  16s,  0.3mn, current cost  7342.276)\n",
      "Iteration  320 (elapsed time:  16s,  0.3mn, current cost  7342.276)\n",
      "Iteration  321 (elapsed time:  16s,  0.3mn, current cost  7342.276)\n",
      "Iteration  322 (elapsed time:  16s,  0.3mn, current cost  7342.276)\n",
      "--- Convergence reached after 322 iterations\n",
      "Iteration   0 (elapsed time:   0s,  0.0mn, current cost     nan)\n",
      "Iteration   1 (elapsed time:   0s,  0.0mn, current cost  425232.317)\n",
      "Iteration   2 (elapsed time:   0s,  0.0mn, current cost  310803.632)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.7/site-packages/sklearn/decomposition/sparse_pca.py:170: DeprecationWarning: normalize_components=False is a backward-compatible setting that implements a non-standard definition of sparse PCA. This compatibility mode will be removed in 0.22.\n",
      "  DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration   3 (elapsed time:   0s,  0.0mn, current cost  309099.257)\n",
      "Iteration   4 (elapsed time:   0s,  0.0mn, current cost  307471.517)\n",
      "Iteration   5 (elapsed time:   0s,  0.0mn, current cost  305491.571)\n",
      "Iteration   6 (elapsed time:   0s,  0.0mn, current cost  303633.811)\n",
      "Iteration   7 (elapsed time:   0s,  0.0mn, current cost  302109.184)\n",
      "Iteration   8 (elapsed time:   0s,  0.0mn, current cost  300763.339)\n",
      "Iteration   9 (elapsed time:   0s,  0.0mn, current cost  299667.004)\n",
      "Iteration  10 (elapsed time:   0s,  0.0mn, current cost  298901.070)\n",
      "Iteration  11 (elapsed time:   0s,  0.0mn, current cost  298461.625)\n",
      "Iteration  12 (elapsed time:   0s,  0.0mn, current cost  298137.232)\n",
      "Iteration  13 (elapsed time:   0s,  0.0mn, current cost  297827.591)\n",
      "Iteration  14 (elapsed time:   0s,  0.0mn, current cost  297512.796)\n",
      "Iteration  15 (elapsed time:   0s,  0.0mn, current cost  297223.427)\n",
      "Iteration  16 (elapsed time:   0s,  0.0mn, current cost  296996.792)\n",
      "Iteration  17 (elapsed time:   0s,  0.0mn, current cost  296832.429)\n",
      "Iteration  18 (elapsed time:   0s,  0.0mn, current cost  296718.814)\n",
      "Iteration  19 (elapsed time:   0s,  0.0mn, current cost  296636.687)\n",
      "Iteration  20 (elapsed time:   1s,  0.0mn, current cost  296569.850)\n",
      "Iteration  21 (elapsed time:   1s,  0.0mn, current cost  296509.339)\n",
      "Iteration  22 (elapsed time:   1s,  0.0mn, current cost  296453.497)\n",
      "Iteration  23 (elapsed time:   1s,  0.0mn, current cost  296402.528)\n",
      "Iteration  24 (elapsed time:   1s,  0.0mn, current cost  296354.733)\n",
      "Iteration  25 (elapsed time:   1s,  0.0mn, current cost  296313.698)\n",
      "Iteration  26 (elapsed time:   1s,  0.0mn, current cost  296278.757)\n",
      "Iteration  27 (elapsed time:   1s,  0.0mn, current cost  296254.315)\n",
      "Iteration  28 (elapsed time:   1s,  0.0mn, current cost  296237.464)\n",
      "Iteration  29 (elapsed time:   1s,  0.0mn, current cost  296226.062)\n",
      "Iteration  30 (elapsed time:   1s,  0.0mn, current cost  296218.732)\n",
      "Iteration  31 (elapsed time:   1s,  0.0mn, current cost  296213.701)\n",
      "Iteration  32 (elapsed time:   1s,  0.0mn, current cost  296210.158)\n",
      "Iteration  33 (elapsed time:   1s,  0.0mn, current cost  296207.527)\n",
      "Iteration  34 (elapsed time:   1s,  0.0mn, current cost  296205.498)\n",
      "Iteration  35 (elapsed time:   1s,  0.0mn, current cost  296203.899)\n",
      "Iteration  36 (elapsed time:   1s,  0.0mn, current cost  296202.591)\n",
      "Iteration  37 (elapsed time:   1s,  0.0mn, current cost  296201.490)\n",
      "Iteration  38 (elapsed time:   1s,  0.0mn, current cost  296200.543)\n",
      "Iteration  39 (elapsed time:   1s,  0.0mn, current cost  296199.718)\n",
      "Iteration  40 (elapsed time:   1s,  0.0mn, current cost  296198.993)\n",
      "Iteration  41 (elapsed time:   1s,  0.0mn, current cost  296198.351)\n",
      "Iteration  42 (elapsed time:   1s,  0.0mn, current cost  296197.779)\n",
      "Iteration  43 (elapsed time:   2s,  0.0mn, current cost  296197.268)\n",
      "Iteration  44 (elapsed time:   2s,  0.0mn, current cost  296196.811)\n",
      "Iteration  45 (elapsed time:   2s,  0.0mn, current cost  296196.399)\n",
      "Iteration  46 (elapsed time:   2s,  0.0mn, current cost  296196.027)\n",
      "Iteration  47 (elapsed time:   2s,  0.0mn, current cost  296195.690)\n",
      "Iteration  48 (elapsed time:   2s,  0.0mn, current cost  296195.382)\n",
      "Iteration  49 (elapsed time:   2s,  0.0mn, current cost  296195.102)\n",
      "Iteration  50 (elapsed time:   2s,  0.0mn, current cost  296194.845)\n",
      "Iteration  51 (elapsed time:   2s,  0.0mn, current cost  296194.611)\n",
      "Iteration  52 (elapsed time:   2s,  0.0mn, current cost  296194.395)\n",
      "Iteration  53 (elapsed time:   2s,  0.0mn, current cost  296194.197)\n",
      "Iteration  54 (elapsed time:   2s,  0.0mn, current cost  296194.015)\n",
