experiments/experiments_pbmc3k.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Please run code in `seurat_pbmc.ipynb` first."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.6/importlib/_bootstrap.py:219: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88\n",
      "  return f(*args, **kwds)\n",
      "/opt/conda/lib/python3.6/importlib/_bootstrap.py:219: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88\n",
      "  return f(*args, **kwds)\n"
     ]
    }
   ],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "import sys\n",
    "sys.path.insert(0, '../')\n",
    "\n",
    "import os\n",
    "import time\n",
    "import pickle\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import figure_utils as utils\n",
    "\n",
    "from truncated_normal import truncated_normal as tn\n",
    "from scipy.stats import ttest_ind, ranksums, gmean\n",
    "from sklearn.svm import SVC\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load data\n",
    "data_dir = './'\n",
    "data_R = pd.read_csv(os.path.join(data_dir, 'pbmc_seurat_preprocessed.txt'), delimiter=' ')\n",
    "labels_R = pd.read_csv(os.path.join(data_dir, 'pbmc_seurat_labels.txt'), delimiter=' ')\n",
    "tsne_R = pd.read_csv(os.path.join(data_dir, 'pbmc_seurat_tsne.txt'), delimiter=' ')\n",
    "data = np.array(data_R).T\n",
    "labels = np.array(labels_R).reshape(-1)\n",
    "tsne = np.array(tsne_R)\n",
    "genes = np.array(data_R.index)\n",
    "\n",
    "tests = [\n",
    "    \"t\",\n",
    "    \"wilcox\",\n",
    "    \"bimod\",\n",
    "    \"tobit\",\n",
    "    \"poisson\",\n",
    "    \"negbinom\",\n",
    "    \"MAST\"\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Split the datasets\n",
    "ident0, ident1 = 0, 1\n",
    "cellsamp1 = pd.read_csv(os.path.join(data_dir, 'pbmc_seurat_%sv%s_cellsamp1.txt'%(ident0, ident1)), delimiter=' ')\n",
    "pdict = {j:i for i, j in enumerate(data_R.columns)}\n",
    "cellsamp_inds = np.array([pdict[i] for i in np.array(cellsamp1).reshape(-1)])\n",
    "cellsamp1labels = np.array(pd.read_csv(os.path.join(data_dir, 'pbmc_seurat_%sv%s_cellsamplabels1.txt'%(ident0, ident1)), delimiter=' ')).reshape(-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Consistency of new labels with old: 1.000\n"
     ]
    }
   ],
   "source": [
    "# get hyperplane using first half of dataset\n",
    "svm = SVC(kernel='linear', C=100)\n",
    "svm.fit(data[cellsamp_inds], cellsamp1labels)\n",
    "a = svm.coef_.reshape(-1)\n",
    "b = svm.intercept_[0]\n",
    "labels1hat = svm.predict(data[cellsamp_inds])\n",
    "print('Consistency of new labels with old: %.3f'\\\n",
    "      %(np.sum(labels1hat == cellsamp1labels)/float(len(cellsamp1labels))))\n",
    "\n",
    "# assign points in second half of dataset\n",
    "inds = np.ones(len(data)).astype(bool)\n",
    "inds[cellsamp_inds] = 0\n",
    "for i in np.unique(labels):\n",
    "    if i != 0 and i != 1:\n",
    "        inds[labels == i] = 0\n",
    "labels2 = svm.predict(data[inds])\n",
    "y, z = data[inds][labels2 == 0], data[inds][labels2 == 1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Grab list of genes of interest based on other DE methods\n",
    "def grab_interesting_de_genes(ident0, ident1):\n",
    "    genes_list = []\n",
    "    ngenes = 10\n",
    "    for i, test in enumerate(tests):\n",
    "        seurat_de = pd.read_csv(os.path.join(data_dir, 'pbmc_seurat_%sv%s_de_%s.txt' \\\n",
    "                                             %(ident0, ident1, test)), delimiter=' ')\n",
    "        genes_list.extend(list(seurat_de.index[:10]))\n",
    "    genes_list = np.unique(genes_list)\n",