      "Iteration  55 (elapsed time:   2s,  0.0mn, current cost  296193.848)\n",
      "Iteration  56 (elapsed time:   2s,  0.0mn, current cost  296193.694)\n",
      "Iteration  57 (elapsed time:   2s,  0.0mn, current cost  296193.552)\n",
      "Iteration  58 (elapsed time:   2s,  0.0mn, current cost  296193.421)\n",
      "Iteration  59 (elapsed time:   2s,  0.0mn, current cost  296193.300)\n",
      "Iteration  60 (elapsed time:   2s,  0.0mn, current cost  296193.189)\n",
      "Iteration  61 (elapsed time:   2s,  0.0mn, current cost  296193.087)\n",
      "Iteration  62 (elapsed time:   2s,  0.0mn, current cost  296192.992)\n",
      "Iteration  63 (elapsed time:   2s,  0.0mn, current cost  296192.905)\n",
      "Iteration  64 (elapsed time:   2s,  0.0mn, current cost  296192.825)\n",
      "Iteration  65 (elapsed time:   2s,  0.0mn, current cost  296192.750)\n",
      "Iteration  66 (elapsed time:   2s,  0.0mn, current cost  296192.681)\n",
      "Iteration  67 (elapsed time:   2s,  0.0mn, current cost  296192.615)\n",
      "Iteration  68 (elapsed time:   2s,  0.0mn, current cost  296192.554)\n",
      "Iteration  69 (elapsed time:   2s,  0.0mn, current cost  296192.496)\n",
      "Iteration  70 (elapsed time:   2s,  0.0mn, current cost  296192.441)\n",
      "Iteration  71 (elapsed time:   2s,  0.0mn, current cost  296192.390)\n",
      "Iteration  72 (elapsed time:   3s,  0.1mn, current cost  296192.341)\n",
      "Iteration  73 (elapsed time:   3s,  0.1mn, current cost  296192.294)\n",
      "Iteration  74 (elapsed time:   3s,  0.1mn, current cost  296192.250)\n",
      "Iteration  75 (elapsed time:   3s,  0.1mn, current cost  296192.208)\n",
      "Iteration  76 (elapsed time:   3s,  0.1mn, current cost  296192.168)\n",
      "Iteration  77 (elapsed time:   3s,  0.1mn, current cost  296192.130)\n",
      "Iteration  78 (elapsed time:   3s,  0.1mn, current cost  296192.093)\n",
      "Iteration  79 (elapsed time:   3s,  0.1mn, current cost  296192.059)\n",
      "Iteration  80 (elapsed time:   3s,  0.1mn, current cost  296192.026)\n",
      "Iteration  81 (elapsed time:   3s,  0.1mn, current cost  296191.994)\n",
      "Iteration  82 (elapsed time:   3s,  0.1mn, current cost  296191.963)\n",
      "Iteration  83 (elapsed time:   3s,  0.1mn, current cost  296191.934)\n",
      "Iteration  84 (elapsed time:   3s,  0.1mn, current cost  296191.907)\n",
      "Iteration  85 (elapsed time:   3s,  0.1mn, current cost  296191.880)\n",
      "Iteration  86 (elapsed time:   3s,  0.1mn, current cost  296191.854)\n",
      "Iteration  87 (elapsed time:   3s,  0.1mn, current cost  296191.829)\n",
      "Iteration  88 (elapsed time:   3s,  0.1mn, current cost  296191.805)\n",
      "Iteration  89 (elapsed time:   3s,  0.1mn, current cost  296191.782)\n",
      "Iteration  90 (elapsed time:   3s,  0.1mn, current cost  296191.759)\n",
      "Iteration  91 (elapsed time:   3s,  0.1mn, current cost  296191.737)\n",
      "Iteration  92 (elapsed time:   3s,  0.1mn, current cost  296191.716)\n",
      "Iteration  93 (elapsed time:   3s,  0.1mn, current cost  296191.694)\n",
      "Iteration  94 (elapsed time:   3s,  0.1mn, current cost  296191.673)\n",
      "Iteration  95 (elapsed time:   3s,  0.1mn, current cost  296191.651)\n",
      "Iteration  96 (elapsed time:   3s,  0.1mn, current cost  296191.629)\n",
      "Iteration  97 (elapsed time:   3s,  0.1mn, current cost  296191.607)\n",
      "Iteration  98 (elapsed time:   3s,  0.1mn, current cost  296191.585)\n",
      "Iteration  99 (elapsed time:   3s,  0.1mn, current cost  296191.563)\n",
      "Iteration  100 (elapsed time:   3s,  0.1mn, current cost  296191.541)\n",
      "Iteration  101 (elapsed time:   4s,  0.1mn, current cost  296191.519)\n",
      "Iteration  102 (elapsed time:   4s,  0.1mn, current cost  296191.496)\n",
      "Iteration  103 (elapsed time:   4s,  0.1mn, current cost  296191.474)\n",
      "Iteration  104 (elapsed time:   4s,  0.1mn, current cost  296191.451)\n",
      "Iteration  105 (elapsed time:   4s,  0.1mn, current cost  296191.428)\n",
      "Iteration  106 (elapsed time:   4s,  0.1mn, current cost  296191.405)\n",
      "Iteration  107 (elapsed time:   4s,  0.1mn, current cost  296191.381)\n",
      "Iteration  108 (elapsed time:   4s,  0.1mn, current cost  296191.358)\n",
      "Iteration  109 (elapsed time:   4s,  0.1mn, current cost  296191.334)\n",
      "Iteration  110 (elapsed time:   4s,  0.1mn, current cost  296191.310)\n",
      "Iteration  111 (elapsed time:   4s,  0.1mn, current cost  296191.286)\n",
      "Iteration  112 (elapsed time:   4s,  0.1mn, current cost  296191.262)\n",
      "Iteration  113 (elapsed time:   4s,  0.1mn, current cost  296191.238)\n",
      "Iteration  114 (elapsed time:   4s,  0.1mn, current cost  296191.214)\n",
      "Iteration  115 (elapsed time:   4s,  0.1mn, current cost  296191.189)\n",
      "Iteration  116 (elapsed time:   4s,  0.1mn, current cost  296191.164)\n",
      "Iteration  117 (elapsed time:   4s,  0.1mn, current cost  296191.139)\n",
      "Iteration  118 (elapsed time:   4s,  0.1mn, current cost  296191.114)\n",
      "Iteration  119 (elapsed time:   4s,  0.1mn, current cost  296191.089)\n",
      "Iteration  120 (elapsed time:   4s,  0.1mn, current cost  296191.063)\n",
      "Iteration  121 (elapsed time:   4s,  0.1mn, current cost  296191.037)\n",
      "Iteration  122 (elapsed time:   4s,  0.1mn, current cost  296191.010)\n",
      "Iteration  123 (elapsed time:   4s,  0.1mn, current cost  296190.983)\n",
      "Iteration  124 (elapsed time:   4s,  0.1mn, current cost  296190.956)\n",
      "Iteration  125 (elapsed time:   4s,  0.1mn, current cost  296190.928)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration  126 (elapsed time:   4s,  0.1mn, current cost  296190.900)\n",
      "Iteration  127 (elapsed time:   4s,  0.1mn, current cost  296190.871)\n",
      "Iteration  128 (elapsed time:   4s,  0.1mn, current cost  296190.841)\n",
      "Iteration  129 (elapsed time:   4s,  0.1mn, current cost  296190.812)\n",
      "Iteration  130 (elapsed time:   4s,  0.1mn, current cost  296190.781)\n",
      "Iteration  131 (elapsed time:   4s,  0.1mn, current cost  296190.750)\n",
      "Iteration  132 (elapsed time:   5s,  0.1mn, current cost  296190.719)\n",
      "Iteration  133 (elapsed time:   5s,  0.1mn, current cost  296190.687)\n",
      "Iteration  134 (elapsed time:   5s,  0.1mn, current cost  296190.655)\n",
      "Iteration  135 (elapsed time:   5s,  0.1mn, current cost  296190.622)\n",
      "Iteration  136 (elapsed time:   5s,  0.1mn, current cost  296190.589)\n",
      "Iteration  137 (elapsed time:   5s,  0.1mn, current cost  296190.555)\n",
      "Iteration  138 (elapsed time:   5s,  0.1mn, current cost  296190.520)\n",
      "Iteration  139 (elapsed time:   5s,  0.1mn, current cost  296190.485)\n",
      "Iteration  140 (elapsed time:   5s,  0.1mn, current cost  296190.449)\n",
      "Iteration  141 (elapsed time:   5s,  0.1mn, current cost  296190.413)\n",
      "Iteration  142 (elapsed time:   5s,  0.1mn, current cost  296190.374)\n",
      "Iteration  143 (elapsed time:   5s,  0.1mn, current cost  296190.332)\n",
      "Iteration  144 (elapsed time:   5s,  0.1mn, current cost  296190.286)\n",
      "Iteration  145 (elapsed time:   5s,  0.1mn, current cost  296190.235)\n",
      "Iteration  146 (elapsed time:   5s,  0.1mn, current cost  296190.180)\n",
      "Iteration  147 (elapsed time:   5s,  0.1mn, current cost  296190.120)\n",
      "Iteration  148 (elapsed time:   5s,  0.1mn, current cost  296190.055)\n",
      "Iteration  149 (elapsed time:   5s,  0.1mn, current cost  296189.983)\n",
      "Iteration  150 (elapsed time:   5s,  0.1mn, current cost  296189.905)\n",
      "Iteration  151 (elapsed time:   5s,  0.1mn, current cost  296189.820)\n",
      "Iteration  152 (elapsed time:   5s,  0.1mn, current cost  296189.727)\n",
      "Iteration  153 (elapsed time:   5s,  0.1mn, current cost  296189.625)\n",
      "Iteration  154 (elapsed time:   5s,  0.1mn, current cost  296189.514)\n",
      "Iteration  155 (elapsed time:   5s,  0.1mn, current cost  296189.393)\n",
      "Iteration  156 (elapsed time:   5s,  0.1mn, current cost  296189.261)\n",
      "Iteration  157 (elapsed time:   5s,  0.1mn, current cost  296189.117)\n",
      "Iteration  158 (elapsed time:   5s,  0.1mn, current cost  296188.960)\n",
      "Iteration  159 (elapsed time:   5s,  0.1mn, current cost  296188.789)\n",
      "Iteration  160 (elapsed time:   5s,  0.1mn, current cost  296188.602)\n",
      "Iteration  161 (elapsed time:   5s,  0.1mn, current cost  296188.399)\n",
      "Iteration  162 (elapsed time:   5s,  0.1mn, current cost  296188.177)\n",
      "Iteration  163 (elapsed time:   6s,  0.1mn, current cost  296187.937)\n",
      "Iteration  164 (elapsed time:   6s,  0.1mn, current cost  296187.693)\n",
      "Iteration  165 (elapsed time:   6s,  0.1mn, current cost  296187.445)\n",
      "Iteration  166 (elapsed time:   6s,  0.1mn, current cost  296187.192)\n",
      "Iteration  167 (elapsed time:   6s,  0.1mn, current cost  296186.936)\n",
      "Iteration  168 (elapsed time:   6s,  0.1mn, current cost  296186.680)\n",
      "Iteration  169 (elapsed time:   6s,  0.1mn, current cost  296186.422)\n",
      "Iteration  170 (elapsed time:   6s,  0.1mn, current cost  296186.164)\n",
      "Iteration  171 (elapsed time:   6s,  0.1mn, current cost  296185.904)\n",
      "Iteration  172 (elapsed time:   6s,  0.1mn, current cost  296185.643)\n",
      "Iteration  173 (elapsed time:   6s,  0.1mn, current cost  296185.380)\n",
      "Iteration  174 (elapsed time:   6s,  0.1mn, current cost  296185.115)\n",
      "Iteration  175 (elapsed time:   6s,  0.1mn, current cost  296184.845)\n",
      "Iteration  176 (elapsed time:   6s,  0.1mn, current cost  296184.572)\n",
      "Iteration  177 (elapsed time:   6s,  0.1mn, current cost  296184.295)\n",
      "Iteration  178 (elapsed time:   6s,  0.1mn, current cost  296184.022)\n",
      "Iteration  179 (elapsed time:   6s,  0.1mn, current cost  296183.776)\n",