    "    return genes_list\n",
    "\n",
    "# Run TN test on genes that actually change across the two clusters\n",
    "keep_inds = np.var(np.vstack((y, z)), 0) > 0\n",
    "y = y[:, keep_inds]\n",
    "z = z[:, keep_inds]\n",
    "a = a[keep_inds]\n",
    "\n",
    "# sweep through iterations\n",
    "genes_list = grab_interesting_de_genes(ident0, ident1)\n",
    "genes_to_test = [np.where(genes[keep_inds] == gene)[0][0] for gene in genes_list]\n",
    "iter_sweep = np.logspace(0, 5, 11).astype(int)\n",
    "\n",
    "tn_test_multiple_runs = {}\n",
    "num_sims = 10\n",
    "\n",
    "picklefile = 'pbmc_seurat_0v1_de_tntest_numiter_sweep.pickle'\n",
    "if not os.path.isfile(picklefile):\n",
    "    np.random.seed(0)\n",
    "    for j in range(num_sims):\n",
    "        for num_iters in iter_sweep:\n",
    "            p_tn, likelihood = tn.tn_test(y, z, a=a, b=b, verbose=True, learning_rate=0.5,\n",
    "                                          eps=1e-1, genes_to_test=genes_to_test,\n",
    "                                          return_likelihood=True, num_iters=num_iters)\n",
    "            tn_test_multiple_runs[(num_iters, j)] = (p_tn, likelihood[-1])\n",
    "    pickle.dump(tn_test_multiple_runs, open(picklefile, 'wb'))\n",
    "else:\n",
    "    tn_test_multiple_runs = pickle.load(open(picklefile, 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAg0AAAEOCAYAAAAKZCSnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzs3XlYVdX6wPHvOod5kEEUJwQVJ5wVp6zUMlNT09RSu81lZenvmmXTtclb3crKNDPHyinTykyzQcuyzAkHTHHCCZRZBpnhnLN+fxxAQFBQ4IC8n+fhOWevtfY6L2Tsl7XXXktprRFCCCGEuBKDrQMQQgghRM0gSYMQQgghykSSBiGEEEKUiSQNQgghhCgTSRqEEEIIUSaSNAghhBCiTCRpEEIIIUSZSNIghBBCiDKRpEEIIYQQZSJJgxBCCCHKxM7WAVQmHx8fHRAQYOswhBCixtizZ0+C1rqereMQ1dN1nTQEBAQQEhJi6zCEEKLGUEqdsXUMovqS2xNCCCGEKBNJGoQQQghRJpI0CCGEEKJMrus5DUIIIURp9uzZU9/Ozm4R0B75IxrAAhw0mUyPduvWLa6kBpI0CCGEqJXs7OwWNWjQoG29evWSDAaDtnU8tmaxWFR8fHxQTEzMImB4SW0ksxJCCFFbta9Xr94FSRisDAaDrlevXgrWkZeS21RhPEIIISpZeFI4++P22zqMmsIgCUNReT+PUnMDSRqEEOI68l7Ie0z6bRKZpkxbhyLKaNmyZZ5KqW779u1zyi87cOCAY9++fQP9/f3bBwUFtR0yZEjzhQsXerVp0yaoTZs2QS4uLl0CAgLat2nTJmjkyJEBGzZscHd3d+/cpk2boFatWgXdcMMNrc6dO1fhUxAkaRBCiOvEzuid/B31N491eAxnO2dbhyPKaNWqVd5du3ZNW7p0qTdARkaGGjZsWMvHH388/syZMwfDwsIOT5w4Mb5z586ZR44cCTty5EhY+/btM5YuXXryyJEjYWvXrj0NEBwcnHbkyJGwY8eOhXXp0iV95syZ9Ss6VkkahBDiOqC1ZtaeWTRwbcA9be6xdTiijFJSUgy7d+92++yzz06vXbvWG2DBggXeXbt2TRs/fnxKfruhQ4emdu/ePassfVosFlJTU41eXl6mio5Xnp4QQojrwK8Rv3Lw/EHeuOENHI2Otg6nxnnu61C/YzGpLhXZZ6sG7hnvje4Uebk2K1eu9OzXr19Kx44ds728vEx//vmny8GDB527du2aUd7PCwkJcWvTpk1QcnKynbOzs3nWrFlnrz76kslIgxBC1HAmi4nZ+2bT3KM5w1oMs3U4ohxWr17tPW7cuCSAUaNGJS5btsz7avvKvz0RExNzYPz48eeffvrpJhUXqZWMNAghRA23/sR6TqWcYla/WdgZ5Nf61bjSiEBliI2NNe7YscP96NGjzk8//TRms1kppfS0adOit27d6nYtfY8aNSp5zJgxLSoq1nwy0iCEEDVYtjmbufvn0tGnI7c0vcXW4YhyWLZsmdfIkSMTo6Ki/jl37tw/MTExB5o0aZLTqlWrrD179ritWrXKI7/tjz/+6LZ7926ny/VX2JYtW9z8/f2zKzpmSUmFEKIGW3VkFbEZsbx909sopWwdjiiHNWvWeD/33HMxhcvuvPPOpJUrV3qvW7cufPLkyX7PP/+8n52dnW7btm3mvHnzIi7XX/6cBq017u7u5iVLlpyu6JiV1tfvuhbBwcE6JCTE1mEIIUSlSM1JZfC3g2lftz2f3vZphfSplNqjtQ6ukM6qudDQ0NOdOnVKsHUc1U1oaKhPp06dAkqqk9sTQghRQ31+6HNSslOY3HWyrUMRtYQkDTXF8U2w/t/WV1OOraMRQthYQmYCy8KWMShgEEF1g2wdjqglqmxOg1JqCTAUiNNat88r+wpondfEE0jWWncu4dzTQCpgBky1ZeisQOwhWH0/5GbAns/AyRPaDoV2I6FZXzDa2zpCIUQVW3BgATnmHJ7u8rStQxG1SFVOhPwc+BhYml+gtS5Ytkwp9T6QculpBfprrWvfvaeMRFg1Hpw8YOIOiAuDQ2vh0DrYtxycvaDtMGsCEXAzGGVuqxDXu8jUSNYcW8NdLe/Cv46/rcMRtUiVXWG01luVUgEl1SnrlN+7AXleqDCLGb55BC5EwYMbwcvf+tV6MORmwYlfrQnEwW9h71Jw9oag4dYEwv9GSSCEuE7N3T8XO2XHE52esHUoopapLleVm4BYrfXxUuo18ItSSgPztdYLqi40G/r1dTjxGwybDX7di9bZO0GbO6xfuZkQnpdAHFgDez4HF59CIxA3gsFok29BCFGxjiYeZePJjTzc/mHqu1T4fkRCXFZ1mQg5DvjyMvU3aq27AoOBp