      "Iteration  180 (elapsed time:   6s,  0.1mn, current cost  296183.552)\n",
      "Iteration  181 (elapsed time:   6s,  0.1mn, current cost  296183.345)\n",
      "Iteration  182 (elapsed time:   6s,  0.1mn, current cost  296183.154)\n",
      "Iteration  183 (elapsed time:   6s,  0.1mn, current cost  296182.977)\n",
      "Iteration  184 (elapsed time:   6s,  0.1mn, current cost  296182.811)\n",
      "Iteration  185 (elapsed time:   6s,  0.1mn, current cost  296182.651)\n",
      "Iteration  186 (elapsed time:   6s,  0.1mn, current cost  296182.495)\n",
      "Iteration  187 (elapsed time:   6s,  0.1mn, current cost  296182.339)\n",
      "Iteration  188 (elapsed time:   6s,  0.1mn, current cost  296182.185)\n",
      "Iteration  189 (elapsed time:   6s,  0.1mn, current cost  296182.030)\n",
      "Iteration  190 (elapsed time:   6s,  0.1mn, current cost  296181.874)\n",
      "Iteration  191 (elapsed time:   6s,  0.1mn, current cost  296181.718)\n",
      "Iteration  192 (elapsed time:   6s,  0.1mn, current cost  296181.561)\n",
      "Iteration  193 (elapsed time:   6s,  0.1mn, current cost  296181.402)\n",
      "Iteration  194 (elapsed time:   6s,  0.1mn, current cost  296181.241)\n",
      "Iteration  195 (elapsed time:   7s,  0.1mn, current cost  296181.079)\n",
      "Iteration  196 (elapsed time:   7s,  0.1mn, current cost  296180.916)\n",
      "Iteration  197 (elapsed time:   7s,  0.1mn, current cost  296180.750)\n",
      "Iteration  198 (elapsed time:   7s,  0.1mn, current cost  296180.582)\n",
      "Iteration  199 (elapsed time:   7s,  0.1mn, current cost  296180.413)\n",
      "Iteration  200 (elapsed time:   7s,  0.1mn, current cost  296180.241)\n",
      "Iteration  201 (elapsed time:   7s,  0.1mn, current cost  296180.069)\n",
      "Iteration  202 (elapsed time:   7s,  0.1mn, current cost  296179.905)\n",
      "Iteration  203 (elapsed time:   7s,  0.1mn, current cost  296179.747)\n",
      "Iteration  204 (elapsed time:   7s,  0.1mn, current cost  296179.596)\n",
      "Iteration  205 (elapsed time:   7s,  0.1mn, current cost  296179.451)\n",
      "Iteration  206 (elapsed time:   7s,  0.1mn, current cost  296179.311)\n",
      "Iteration  207 (elapsed time:   7s,  0.1mn, current cost  296179.176)\n",
      "Iteration  208 (elapsed time:   7s,  0.1mn, current cost  296179.046)\n",
      "Iteration  209 (elapsed time:   7s,  0.1mn, current cost  296178.921)\n",
      "Iteration  210 (elapsed time:   7s,  0.1mn, current cost  296178.801)\n",
      "Iteration  211 (elapsed time:   7s,  0.1mn, current cost  296178.684)\n",
      "Iteration  212 (elapsed time:   7s,  0.1mn, current cost  296178.572)\n",
      "Iteration  213 (elapsed time:   7s,  0.1mn, current cost  296178.461)\n",
      "Iteration  214 (elapsed time:   7s,  0.1mn, current cost  296178.353)\n",
      "Iteration  215 (elapsed time:   7s,  0.1mn, current cost  296178.246)\n",
      "Iteration  216 (elapsed time:   7s,  0.1mn, current cost  296178.141)\n",
      "Iteration  217 (elapsed time:   7s,  0.1mn, current cost  296178.038)\n",
      "Iteration  218 (elapsed time:   7s,  0.1mn, current cost  296177.937)\n",
      "Iteration  219 (elapsed time:   7s,  0.1mn, current cost  296177.837)\n",
      "Iteration  220 (elapsed time:   7s,  0.1mn, current cost  296177.739)\n",
      "Iteration  221 (elapsed time:   7s,  0.1mn, current cost  296177.643)\n",
      "Iteration  222 (elapsed time:   7s,  0.1mn, current cost  296177.549)\n",
      "Iteration  223 (elapsed time:   7s,  0.1mn, current cost  296177.457)\n",
      "Iteration  224 (elapsed time:   8s,  0.1mn, current cost  296177.368)\n",
      "Iteration  225 (elapsed time:   8s,  0.1mn, current cost  296177.280)\n",
      "Iteration  226 (elapsed time:   8s,  0.1mn, current cost  296177.194)\n",
      "Iteration  227 (elapsed time:   8s,  0.1mn, current cost  296177.110)\n",
      "Iteration  228 (elapsed time:   8s,  0.1mn, current cost  296177.027)\n",
      "Iteration  229 (elapsed time:   8s,  0.1mn, current cost  296176.945)\n",
      "Iteration  230 (elapsed time:   8s,  0.1mn, current cost  296176.863)\n",
      "Iteration  231 (elapsed time:   8s,  0.1mn, current cost  296176.783)\n",
      "Iteration  232 (elapsed time:   8s,  0.1mn, current cost  296176.703)\n",
      "Iteration  233 (elapsed time:   8s,  0.1mn, current cost  296176.624)\n",
      "Iteration  234 (elapsed time:   8s,  0.1mn, current cost  296176.545)\n",
      "Iteration  235 (elapsed time:   8s,  0.1mn, current cost  296176.468)\n",
      "Iteration  236 (elapsed time:   8s,  0.1mn, current cost  296176.391)\n",
      "Iteration  237 (elapsed time:   8s,  0.1mn, current cost  296176.316)\n",
      "Iteration  238 (elapsed time:   8s,  0.1mn, current cost  296176.242)\n",
      "Iteration  239 (elapsed time:   8s,  0.1mn, current cost  296176.169)\n",
      "Iteration  240 (elapsed time:   8s,  0.1mn, current cost  296176.097)\n",
      "Iteration  241 (elapsed time:   8s,  0.1mn, current cost  296176.026)\n",
      "Iteration  242 (elapsed time:   8s,  0.1mn, current cost  296175.956)\n",
      "Iteration  243 (elapsed time:   8s,  0.1mn, current cost  296175.888)\n",
      "Iteration  244 (elapsed time:   8s,  0.1mn, current cost  296175.822)\n",
      "Iteration  245 (elapsed time:   8s,  0.1mn, current cost  296175.757)\n",
      "Iteration  246 (elapsed time:   8s,  0.1mn, current cost  296175.693)\n",
      "Iteration  247 (elapsed time:   8s,  0.1mn, current cost  296175.631)\n",
      "Iteration  248 (elapsed time:   8s,  0.1mn, current cost  296175.571)\n",
      "Iteration  249 (elapsed time:   8s,  0.1mn, current cost  296175.511)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration  250 (elapsed time:   8s,  0.1mn, current cost  296175.453)\n",
      "Iteration  251 (elapsed time:   8s,  0.1mn, current cost  296175.396)\n",
      "Iteration  252 (elapsed time:   8s,  0.1mn, current cost  296175.341)\n",
      "Iteration  253 (elapsed time:   8s,  0.1mn, current cost  296175.286)\n",
      "Iteration  254 (elapsed time:   8s,  0.1mn, current cost  296175.233)\n",
      "Iteration  255 (elapsed time:   8s,  0.1mn, current cost  296175.181)\n",
      "Iteration  256 (elapsed time:   9s,  0.2mn, current cost  296175.130)\n",
      "Iteration  257 (elapsed time:   9s,  0.2mn, current cost  296175.080)\n",
      "Iteration  258 (elapsed time:   9s,  0.2mn, current cost  296175.030)\n",
      "Iteration  259 (elapsed time:   9s,  0.2mn, current cost  296174.982)\n",
      "Iteration  260 (elapsed time:   9s,  0.2mn, current cost  296174.935)\n",
      "Iteration  261 (elapsed time:   9s,  0.2mn, current cost  296174.890)\n",
      "Iteration  262 (elapsed time:   9s,  0.2mn, current cost  296174.845)\n",
      "Iteration  263 (elapsed time:   9s,  0.2mn, current cost  296174.802)\n",
      "Iteration  264 (elapsed time:   9s,  0.2mn, current cost  296174.760)\n",
      "Iteration  265 (elapsed time:   9s,  0.2mn, current cost  296174.720)\n",
      "Iteration  266 (elapsed time:   9s,  0.2mn, current cost  296174.680)\n",
      "Iteration  267 (elapsed time:   9s,  0.2mn, current cost  296174.642)\n",
      "Iteration  268 (elapsed time:   9s,  0.2mn, current cost  296174.604)\n",
      "Iteration  269 (elapsed time:   9s,  0.2mn, current cost  296174.568)\n",
      "Iteration  270 (elapsed time:   9s,  0.2mn, current cost  296174.532)\n",
      "Iteration  271 (elapsed time:   9s,  0.2mn, current cost  296174.498)\n",
      "Iteration  272 (elapsed time:   9s,  0.2mn, current cost  296174.465)\n",
      "Iteration  273 (elapsed time:   9s,  0.2mn, current cost  296174.433)\n",
      "Iteration  274 (elapsed time:   9s,  0.2mn, current cost  296174.401)\n",
      "Iteration  275 (elapsed time:   9s,  0.2mn, current cost  296174.370)\n",
      "Iteration  276 (elapsed time:   9s,  0.2mn, current cost  296174.341)\n",
      "Iteration  277 (elapsed time:   9s,  0.2mn, current cost  296174.312)\n",
      "Iteration  278 (elapsed time:   9s,  0.2mn, current cost  296174.283)\n",
      "Iteration  279 (elapsed time:   9s,  0.2mn, current cost  296174.256)\n",
      "Iteration  280 (elapsed time:   9s,  0.2mn, current cost  296174.229)\n",
      "Iteration  281 (elapsed time:   9s,  0.2mn, current cost  296174.203)\n",
      "Iteration  282 (elapsed time:   9s,  0.2mn, current cost  296174.178)\n",
      "Iteration  283 (elapsed time:   9s,  0.2mn, current cost  296174.153)\n",
      "Iteration  284 (elapsed time:   9s,  0.2mn, current cost  296174.129)\n",
      "Iteration  285 (elapsed time:   9s,  0.2mn, current cost  296174.105)\n",
      "Iteration  286 (elapsed time:   9s,  0.2mn, current cost  296174.082)\n",
      "Iteration  287 (elapsed time:   9s,  0.2mn, current cost  296174.060)\n",
      "Iteration  288 (elapsed time:   9s,  0.2mn, current cost  296174.038)\n",
      "Iteration  289 (elapsed time:  10s,  0.2mn, current cost  296174.017)\n",
      "Iteration  290 (elapsed time:  10s,  0.2mn, current cost  296173.996)\n",
      "Iteration  291 (elapsed time:  10s,  0.2mn, current cost  296173.976)\n",
      "Iteration  292 (elapsed time:  10s,  0.2mn, current cost  296173.956)\n",
      "Iteration  293 (elapsed time:  10s,  0.2mn, current cost  296173.937)\n",
      "Iteration  294 (elapsed time:  10s,  0.2mn, current cost  296173.918)\n",
      "Iteration  295 (elapsed time:  10s,  0.2mn, current cost  296173.899)\n",
      "Iteration  296 (elapsed time:  10s,  0.2mn, current cost  296173.881)\n",
      "Iteration  297 (elapsed time:  10s,  0.2mn, current cost  296173.864)\n",
      "Iteration  298 (elapsed time:  10s,  0.2mn, current cost  296173.847)\n",
      "Iteration  299 (elapsed time:  10s,  0.2mn, current cost  296173.830)\n",
      "Iteration  300 (elapsed time:  10s,  0.2mn, current cost  296173.814)\n",
      "Iteration  301 (elapsed time:  10s,  0.2mn, current cost  296173.798)\n",
      "Iteration  302 (elapsed time:  10s,  0.2mn, current cost  296173.782)\n",
      "Iteration  303 (elapsed time:  10s,  0.2mn, current cost  296173.767)\n",
      "Iteration  304 (elapsed time:  10s,  0.2mn, current cost  296173.752)\n",