5RSN5fWUCk1QSkVopQKiY+Pr+g4q87Bb2DbRxD8MHR74PJt7Z2tcxxGL4ZpJ+DuZdDsZjjwFSwdDu+3hg3PwKk/raMXQogaa/a+2bg5uPFQ+4dsHYqoIMW3xj569KiDUqrbm2++WZAV3n///U1nz55dF+CWW24J/Pjjj+vm140dO9Z/+vTpvvnH0dHRdnZ2dl3ffffdevllqamphn79+gU2a9asXWBgYLuJEyc2vppYbT7SoJSyA+4CupXWRmt9Lu81Tim1FugBbC2l7QJgAVjXaajwgKtCzD+w7mnw6wWD3infufbO1lsUQcMhJwOO/2Idgdi/EkIWg2t9CLrTOgLRtJeMQFQirTVagzZrLFqjLRqLWVvLLUXbXTzIL7t4UKQ6/6BIWdGTtS5aX/wzCt5q0JTWX8l9lbsfrSka3qX96kKd6cJ9lNLvtaqZvxSswpOOc+qfOB4MfJrkcBPJnC+xndHOQJPWXlUcnbhahbfG7tKlSxSAt7e3af78+fWnTp0a7+TkVOSf7bx58yJuueWW1mPHjk3et2+f0969e12XLVt2Jr9+6dKlXp06dUpfs2aN97Rp0wr+ep46dWrssGHDUrOyslSfPn1arV69us7dd999oTyx2jxpAAYAR7TWJe7GpZRyBQxa69S89wOBN6oywCqVkQir7rVOfLx7Kdg5XH1fDi7QboT1Kyf9YgKxbznsXghuDS4mEH49wVAxA09aaywWbb1Y5l0oLea8i6ZFYzFbrGWWixdSS6G2BecValvQzqKLHps1Fkuh/gp/Zv6xtr7mf771lSLHRevK2obL1mlLTb48ierqDp4k9whsILTUNs51HHj43RurMCpxtfK3xt68efPR4cOHt/zwww8Lkobu3bunzZ07t+7UqVOLPATQunXrnPvvvz9+0qRJTfbu3ev60UcfRdjbX3yKbs2aNd4zZ86MfOCBB5qfOHHCvkWLFrnu7u6WYcOGpQI4OTnpjh07ZkRGRpb7AlOVj1x+CfQDfJRSZ4FXtdaLgbEUuzWhlGoELNJaDwF8gbV5y6PaASu11j9VVdxVymyCrx+G1Gh46Edw973yOWXl4GpNDtqNhOw0OP6zNYHY+wXsmg/uDSEoL8Fo0gONIisjl8wLuWSm5ZCZmktmag6ZablkpeaQkZpLVpr1ODM1B1NO0UTAVgxGhcGgUHmvBqNCGfLK1MVyaxkX6/LqDca8L3uDtcxQuH1eOwMllF3ab0nnGvLOz1/ut2DV30LL/xZeCbhgWeAiZQXvLh4Xb1ZSfwrUJQ25ZOnh/P5KbnuxQ1VaP6pwHHn9FInhYmdF2hU9qdBnFPtZ1VJ74/byfsj7PNL+kSvuMWEw1vIf1tX47ik/4sIqdGts6gdlMGJuubfGrl+/vgngP//5T/SQIUNa/d///d8lTw6+9tprsQEBAe27d++eNnjw4LT88vDwcPv4+Hj7/v37ZwwfPjxp6dKl3q+//nps4XMTEhKMmzZt8nzuuedii/d7JVX59MS4UsofLKEsChiS9/4k0KlSg6sufn0dTm6B4R9Dk4pdisJi0WSn51ov/mm5ZGbfRKZvTzId/0Pm2ZNkxkaR9WMmGT8cIUvHkmV2Q1PyLx5HFzuc3R1wdrfHs74LDVp4YO9otF6kDRcvvMqgMBgNJZTlfRku1hW+yBe96F/ufOux0WhAGRVKXXoBFOJ6YLaYWfjPbJwbw5ibh2JvkLVZrherV6/2njx5chxc3Bp76tSpcQBBQUE5Xbp0SZs/f/4l22Xv2rXL2WKxqBMnTjiZzWaMRuut5qVLl3oPHz48CeC+++5LfOSRRwIKJw25ubncddddzSdMmBAbFBRU7pUCq8PtCQHwz9fw92zo/ih0ve+KzfOTgIzUHLLyE4HUnILRgCIjA2k5ZKXlUto2I44ubji7t8e5iQEvfR6nrJO4pIXhRBLOrnY4t+iIc9BNOLfohJO7A0ZjdZk/K0TtsPHURsKTw3mv73uSMFSWK4wIVIbStsZ+5pln4vLbvPLKKzF33313i169eqXml5nNZp566qmmixcvPvXpp5/We/fdd+u9+OKL8QDffPONd3x8vP23337rDRAXF2f/zz//OHbo0CEbYPz48QHNmzfPeuWVV+KKx1MWkjRUB9EHrBMfm/aG298uscnJffEc2BJJRl4ykJWeW+qMrsIjAV6+LjgHeuDs7oCTmz0u7g44udvj7Gatd3KzLzkJyEqBoz9Zb2GceAdOzAAPv7w5EiOhUVcZLxaiCuSYc5i7fy5tvdsy0H+grcMRFSh/a+yVK1cWTGLs3r1761OnThXMNejSpUtWy5YtMzdv3uzRvXv3dICZM2fWa9asWfbQoUNT27dvn9W7d+82DzzwQFJCQoIxPT3dGBcXdyD//ClTpjT64osvvGfOnBk9efLkRhcuXDCuWrXq9NXGLEmDrWUkwlf3Wld2LGXiY0TYeX5eeBD3uk7UbeJGo7wkwDnv4u/knpcMuF0mCSgvJw/odI/1KzMZjv5oTSB2fAp/zwHPptbkIWgENOpS+QmE1tbHRbW52KsFLKYS6vLKtcV6rC15dbrYsaVYm+LHhdsUPjYXKyup38v0WfQxiRKOi33vJbYp6RGKSm5T7EmNy8d/pTZlPL6k32tVsyaorjHFc84UxSv2jhhW31+2k5w84M65lRuYuGalbY399ttvNyxcNn369Og+ffoEAZw7d85u9uzZDbZv334YICAgIPfxxx+Pmzx5chN/f//sIUOGJBU+d+zYsUnjxo1r/uSTTybMmTOnYbNmzbLatWsXBDBhwoS4Z555plwrLcvW2LZkNsHyuyBiBzz8IzS+9KnTuDMX+O6DfdTxcWbks11xdLZxnpeZBEc2WhOIk1usF2ZPf6jXJu8iXfhinn8BN5VQZr54US8pGSheVvgZxRpLWR9xVQbrV+FyKDpjscjx5dpQhjZl6ac8ba4US0mzNq+2zxLiqgg1ZJQsHc0QlwwCLQYWZTmhyvozcPaGh364qs+UrbHF5bbGlpEGW9r8Kpz6A+78pMSEISU+gw0fh+Lkas+wSZ1snzCAdUSky73Wr4xEOLoRwtZBWmzeBdFofTXYgZ1d0bKCuuJlhrzX4u0NF/sqXlakvaGEPu0u9ltQX/jYcPHifUkbQynnGK0XmxLb59WVeE5+m5pxoRLVx9LQeSTu/4R/D1uGqtfB1uEIIUmDzRxYA9s/hh4TrBfgYjIu5LB+digWi2bY5E64ejraIMgrcPGGLv+yfgkhKlRiViJfHPqCAU0H0EESBlFNyDR4W4gOhe+fBv8+cPtbl1TnZJn4YW4o6cnZDH2qE14NXG0QpBDClhb9s4hMUyaTukyydShCFJCkoaqln4dV/wKXujDmCzAWfXzKbLbw88KDxEekMvDRdjRo7mGjQIUQthKdFs2qI6u4s8WdNPdsbutwhCggtyeqktkEXz9ovf//8E/gVq9ItdaaLcuOEHEokX73tqZZp3ol9yOEuK7N3T8XheLJTk/aOhQhipCRhqq06RU4tRWGfQSNu15SvWPdSY7uiKH70Ga0u+mqNiATQtRw4UnhrD+5nrFtxtLQreGVTxCiCknSUFVCv4Idc6HnE9D50hW1D2w5y96fzhB0UyO63xFQ9fEJIaqFOfvm4GLnwmMdHrN1KKIKGI3Gbm3atAlq3bp1UFBQUNtNmza5Avz999/OnTt3bhMYGNiuVatWQQsXLizYtrRHjx6tGzZs2MFiufgo+oABA1q4uLh0qex4JWmoClH7Yf1k8L8RBv73kurwPXH8ufoYAR196Du2leyfIEQtFRofym+Rv/FguwfxdPK0dTiiCjg6OlqOHDkSdvTo0bAZM2ace+mll5oAuLm5WZYtW3YqPDz80C+//HL8pZde8ktISDDmn+fu7m7etGmTG1g3oIqLi6uS9cUlaahs6Qnw1b/AxQfGfH7JxMdzx5LY9NkhGjSrw8BH22GQfR2EqJW01szaMwtvJ2/uC7ry/jPi+pOSkmL08PAwAXTs2DE7f7+IgICAXG9vb1N0dHTBPMS77rorccWKFd4Ay5cv9xw2bFhyVcQoEyErkzkX1jwI6fElTnw8fy6NjfP+wcPHmTsmdsLewVhyP0KI697fUX8TEhvCiz1exMW+YndoFlc2fdt0v/Ck8Ar9wQd6BWbM6DPjshthZWdnG9q0aROUnZ2tEhIS7Ddu3HiseJstW7a45ObmqqCgoOz8soEDB6Y+8cQT/iaTiTVr1ngvWbLkzIcffljpk2AkaahMv0yH03/CyPnW/RkKSU3MYv2cUOwdDAyd1AknN9m5TojayqItzNo7i8ZujRnTaoytwxFVKP/2BMDmzZtdH3rooWbHjh07ZDBYR53PnDlj/9BDDzVfvHjxqfztrwHs7Ox0jx490hYuXOidlZVlaN26dbm3ub4akjRUltBVsHMe9HwSOo0tUpWVnsv62fvJzTIx8tlu1KnrbKMgax+tNZitG1ppiwXM5qKvFgvabN0PI/+15LY6r01evdkMBWVFz7/Yb6G2moubMJWyYZMuXl/QrIztSzrniueWsW0Z+r1i/IXOKXvbilD99ts5mXySrpGH6NukL0mnZ19zfwZXV3yeeKICIqs9rjQiUBUGDBiQnpSUZBcdHW3XuHFjU2JiomHw4MGBr7766rlbb701vXj7e++9N3HcuHGBzz33XFRVxShJQ2U4txe+nwwBN8HAGUWqTDlmNn5ygJSETIZP6oxPEzcbBVl+Wmt0ZibmlBTrV3Lea0oyOis776JoRpsuvmqLGUx5F1ezuUgbbTaB2VL2urzXksouW1fool+xFx9RYZS6uDdHKa8VOj24mk02drHkcgfgsPdvEvn7mvsz1q0rSUMNtG/fPieLxYKvr68pKytL3XHHHYFjx449/9BDDyWV1P72229Pmzx5cvTDDz+cWFUxStJQ0dLirRMf3epfMvHRYtH8svgQ0SdTGPhIOxq39iq9n0qktcaSllbowp+MJT8RKJIMXEwKzCkpWJJT0Lm55fswoxFlMICdXbFXI8pYQpnBWPCqjMaCOuXoULTOzro5lTIarZ9hLKnOAEY766sygNFg7SP/1WCw1uW3NRjBoIq2MRqsfatS2hrz+8l7NRRtY43BuplV8bZWeRfEgs0dL3/hLL6jZJEnba54brFzLvtahrhKu6hfpk1B22p20bal1UdXM2PHDObeOpebm9xs63BEFcuf0wDW383z5s07bWdnx4IFC7x2797tlpSUZLdy5UofgCVLlpy64YYbMvPPNRgMvPHGG7FVGW+VJQ1KqSXAUCBOa90+r+w14DEgPq/ZS1rrjSWcOwj4CDACi7TW/6uSoMvLnAtrHoCM8/Dwz+DqU1CltWbrqmOcCk3gxrtb0jLY95o/TpvNWFJTi17gk1MwJyeXeMEvKLtwwfpXdykMLi4YPD0wenhi9PDAsUUgRg8P65en9dWQf+zhidGjDgYnp6KJQf4FUi4OQpQq05TJvNB5dK3flZsa32TrcIQNmM3mPSWVT5w4MXHixIkljiDs2rXraEnlGRkZ+yoytpJU5UjD58DHwNJi5R9qrWeWdpJSygjMBW4DzgK7lVLfa63DKivQq/bzy3BmG9y1EBp1LlK158fTHNp6ji4Dm9LpFr9ydWvJySF51Vekbd1aJEGwXLhw2eF2g7v7xYu9hwf2jRsVu9hfTALyvwweHhgcHK7q2xdClM+KwytIyEzg/b7vS4ItaoQqSxq01luVUgFXcWoPIFxrfRJAKbUKuBOoXknDvhWwaz70ego63l2kKmxbFDu/P0Xrng3oPaJFmbvUZjMp368nfs5sTFHROLZsiV2DBjj4+2P09Lzkr/+CEQBPT4zu7ig7ufskRHWVkp3Ckn+W0LdJX7r6XrqsvBDVUXW4qjytlLofCAGmaq2LT/hoDBSe1XoW6FlVwZXJuT2wYQo0uxlue6NI1el/Evh9xVH8grzpf38blOHKf01orUnbsoX4Dz8k+3g4Tu3a0ei//8X1hhsq6zsQQlSxxQcXk5abxuSuk20dihBlZuvlB+cBLYDOQDTw/rV2qJSaoJQKUUqFxMfHX/mEa5UWB1/dB26+MPpzMF7Mw2JOpfDzgoP4NHFj0IT2GMuw2mNGSAhnxt/L2YlPoXNyaTzrQwLWrJaEQYjrSGx6LCsPr2Ro86G08mpl63CEKDObjjRorQtmfSqlFgIbSmh2Dig8CaBJXllpfS4AFgAEBwdX7vN1phxY/QBkJMIjv4Br3YKqpJh0fvj4AC6ejgx9uhMOTpf/UWcdOULchx+S/sdW7OrXp8Hrr+N510iUvSz6JMT15tMDn2LWZiZ2nmjrUIQoF5smDUqphlrr6LzDkcDBEprtBloqpZphTRbGAuOrKMTL+/kliPgbRi2Ghh0LitNTslk/JxRlgGGTOuFSp/SJhTmRkcTPnsOFDRswuLtT/9mpeN17LwZnWfBJiOvR6ZTTrD2+lnta30MT9ya2DkeIcqnKRy6/BPoBPkqps8CrQD+lVGesS7SdBh7Pa9sI66OVQ7TWJqXU08DPWB+5XKK1PlRVcZdq33LYvRB6Pw0dRhcU52SaWD8nlMy0XEY+0wXP+iUvZW5KSCDhk3kkrVmDMhqp++ij1H30EYweHlX1HQghbGDOvjk4GB14rKNsfS0gIiLCbuLEiU1DQ0Nd6tSpY/bx8cmdM2dOJMCkSZP8Tp8+7eTq6moOCAjInj9/fkRoaKjzuHHjWjRu3DgnOztb3XbbbSkLFiw4CzB9+nTfNWvW1AUwm83q5MmTTlFRUft9fX1Lf8a+nKry6YlxJRQvLqVtFDCk0PFG4JL1G2zmbP7Ex74w4PWCYrPJwo/z/yEpKp0hT3Wkvn+dS041p6ZyfskSEr9Yis7OxnP0aHwmTsTet35VfgdCCBs4lHCIX878wuMdH8fH2efKJ4jrmsViYfjw4YHjx48/v2HDhpMA27dvd46KirJ//PHHA95+++3I8ePHpwBs2LDBPSYmxg4gODg4bcuWLeFpaWmqQ4cOQb/88kvSwIED02fMmBE7Y8aMWICVK1d6zJ4927ciEwaoHk9P1CypsdYVH90b5K34aP0Raovm1y8Oc/ZIErc+2Bb/dnWLnGbJziZpxUrOz5+POSWFOkMGU2/yZBwCAqr+exBC2MRHez/C09GTB9s9aOtQRDWwYcMGdzs7Oz1t2rSCWfu9e/fOnDVrVt2uXbum5ScMAEOHDk3NO6dgopubm5tu165dZkREhANQZG+KL7/80nvMmDEVvry0JA3lYcqxrviYmQSPbgIX74Kqbd+Gc3x3LL1GNKdNr4u7k2qTiZR164if8zGmmBhc+/Sh3pQpOLdvZ4vvQAhhIzuid7A9ejvPBT+Hm0PN2XOmtoh66WW/7OPHK3RrbMeWLTMavfVmqRthHThwwLlTp04ZxcsPHjzo3LVr10vKi4uPjzeeOnXKceDAgamFy1NTUw1bt271WLRoUcTVRV46SRrK46cXIGK7deJjgw4Fxfs2RRC6OZIO/ZvQ9XZ/wLrWQurmzcTP+oicEydw6tiRRv97G9devWwVvRDCRrTWfLTnIxq4NuCeNvfYOhxRw4WEhLi1bt06KCIiwvGRRx6Ja9q0qalw/apVqzy6deuWVtG3JkCShrLbuxRCFsMNk4tMfDy2K4a/vwmnRdd63DimJUop0nfuIu6D98kKPYBDs2Y0nv0R7rfdJsvEClFLbY7YzMHzB3njhjdwNDraOhxRgsuNCFSWDh06ZH733XeX7FzYrl27rK1bt5Y6HJU/p+HIkSMOffr0aTt+/PjEwhtZrV692vvuu++ulJ0vbb24U80QuRt+mArN+8OA1y4WH0nk1y8O06ilJwMeCiL7cBgRjz5GxAMPYIqNo+F/Z9B8/ffUGThQEgYhaimTxcTsvbNp7tGcYS2G2TocUY0MGzYsNScnR82cObNgVuzOnTud27Ztm7Vnzx63VatWFTxO9+OPP7rt3r3bqfD5bdq0yZk8eXL022+/3SC/7Pz588Zdu3a5jx8/PrkyYpak4UpSY2H1feDeEEYvsW6JDMRHpPLjp//g1cCFAUM9iX3+OU6PGk3WP/9Qf9o0Wvz0I56jR8v+D0LUct+f+J7TF04zuctk7Azy+0BcZDAY+P7770/89ttvdfz8/NoHBga2e/755xs3btw4d926deFz586t7+/v375Fixbt5s6dW79Bgwam4n1MnTo1fufOne5Hjx51AFixYoXnTTfddKFOnTqWyohZ6cvskljTBQcH65CQkKvvwJQDXwyDmAPwyCZo0B6ACwmZfP3uHgxKc7PDn2Sv/RJlb4/3A/dT95FHMLq7V9B3IISoybJMWQxdOxRfF1+WD1leI0YclVJ7tNbBto6jKoSGhp7u1KlTgq3jqG5CQ0N9OnXqFFBSnaS9l/PT8xC5A0Z/VpAwZKbm8P2svZjSMui2932y06LwunsMPk8+iV29ejYOWIirdyzpGCExIbg5uOFmn/fl4Ia7vTuuDq6427tjb5Rlzcvjq6NfEZsRy9s3vV0jEgYhrkSShtLs+RxClkCff0P7uwDITknnuzf+IDXVSOf9s2lwc0fqTV6IQ9Omto1ViGsUmRrJAz8+QFpu2mXbORodcbV3xd3B3fpq746bg1uJZflJR+EExM3eDRd7Fwzq+r8zmpqTysJ/FtKnUR+6N+hu63CEqBCSNJQkchf88Cy0uBVufQVtMpH49bdsXp9IolsgwWyj0xczcWrb1tbQ0yBxAAAgAElEQVSRCnHNcs25TPtjGgrF18O+xsXOhdTcVNJz00nNSSUtN420nDTra/77QscRqRFFyjSXv+WpULjZu+Hq4FrqiIarvau1LC8RaebRjOYezavoJ1IxPjv4GSnZKbL1tbiuSNJQXEaidatrj8bouxaS+stm4mbN4oBjHxIa3kDvGxzpev+rto5SiAoza+8sDp4/yIf9PqS1d+tr6ktrTYYpoyCBSM3JSz5yU0nPSS9aVqguKSuJs6lnC5KUbHP2JX138OnAiMARDG42GHeH6j1vKCEzgeWHlzMoYBBBdYNsHY4QFUaShuKcvaDP/5Ge6EXcA0+QdfAgZ7reR3SdXgQPCaDr8Jr1144Ql/N75O8sDVvKuDbjGOA/4Jr7U0rhau+Kq70rvvhedT+55tyCkYzUnFRCYkJYG76WGTtm8O7udxngP4ARgSPo0aBHtbzVMT90PrnmXJ7u8rStQxGiQknSUIw5NZVzC0NI/3s7do0akvrkTE4cdiaoT0N6DGtm6/CEqDDRadG8/NfLtPVuy9TgqbYOpwh7oz1eRi+8nKzr3gTVDeK+oPsIOx/G2vC1bDy1kR9O/kAj10bcGXgndwbeSWO3xjaO2ioyNZKvj33NXS3vwr+Ov63DEaJCSdJQjMHdHeXkjO+LL5DY9jZ2f3aEgA516Tu+tcx+FteNXEsu07ZOw2Qx8V7f96rtKoU5JgubD8dyNMa6tL5S9rhzD6PrjeBM5i6Opv/GvNBPmRc6j8aOHWjjdistXHpjn/f9KEAp6zyKwv/7KqUK1VHw/3b+MUXqi9YVlBU6t6BcwdcR76EwEmB3J+tDo4p8fn47UBjUpecW7teg1CXnkndccG6xPvOP7Y0G2jcuWBdIVGOlbY09atSowOPHjx8q3HbUqFEBQ4cOTXnooYeSsrOz1ZQpUxr98MMPXq6urmYHBwf90ksvRd19990XJk2a1HjNmjV1L1y4YMzIyNhXkfFK0lCMUgq/T+YSFZ7M5ln78Q2ow8DH2mMwVr8hUCGu1if7P2F//H7euemdavnX8NmkDFbtimTV7kgS0i6d32DlC4xD2Q3G3nMPkR57OJc9i81xn5J7oRO5ycFYspqQd7mtEgbHKFya/U7O+b5MD4sEqnxlYgB83BwJ+c+1324SletyW2Nf6dwpU6Y0iomJsT9y5MghZ2dnHRkZaffzzz+7A4wYMSL52WefjWvbtm37io5ZkoYSnI9KY+MnB3Cv68QdT3XE3sFo65CEqDDbzm1j0T+LGNVyFEOaD7F1OAXMFs3WY/Es33GGLUfj0MAtrevzr17+3NyqHgYFWoPGOuES8t+DZixmi4V9cXv4/sQ6fo3cTLbXTprVac7Q5ndyu/8QvJys29VrbT1Rowv6I6/Pi/1drChcdsnn66J1r+96hsNJbswf9yKudnUAjaVYm1LfF4uhyLklxGYp5Vy0daRBVH+lbY2dv7pjaVJTUw0rV66sd/LkyQPOzs4awM/Pz/Too48mAdx6663plzv/WkjSUExWWi4b5oRitDMwbFInnN0u+99OiBolPiOel/56iUDPQJ7v8bytwwEgPjWb1SGRfLkrgrNJmfi4OTKxXyBje/jRxKvoTsUXbzGUNHpgpE+TXvRp0ou0nJf56fRPfBf+HXP2f8i80Dnc1OQmRgSO4KYmN2FvqPhFqkJiQgiJ/5t/d/03nRo1qvD+ReX6delhv8RzaRW6NbZ3Y7eMW+9vW+6tsa8kLCzMsWHDhjne3t6VslT05UjSUIyjqx3t+zamabu61PFxtnU4QlQYs8XMC3++QKYpk5l9Z+JsZ7t/31prdp5KZPmOM/x8KIZcs6Z387q8OLgttwX54mB3bX8puzm4MbrVaEa3Gs3J5JN8d+I71p9Yz5bILXg7eTOs+TBGBI4g0Cuwwr6fj/Z+RH3n+oxvO75C+hSiOpKkoRilFN0GBdg6DCEq3IIDC9gVs4sZfWbQwrOFTWJIyczl271nWbEzgvC4NOo42XFfrwDG92xKYP1SdwK+Js09m/NMt2eY3GUy285tY234WlYcXsEXYV9U2NoPf5z9g/3x+3ml9ys2TcbE1bvciEBlKW1r7CsJCgrKjo6OdkhMTDRU9WhDld34UkotUUrFKaUOFip7Tyl1RCl1QCm1VinlWcq5p5VS/yil9iulrmEHKiFqp13Ru5gXOo9hzYdxZ4s7q/zzD5xNZtrXofR8azOvrw/D1dGO90Z3ZOdLA3hlWFClJQyF2Rns6OvXl1n9Z7F5zGaeC36OTFMmM3bMoP/q/rzw5wvsiN6BRZfvd7DZYuajvR/hX8efEYEjKil6cT0qbWvsU6dOXfa+uLu7u2Xs2LEJEyZMaJqVlaUAoqKi7JYsWVLuBKS8qnK2zOfAoGJlm4D2WuuOwDHgxcuc319r3bm27L4mREU5n3meF/58Af86/vyn13+q7NHhjBwTX+2OYPjHfzH8422sD41mZJfGbJh0I+ue6sOYYD+cbTTJuK5zXe5vdz/fDv+WVXesYkTgCLae3cpjvzzG4G8G88n+TziXdq5Mff1w6gfCk8N5usvTlTJXQly/Lrc19qlTpxx9fX075n8VTwhmzZp1zsfHx9SqVat2LVu2bDdo0KBADw8PM8ATTzzRxNfXt2NWVpbB19e34zPPPFNhk2yqdGtspVQAsEFrfcljIEqpkcBorfW9JdSdBoK11uXawvSat8YWooazaAsTN09kd8xuVt6x8pqXiS6L47GprNgZwTd7z5KaZaKVrxv/6uXPiC6NqeNUfS+qWaYsfov4je/Cv2NH9A40mp4NejKi5QgGNB2Ak53TJefkmHMYtnYYHo4erBq6qlquTllesjW2qClbYz8MfFVKnQZ+UUppYL7WekHVhSVEzfXZwc/YFrWN6b2mV2rCkGOy8NOhGFbsOMPOU4k4GA0M7tCAe3v60z3Aq0YsjOZk58SQ5kMY0nwI0WnRrDuxju/Cv+PFP1/kLfu3GNRsECMDR9Lep33B97Pm2Bqi0qN4tfer10XCIMSVVIukQSn1MmACVpTS5Eat9TmlVH1gk1LqiNZ6ayl9TQAmADSVLatFLbYvbh9z9s3h9oDbGdNqTKV8RmRiBl/uimB1SCQJaTn4eTvz/KA2jAlugo9b9VxlsiwaujXkiU5PMKHjBPbE7mHt8bWsP7GeNcfW0MKjBSNbjuQWv1tYcGABPRv0pHej3rYOWYgqYfPbE0qpB4HHgVu11ld8XlUp9RqQprWeeaW2cntC1FbJWcmM2TAGO2XH6mGrK3RXSLNF8/vROJbvOMPvx+JRwC1tfPlXr6bc3LIeBkP1H1W4Gmk5aQVrP4TGhxaUrxyykg71OtgwsooltydEtb09oZQaBEwD+paWMCilXAGD1jo17/1A4I0qDFOIGkVrzfRt063bMw9eXmEJQ1xqFqt3R/LlrkjOJWdS392RSf0DGdujKY08r//HDIuv/bDuxDrc7N2uq4ShFrJYLBZlMBiq7q/nas5isSig1EeIqixpUEp9CfQDfJRSZ4FXsT4t4Yj1lgPADq31E0qpRsAirfUQrAvMr82rtwNWaq1/qqq4hahplh9ezu9nf+f57s/TzqfdNfWltWb7yfOs2BHBz4diMFk0fQLr8p872jIgyLfWLlfc3LM5U7pNsXUY4todjI+PD6pXr16KJA7WhCE+Pt4DOFhamypLGrTW40ooXlxK2yhgSN77k0CnSgxNiOvGwYSDfLDnA/r79efetpc8iFRmKRm5fL33LCt2nuFkfDoezvY8eIN1Eabm9Sp/TQUhqoLJZHo0JiZmUUxMTHuqdgmC6soCHDSZTI+W1qBK5zRUNZnTIGqT1JxUxqwfg0VbWDNsDR6O5dsaWWtN6NkUlu84w/rQKLJNFro09eRfPf25o2NDnOxl47baoDbNaRDlVy2enhBCXButNa/+/Sox6TF8PujzcicMeyOSeOuHw4ScScLFwciobk24t2dT2jUqXz9CiOtbuZOGvMmIWVprcyXEI4S4CquPrmbTmU1M6TaFzvU7l/m8yMQM/vfTEX44EE09d0deH96Ou7o2xr0aL8IkhLCdKyYNSikDMBa4F+gOZAOOSqkE4Aesiy2FV2qUQohSHUk8wru73+XGxjfyYLsHy3ROSmYuc7eE8/m20xgMMPnWljx+c3NcHWXwUQhRurL8htgCbMb6pMNBra27uSilvIH+wDtKqbVa6+WVF6YQoiTpuek898dzeDp68uaNb15xVcJcs4XlO87w0a/HScnMZVTXJjw7sDUNPC5dIlkIIYorS9IwQGudW7xQa50IfAN8o5SSsUwhqpjWmhk7ZhCRGsGigYvwdvK+bNtfwmL5349HOJWQzg0t6vLyHW1lzoIQolyumDSUlDBcTRshRMX6Lvw7fjj5A091foruDbqX2u7A2WT++8Nhdp1KJLC+G0seDKZ/6/o1Yj8IIUT1UuYbmEqpwcBrgCcQCnyotd5eSXEJIS4jPCmct3a+Rc8GPXmsw2MltjmXnMl7Px3hu/1R1HV1YMaI9ozr7oddLV2QSQhx7coz6+kT4F9AGNANeE8pNVdr/WWlRCaEKFGmKZNn/3gWF3sX3r7pbYyGousnpGblMu/3Eyz+6xQamNivBU/2ayFPRAghrll5koY4rfW2vPeblVLbgZ2AJA1CVKH/7fofJ1NO8ultn1LPpV5BuclsYdXuSGZtPkZCWg4jOjfiuUFtaFwL9oUQQlSN8iQNp5RS/wXe0FrnALlYt7MWQlSRDSc38O3xb3msw2Pc0OgGwDrJccvRON7aeITwuDR6BHiz+IG2dPLztHG0QojrTXmSBgswEnhMKXUcaAp8q5RqqbU+XinRCSEKnE45zYztM+havysTO08EICzqAm9uDGNb+Hma+bgy/75uDAzylUmOQohKUeakQWs9HkAp5Qi0x7qJVCdgoVKquda6aeWEKITINmfz3NbncDA68M7N73A+zcTMnw/x9d6zeDjb8+qwIO7t6Y+DnUxyFEJUnrKsCKl0oV2ttNbZwJ68r/w28ptKiEr03u73OJJ4hPdvns3Kvy+wcOt+zBbNYzc156l+gXi4yCRHIUTlK9OKkEqpb4B1WuuI/EKllANwI/AA1lUjP6+UCIWo5X45/QtfHf2KXt4j+c9KM3Gpx7mjY0Oev70NTeu62Do8IUQtUpakYRDwMPClUqoZkAw4Y917/BdgltZ6X+WFKETtFZkayct/vYK9KYBN27rRpakz8/7VjW7+XrYOTQhRC5VlRcgsrGs0fJK3XLQPkKm1Tq7s4ISozQ5FJ/Lwz0+RaTFTJ+l+Ph7fnTs6NJRJjkIImynXlnZa61ylVD/gTqWUCetow3pZ4EmIihOfms0Hm46x9swn2Huf4o6GL/DG/XfhaGe88slCCFGJrmYf3H5a67H5B0qpucgCT0Jcs8wcM4v/Osm830+Q63gQxyZ/cVeLe3j9xnttHZoQQgDWkYLyclRK3aGU6qiUGoJ1fkOZKKWWKKXilFIHC5V5K6U2KaWO572WeLNWKfVAXpvjSqkHriJuIaoli0Xz7d6z3PL+78z85RjdAxU+zdbS1rstL/V+ztbhCSFEgatJGiYCXsCQvNeny3Hu51gnVhb2AvCr1rol8GvecRFKKW/gVaAn0AN4tbTkQoiaZPuJ8wyf+xfPrA6lnrsjKx4NxuyzHLM28V7f93A0Oto6RCGEKFDu2xNa6wxg+dV8mNZ6q1IqoFjxnUC/vPdfAL8DzxdrczuwSWudCKCU2oQ1+ZDbIqJGOhGfxtsbj7D5cCyNPJyYdU9nhndqxJz9s9kfv593b34X/zr+tg5TCCGKuJo5DUUopZ7XWr9zDV34aq2j897HAL4ltGkMRBY6PptXVlI8E4AJAE2byiKVonoxWzTv/XyURX+exMneyLRBrXm4TzOc7I1sO7eNRf8sYlTLUQxuNtjWoQohxCXKnTQopVYXPgQ6A9eSNBTQWmullL5yy8v2sQBYABAcHHxNfQlRkdKyTUz+ch+/HYnj7uAmTBvUBh836+2H+Ix4XvrrJQI9A3m+R/GBNiGEqB6uZqThgtb60fwDpdS8a4whVinVUGsdrZRqCMSV0OYcF29hADTBehtDiBohKjmThz/fzfG4NGaMaM99vS7eejBbzLzw5wtkmjKZ2XcmznaylbUQonq6momQbxY7fvkaY/ge61LU5L2uK6HNz8BApZRX3gTIgXllQlR7B84mc+fcbZxNymTJg92LJAwACw4sYFfMLl7q+RItPFvYKEohhLiyMo80KKWeKfQ+/20K1o2rEsvYx5dYRwx8lFJnsT4R8T9gtVLqEeAMcHde22DgCa31o1rrRKXUDGB3Xldv5E+KFKI6++lgNP/+aj91XR1Z/mRPWjdwL1K/K3oX80LnMbzFcEYEjrBRlEIIUTaq0AaWl2+o1EogGFifVzQUOAAEAGu01u9WRoDXIjg4WIeEhNg6DFELaa2Zv/Uk//vxCJ39PFl4fzD13Is+Pnk+8zxj1o/B1d6Vr4Z+hYu9bD4lbE8ptUdrHWzrOET1VJ45DU2ArlrrNACl1KvAD8DNWEcbql3SIIQt5Jot/GftQb4KieSOjg15f0wnnOyLLgFt0RZe/utlUrJTmDdgniQMQogaoTxJQ30gu9BxLtbHJTOVUtmlnCNErZKSkcuTK/bw94nzTLolkCkDWmEwXLrB1GcHP2Nb1Dam95pOa+/WNohUCCHKrzxJwwpgp1JqHdZHLYcCK5VSrkBYZQQnRE1y5nw6D32+m8jEDN4f04lR3ZqU2G5/3H7m7JvD7QG3M6bVmCqOUgghrl6Zkwat9Qyl1I9An7yiJ7TW+RMGZEcdUavtPp3IhKUhaGD5Iz3p2bxuie22nt3KC1tfoKFrQ17t/apscy2EqFHKu05DLmABdN57IWq9tfvO8vzX/9DEy5nFD3anmY/rJW0s2sL80PnMC51Ha+/WzOo/C3cH9xJ6E0KI6qvM6zQopf4P6y0KH6zzG5YrpSZVVmBCVHdaaz7YdIwpX4XS1d+TbyfeUGLCcCHnApN/m8wnoZ8wrMUwlg1eRmO3EldBF0KIaq08Iw2PAD211ukASql3gO3AnMoITIjqLCvXzLSvD/B9aBSjuzXhrZEdcLC7NAc/lnSMf2/5N9Fp0bzc82XuaX2P3JIQQtRY5UkaFGAudGzOKxOiVjmfls1jS0PYG5HMtEGtebJvixITgY0nN/La9tdws3fjs0Gf0bl+ZxtEK4QQFac8ScNnWJ+eWJt3PAJYXPEhCVF9HY9N5eEvdhN3IZtP7u3KkA4NL2mTa8nlwz0fsixsGV3rd+X9fu/j4+xjg2iFEKJilefpiQ+UUn9w8emJh7TW+yonLCGqn7+OJ/Dkij042hn56vHedPbzvKRNQmYCz/7xLHti93Bv23uZGjwVe4O9DaIVQoiKV66nJ7TWe7Cu/ihErbJyZwTT1x2kZX03Fj0QTBOvS1dwDI0P5Zktz3Ah5wJv3fgWw1oMs0GkQghRea6YNCilUrE+YnlJFaC11nUqPCohqgmzRfO/Hw+z8M9T9GtdjznjuuDuVHTkQGvNmmNreHvX2/i6+LJ8yHJZ5VEIcV26YtKgtZaHyUWtlJFj4v9W7WdTWCwP9PZn+tAg7IxFn5DIMmXx5s43+S78O/o07sM7N72Dh6OHjSIWQojKVd7FnVBKDdNar79ySyFqrpiULB75YjeHoy/w2rAgHuzT7JI2UWlRTPl9CmHnw3i84+M82elJjAZjCb0JIcT1odxJA/AmF7fHFuK6c/BcCo98sZu0LBOLH+hO/zb1L2mzPWo707ZOw2QxMeeWOfTz61f1gQohRBW7mqRB1mYQ161NYbFM/nIfXi72fP3kDbRtWHTKjtaaJQeXMHvfbJp7NGdW/1n41/G3UbRCCFG1riZpKGlSpBA1mtaaxX+d4s2Nh+nQ2INF9wdTv45TkTbpuelM3zadTWc2cXvA7bxxwxu42F/6FIUQQlyvriZpEOK6kmu28Nr3h1ixM4JB7Rrw4T2dcXYoOjfhZMpJ/r3l30RciODZ4Ge5P+h+WQ5aCFHr2DxpUEq1Br4qVNQceEVrPatQm37AOuBUXtG3Wus3qixIcd26kJXLUyv28ufxBJ7o24Jpt7fGYCiaDPx65lde3vYyjkZHFty2gB4Ne9goWiGEsK2rSRpiKzIArfVRoDOAUsoInAPWltD0T6310Ir8bFG7RSZm8PDnuzmVkM67ozpyd3e/IvVmi5k5++aw+OBiOvh04IN+H9DAtYGNohVCCNsrd9Kgtb6tMgLJcytwQmt9phI/Qwj2nEliwtIQcs0Wlj7SgxtaFN0bIikriee3Ps/26O2MbjWaF3u8iIPRwUbRCiFE9WDz2xPFjAW+LKWut1IqFIgCntVaH6q6sMT15PvQKJ5dE0pDDyeWPNidFvXcitSHnQ9jypYpxGfG81rv1xjVapSNIhVCiOql2iQNSikHYDjwYgnVewF/rXWaUmoI8B3QspR+JgATAJo2bVpJ0YqaSGvNx7+F8/6mY/QI8ObT+7rh7Vp09OC78O+YsX0GXk5eLB28lPY+7W0UrRBCVD+GKzepMoOBvVrrS+ZMaK0vaK3T8t5vBOyVUiXuNay1XqC1DtZaB9erV69yIxY1RrbJzNTVoby/6Rh3dWnMskd7FEkYcs25/HfHf5m+bTqd63fmq6FfScIghBDFVJuRBmAcpdyaUEo1AGK11lop1QNrsnO+KoMTNVdieg5PLNvDrtOJTL2tFU/fEljkccnY9Fim/jGV0PhQHmr3EJO7TsbOUJ3+1xBCiOqhWvxmVEq5ArcBjxcqewJAa/0pMBp4UillAjKBsVprWWRKXNGJ+DQe/nw30SlZzB7XheGdGhWpD4kJ4dk/niXDlMHMvjO5PeB2G0UqhBDVX7VIGrTW6UDdYmWfFnr/MfBxVcclaq70bBN/HIvnhW8OYG808OVjvejm71VQr7VmxeEVzAyZiZ+7H4sGLiLQK9CGEQshRPVXLZIGIa5V7IUsQk4nsft0IiFnEjkcnYrZomlZ340lD3bHz/vics8ZuRm8vv11Np7aSH+//rx545u4O8gO8EIIcSWSNIgax2LRnIhPY/fpJELOJBJyOomIxAwAnOwNdPHz4ql+LQgO8KZHM2+c7C8uCR15IZL/+/3/CE8KZ1KXSTza4VEMqjrNBxZCiOpLkgZxzXbH7CY2IxZvR2+8nb3xcvTC28kbe6N9hfSfbTLzz9kUQs4kEXI6kZAzSSRn5ALg4+ZAsL839/f2p3uAN0GN6mBvLDkJ2Hp2Ky/8+QIKxScDPuHGxjdWSHxCCFFbSNIgrtq5tHP8b9f/+D3y9xLr3ezd8HbyxsvJCy8nL+o61bW+dyx27GRNMvJXXEzJyGVPRKJ1JOF0IqFnU8gxWQBoXs+V24MaEBzgRXCANwF1Xa64cZRFW5gfOp95ofNo7d2aD/p9gJ+732XPEUIIcSlJGkS55Zhz+PzQ5yw8sBClFFO6TaGfXz+Ss5JJzEokMSuRpKyki6/ZiUSlRXEw4SDJWcmYtKnEfu1wRptdycl2wWJ2RZnd8HHxpldnX9rUb0iXRk3w96qPt5N3kSTjci7kXOClP1/ij7N/