      "Iteration  305 (elapsed time:  10s,  0.2mn, current cost  296173.737)\n",
      "Iteration  306 (elapsed time:  10s,  0.2mn, current cost  296173.723)\n",
      "Iteration  307 (elapsed time:  10s,  0.2mn, current cost  296173.709)\n",
      "Iteration  308 (elapsed time:  10s,  0.2mn, current cost  296173.695)\n",
      "Iteration  309 (elapsed time:  10s,  0.2mn, current cost  296173.682)\n",
      "Iteration  310 (elapsed time:  10s,  0.2mn, current cost  296173.668)\n",
      "Iteration  311 (elapsed time:  10s,  0.2mn, current cost  296173.655)\n",
      "Iteration  312 (elapsed time:  10s,  0.2mn, current cost  296173.643)\n",
      "Iteration  313 (elapsed time:  10s,  0.2mn, current cost  296173.630)\n",
      "Iteration  314 (elapsed time:  10s,  0.2mn, current cost  296173.618)\n",
      "Iteration  315 (elapsed time:  10s,  0.2mn, current cost  296173.605)\n",
      "Iteration  316 (elapsed time:  10s,  0.2mn, current cost  296173.593)\n",
      "Iteration  317 (elapsed time:  10s,  0.2mn, current cost  296173.582)\n",
      "Iteration  318 (elapsed time:  10s,  0.2mn, current cost  296173.570)\n",
      "Iteration  319 (elapsed time:  11s,  0.2mn, current cost  296173.558)\n",
      "Iteration  320 (elapsed time:  11s,  0.2mn, current cost  296173.547)\n",
      "Iteration  321 (elapsed time:  11s,  0.2mn, current cost  296173.536)\n",
      "Iteration  322 (elapsed time:  11s,  0.2mn, current cost  296173.525)\n",
      "Iteration  323 (elapsed time:  11s,  0.2mn, current cost  296173.514)\n",
      "Iteration  324 (elapsed time:  11s,  0.2mn, current cost  296173.504)\n",
      "Iteration  325 (elapsed time:  11s,  0.2mn, current cost  296173.493)\n",
      "Iteration  326 (elapsed time:  11s,  0.2mn, current cost  296173.483)\n",
      "Iteration  327 (elapsed time:  11s,  0.2mn, current cost  296173.473)\n",
      "Iteration  328 (elapsed time:  11s,  0.2mn, current cost  296173.464)\n",
      "Iteration  329 (elapsed time:  11s,  0.2mn, current cost  296173.455)\n",
      "Iteration  330 (elapsed time:  11s,  0.2mn, current cost  296173.447)\n",
      "Iteration  331 (elapsed time:  11s,  0.2mn, current cost  296173.439)\n",
      "Iteration  332 (elapsed time:  11s,  0.2mn, current cost  296173.431)\n",
      "Iteration  333 (elapsed time:  11s,  0.2mn, current cost  296173.424)\n",
      "Iteration  334 (elapsed time:  11s,  0.2mn, current cost  296173.417)\n",
      "Iteration  335 (elapsed time:  11s,  0.2mn, current cost  296173.410)\n",
      "Iteration  336 (elapsed time:  11s,  0.2mn, current cost  296173.404)\n",
      "Iteration  337 (elapsed time:  11s,  0.2mn, current cost  296173.397)\n",
      "Iteration  338 (elapsed time:  11s,  0.2mn, current cost  296173.392)\n",
      "Iteration  339 (elapsed time:  11s,  0.2mn, current cost  296173.386)\n",
      "Iteration  340 (elapsed time:  11s,  0.2mn, current cost  296173.381)\n",
      "Iteration  341 (elapsed time:  11s,  0.2mn, current cost  296173.376)\n",
      "Iteration  342 (elapsed time:  11s,  0.2mn, current cost  296173.371)\n",
      "Iteration  343 (elapsed time:  11s,  0.2mn, current cost  296173.366)\n",
      "Iteration  344 (elapsed time:  11s,  0.2mn, current cost  296173.361)\n",
      "Iteration  345 (elapsed time:  11s,  0.2mn, current cost  296173.357)\n",
      "Iteration  346 (elapsed time:  11s,  0.2mn, current cost  296173.353)\n",
      "Iteration  347 (elapsed time:  11s,  0.2mn, current cost  296173.349)\n",
      "Iteration  348 (elapsed time:  11s,  0.2mn, current cost  296173.345)\n",
      "Iteration  349 (elapsed time:  11s,  0.2mn, current cost  296173.342)\n",
      "Iteration  350 (elapsed time:  11s,  0.2mn, current cost  296173.339)\n",
      "Iteration  351 (elapsed time:  12s,  0.2mn, current cost  296173.335)\n",
      "--- Convergence reached after 351 iterations\n"
     ]
    }
   ],
   "source": [
    "# Compute source PCs\n",
    "source_pc = SparsePCA(n_factors, verbose=10, alpha=1).fit(X_source_filtered).components_\n",
    "source_pc = scipy.linalg.orth(source_pc.transpose()).transpose()\n",
    "source_pc_df = pd.DataFrame(source_pc, columns=np.arange(X_source_filtered.shape[1]))\n",
    "source_pc_df.index = ['Factor %s'%(i+1) for i in range(n_factors)]\n",
    "\n",
    "# Compute target PCs\n",
    "target_pc = SparsePCA(n_factors, verbose=10, alpha=10).fit(X_target_filtered).components_\n",
    "target_pc = scipy.linalg.orth(target_pc.transpose()).transpose()\n",
    "target_pc_df= pd.DataFrame(target_pc, columns=np.arange(X_target_filtered.shape[1]))\n",
    "target_pc_df.index = ['Factor %s'%(i+1) for i in range(n_factors)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x216 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x216 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(7,3))\n",
    "sns.heatmap(source_pc_df,\n",
    "            cmap='seismic_r',\n",
    "            center=0,\n",
    "            cbar=False)\n",
    "plt.ylabel('Factors', fontsize=25, fontweight='bold')\n",
    "plt.xlabel('Genes', fontsize=25, fontweight='bold')\n",
    "plt.yticks([], [])\n",
    "plt.xticks([], [])\n",
    "plt.savefig('./figures/fig1_source_factors_view.png', dpi=300)\n",
    "plt.show()\n",
    "\n",
    "plt.figure(figsize=(7,3))\n",
    "sns.heatmap(target_pc_df,\n",
    "            cmap='seismic_r',\n",
    "            center=0,\n",
    "            cbar=False)\n",
    "plt.yticks([], [])\n",
    "plt.xticks([], [])\n",
    "plt.ylabel('Factors', fontsize=25, fontweight='bold')\n",
    "plt.xlabel('Genes', fontsize=25, fontweight='bold')\n",
    "plt.savefig('./figures/fig1_target_factors_view.png', dpi=300)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<precise.principal_vectors.PVComputation at 0x1c40dd29b0>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "principal_vectors = PVComputation(\n",
    "    n_factors = n_factors,\n",
    "    n_pv = n_pv,\n",
    "    dim_reduction = dim_reduction,\n",
    "    dim_reduction_target = dim_reduction_target\n",
    ")\n",
    "\n",
    "principal_vectors.compute_principal_vectors(source_pc, target_pc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Create clustermap on the columns.\n",
    "g_source = sns.clustermap((principal_vectors.source_components_), row_cluster=False, cmap='seismic_r', center=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,5))\n",
    "sns.heatmap(principal_vectors.source_components_[:,g_source.dendrogram_col.reordered_ind],\n",
    "            cmap='seismic_r',\n",
    "            center=0)\n",
    "plt.show()\n",
    "\n",
    "plt.figure(figsize=(10,5))\n",
    "sns.heatmap(principal_vectors.target_components_[:,g_source.dendrogram_col.reordered_ind],\n",
    "            cmap='seismic_r',\n",
    "            center=0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:10: FutureWarning: sortlevel is deprecated, use sort_index(level= ...)\n",
      "  # Remove the CWD from sys.path while we load stuff.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Vizualisation of principal vectors\n",
    "plt.figure(figsize=(10,5))\n",
    "pv_df = pd.DataFrame(np.concatenate([\n",
    "    principal_vectors.source_components_,\n",
    "    principal_vectors.target_components_]),\n",
    "                    columns=range(X_source_filtered.shape[1]))\n",
    "\n",
    "pv_df['PV'] = ['PV %s'%(i+1) for i in range(n_pv)]*2\n",
    "pv_df['system'] = np.concatenate([['source']*n_pv, ['target']*n_pv])\n",
    "pv_df = pv_df.set_index(['PV', 'system']).sortlevel(axis=0)\n",
    "\n",
    "ax = sns.heatmap(pv_df[g_source.dendrogram_col.reordered_ind],\n",
    "            cmap='seismic_r',\n",
    "            center=0,\n",
    "            cbar=False)\n",
    "\n",
    "for i in range(n_pv+1):\n",
    "    ax.axhline(2*i, color='black')\n",
    "plt.xlabel('Genes', fontsize=30, fontweight='bold')\n",
    "plt.yticks(plt.yticks()[0], ['pre-clinical', 'tumors']*n_pv, fontsize=25, fontweight='bold')\n",
    "plt.xticks([],[])\n",
    "plt.ylabel('')\n",
    "\n",
    "ax2 = ax.twinx()\n",
    "ax2.set_yticks(np.arange(n_pv,2)+1)\n",
    "ax2.set_yticklabels(['PV %s'%(i) for i in range(n_pv)])\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.savefig('./figures/fig1_pv_view.png', dpi=300)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Consensus"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.7/site-packages/sklearn/decomposition/sparse_pca.py:170: DeprecationWarning: normalize_components=False is a backward-compatible setting that implements a non-standard definition of sparse PCA. This compatibility mode will be removed in 0.22.\n",
      "  DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration   0 (elapsed time:   0s,  0.0mn, current cost     nan)\n",
      "Iteration   1 (elapsed time:  11s,  0.2mn, current cost  567981.201)\n",
      "Adding new random atom\n",
      "Iteration   2 (elapsed time:  12s,  0.2mn, current cost  563528.385)\n",
      "Adding new random atom\n",
      "Iteration   3 (elapsed time:  13s,  0.2mn, current cost  563359.797)\n",
      "Adding new random atom\n",
      "Iteration   4 (elapsed time:  14s,  0.2mn, current cost  563275.427)\n",
      "Adding new random atom\n",
      "Iteration   5 (elapsed time:  15s,  0.3mn, current cost  563213.061)\n",
      "Adding new random atom\n",
      "Iteration   6 (elapsed time:  16s,  0.3mn, current cost  563152.535)\n",
      "Adding new random atom\n",
      "--- Convergence reached after 6 iterations\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.7/site-packages/sklearn/decomposition/sparse_pca.py:170: DeprecationWarning: normalize_components=False is a backward-compatible setting that implements a non-standard definition of sparse PCA. This compatibility mode will be removed in 0.22.\n",
      "  DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration   0 (elapsed time:   3s,  0.1mn, current cost     nan)\n",
      "Iteration   1 (elapsed time:  17s,  0.3mn, current cost  8474913.033)\n",
      "Iteration   2 (elapsed time:  20s,  0.3mn, current cost  7881829.834)\n",
      "Iteration   3 (elapsed time:  23s,  0.4mn, current cost  7850355.544)\n",
      "Iteration   4 (elapsed time:  26s,  0.4mn, current cost  7825674.275)\n",
      "Iteration   5 (elapsed time:  29s,  0.5mn, current cost  7809306.426)\n",
      "Iteration   6 (elapsed time:  32s,  0.5mn, current cost  7794234.445)\n",
      "Iteration   7 (elapsed time:  35s,  0.6mn, current cost  7778461.574)\n",
      "Iteration   8 (elapsed time:  38s,  0.6mn, current cost  7762944.971)\n",
      "Iteration   9 (elapsed time:  41s,  0.7mn, current cost  7750823.087)\n",
      "Iteration  10 (elapsed time:  44s,  0.7mn, current cost  7744032.630)\n",
      "Iteration  11 (elapsed time:  46s,  0.8mn, current cost  7741151.287)\n",
      "Iteration  12 (elapsed time:  49s,  0.8mn, current cost  7740008.838)\n",
      "--- Convergence reached after 12 iterations\n"
     ]
    },
    {
     "ename": "IndexError",
     "evalue": "index 4 is out of bounds for axis 0 with size 4",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mIndexError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-14-4e1a5d512b3e>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      9\u001b[0m                                    std_unit=std_unit)\n\u001b[1;32m     10\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 11\u001b[0;31m \u001b[0mconsensus\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX_source\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     12\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     13\u001b[0m \u001b[0mconsensus_representation\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mconsensus\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mconsensus_representation\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda3/lib/python3.7/site-packages/precise-1.0-py3.7.egg/precise/pipeline_routine.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X, y)\u001b[0m\n\u001b[1;32m    451\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    452\u001b[0m         \u001b[0;31m# Compute the consensus representation between each PV\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 453\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_find_common_representation\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    454\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    455\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/anaconda3/lib/python3.7/site-packages/precise-1.0-py3.7.egg/precise/pipeline_routine.py\u001b[0m in \u001b[0;36m_find_common_representation\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    395\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    396\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mi\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mn_pv\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 397\u001b[0;31m             \u001b[0msource_projected\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mflow_vectors\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdot\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msource_data\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtranspose\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    398\u001b[0m             \u001b[0mtarget_projected\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mflow_vectors\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdot\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtarget_data\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtranspose\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    399\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mIndexError\u001b[0m: index 4 is out of bounds for axis 0 with size 4"
     ]
    }
   ],
   "source": [
    "consensus = ConsensusRepresentation(source_data=X_source,\n",
    "                                   target_data=X_target,\n",
    "                                   n_factors=n_factors,\n",
    "                                   n_pv=n_pv,\n",
    "                                   dim_reduction='sparsepca',\n",
    "                                   n_representations=100,\n",
    "                                   use_data=False,\n",
    "                                   mean_center=mean_center,\n",
    "                                   std_unit=std_unit)\n",
    "\n",
    "consensus.fit(X_source)\n",
    "\n",
    "consensus_representation = consensus.consensus_representation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure(figsize=(10,4))\n",
    "sns.heatmap(consensus_representation.transpose()[:3,g_source.dendrogram_col.reordered_ind],\n",
    "            cmap='seismic_r',\n",
    "            center=0,\n",
    "            cbar=False)\n",
    "plt.xticks([], [])\n",
    "plt.yticks([], [])\n",
    "plt.xlabel('Genes', fontsize=35, fontweight='bold')\n",
    "plt.ylabel('Common \\n factors', fontsize=35, fontweight='bold')\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.savefig('./figures/fig1_consensus_view.png', dpi=300)\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python (precise_figures)",
   "language": "python",
   "name": "precise_figures"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}