MKz5MKb3no6znXNF/0iEEKJWkKRBlMv2qO28tfMtTl84zW3+tzGt+7SLmzh5XPl8k9lCSOQ5/j51mn3nznE4LpqUnCSUMQ0cM/Gqk4uLVybKLp0sy2mSs/axL93EvlPw5amifbnauxaMZHg7eheMWOS/OhodmbV3FtFp0bzY40XGtRkn21kLIcQ1kKRBlElseizvhbzHz6d/xs/dj3kD5pVpTkBGjon9EcmEnLE+2bAvIpm0bOtIQyOPetwQ0JruebcaWvm6Yyy2LbXWmgs5F0jKSiIpO6nISEZSVhLns86TlJVEdHo0YefDSMxKLDKS4ePsw5JBS+hSv0vF/kCEEKIWkqRBXFauJZeVh1fyyf5PMFlMTOw8kYfbP4yj0bHE9vGp2ew5c3E+wsGoC5gtGqWgta87I7s0LpiP0NjzyrcJlFJ4OHrg4ehBAAFXbK+1JjU3tSCpaObRDA/HMgyBCCGEuCJJGkSp9sTu4b87/kt4cjg3Nb6JF3u8iF+dohMItdb8dDCGX4/EsedMEqcS0gFwtDPQ2c+TJ/u2IDjAiy5NvfBwrpinKS5HKUUdhzrUcaiDfx3/Sv88IYSoTSRpEJc4n3meD/Z8wPcnvqeha0Nm9Z/FLX63XDIfYH9kMq+vP8S+iGS8XR0I9vdiXA8/ggO8ad/IAwc7Wf9ACCGuJ5I0iAJmi5k1x9Ywe99sMk2ZPNL+ESZ0nICLvUuRdjEpWbz70xG+3XeOeu6OvDu6I6O7NsFgkEmGQghxPZOkQQBwMOEgM3bMIOx8GD0b9OSlXi/R3KN5kTZZuWYWbD3JvN9PYNaaif1aMLF/IG6O8s9ICCFqA/ltX8ulZKfw0d6P+PrY1/g4+/Duze8yKGBQkVsRWmvWH4jmfxsPE5WSxZAODXhxcNsi+zkIIYS4/knSUEtZtIV14ev4cM+HXMi5wL+C/sXEThNxc3Ar0i40Mpk3NoSx50wSQQ3r8ME9nenVvG4pvQohhLieSdJQCx1NPMp/d/yX/fH76VK/Cy/3fJnW3q2LtIm9kMW7Px3lm71n8XFz4J1RHRjdze+SdRSEEELUHpI01CJpOWnM3T+XL498SR2HOszoM4PhLYYX2eUxK9fM4r9OMXdLOCaz5vG+zXm6fyDuTpX/uKQQQojqTZKGWkBrzY+nfmRmyEwSMhMY02oMk7tOLrLokdaajf/E8NbGw5xLzuT2dr68NKQt/nVdbRi5EEKI6kSShuvcyeSTvLXzLXbG7CSobhCzb5lNe5/2RdocPJfCG+vD2HU6kTYN3Fn5WE9uaOFjo4iFEEJUV5I0XKcycjNYcGABX4R9gbOdM//p+R9GtxqN0WAsaBOXmsXMn4+yZjkypnQAABKBSURBVM9ZvF0ceGtkB+7pLvMWhBBClKzaJA1KqdNAKmAGTFrr4GL1CvgIGAJkAA9qrfdWdZzVndaa3yJ/451d7xCdHs2dLe5kSrcp1HW++MRDVq6ZJdtOMfe3cHLMFh69sRmTbm1JHZm3IIQQ4jKqTdKQp7/WOqGUusFAy7yvnsC8vFeRJzI1krd3vs2f5/6kpVdLvrjpC7r6di2o11rz86EY3tx4mMjETAa09eXlO9rSzEfmLQghhLiy6pY0XM6dwFKttQZ2KKU8lVINtdbRtg7M1rLN2Sw5uIRFBxZhZ7DjueDnGNd2HPaGiyMHh6JSmLEhjB0nE2nt687yR3pyY0uZtyCEEKLsqlPSoIFflFIamK+1XlCsvjEQWej4bF5ZrU4a/jr3F2/tfIvI1EgGBQzi2eBn8XX1LaiPT83mg01HWbU7Ek9ne2aMaM+47n7YGWUzKSGEEOVTnZKGG7XW55RS9YFNSqkjWuut5e1EKTUBmADQtGnTio6x2ohJj+Hd3e+y6cwmAuoEsOC2BfRu1LugPttk5vNtp5nzWzhZuWYe7tOMybe0xMNF5i0IIYS4OtUmadBan8t7jVNKrQV6AIWThnOAX6HjJnllxftZACwACA4O1pUWsI3kWnJZHraceaHz0FozuctkHmj3AA5GB8A6b+GXsFje2niYM+czuKVN/f9v797DrKrrPY6/PzDcL5MoaCAXDVQMSZQQFQUUFM3jtVMWWRlPCl4hLS+ZlMo55iU9BDZJGaZZx8qKlC6ikSiJFwRBJtBIEgQN7QyDCsPMfM8fe0HjdoC9B2avPczn9TzzPGv91u27vwwzH357sRdf/0R/PtK1407ObGZmtmNFERokdQBaRERlsnwScEPWbrOASyT9jMwNkBXN7X6GZ9c9y5Snp/C3ir8xsudIrhpyFT069ti2vXztBm58eBnz//YW/bp15N4vDWH4QV1TrNjMzPYkRREagH2BXyVPViwBHoiI30saDxARZcBsMv/d8hUy/+Xy/JRqTcXUhVOZsWQGPTr2YNoJ0xjec/i2bW9t3Mztj67gZ8/8g87tWnHDGR/ls0N6+b4FMzPbrYoiNETESuBj9YyX1VkO4OJC1lUs7n3pXmYsmcHZ/c7mmiHX0LakLQBV1bX8+C+v8j+Pvcy7VTV8/ug+TBzVjw+1b51uwWZmtkcqitBg2/fwyoe57bnbGN17NNcPvZ6WLVoSETxW/iZTZpfz9/XvMPygrnzjtP707dYp7XLNzGwP5tBQxJ5a8xTfePIbDNlvCDcfdzMtW7Rk+bpKbnpkGfNeXs9HunbgR+d/nJEHd0u7VDMzawYcGorU0vVLmTR3En336svtw+9g1frN/PgvK/jJglV0bFPC5P84lM8N7U0r37dgZmYF4tBQZKpranlyVTlXzb8QRSe2rP4Sx/zXfN6tqqFlC/G5ob2ZNOog9urg+xbMzKywHBpSVF1Tyyv/3MiS1RUsXVPBkjUVLHvzNVr2mA4taqhdcz49u5byqcGlHNajlCEHdKFnl/Zpl21mZs2UQ0OBbKmp5eU3NrJ0TQVLX88EhPK1G9i0pRaADq1bckj3VuzT9z7ei81MOWo6o/sO9mOqzcysaDg0NIItNbWseKNy2+zBkjUb+OvaDWyuzgSEjm1KOLR7Z8Ye1ZvDepQyoEcp3fcq4aLHxrPyn68z/cTpHNP94ym/CjMzs/dzaNhFVdWZgLAkCQgvramgfF0lVXUCwke7d+a8ob05bP9MQDhg7w60qDODUFNbw5V/vpLn33ieW46/hWO6H5PWyzEzM9suh4Y8bK6uYcW6jdsCwtI1FSxfV0lVTSYgdGpbwoDupXzxmD4M6JG5D6F3l/bvCwjZIoIpC6Yw5x9zuHrI1ZxywCmFejlmZmZ5cWjYjk1bali+rnJbOFj6eiYgbKnJPAOrc9sSBvQo5fxj/x0Qeu0kINSnbHEZP1/xc8YNGMfY/mMb46WYmZntFg4NWaqqaznrrqdYvq6S6tpMQCht14rDepQybtiBHJYEhJ5d2pE8K6PBHlz+IHctvosz+57J5UdcvjvKNzMzazQODVlal7TgkP06c/xBXbcFhP332vWAkO3RVY9y09M3MXz/4Uw+evJuP7+Zmdnu5tBQj9s/9YFnZ+1Wz657lqueuIqBXQdy6/BbKWnhPwYzMyt+/gziAlv+9nIue/wyenbqyfQTp9OupF3aJZmZmeXEoaGAVleuZvyc8XRo1YHvj/4+pW1K0y7JzMwsZ54XL5C3N73N+Dnjqaqp4t4x97Jfh/3SLsnMzCwvDg0F8O6Wd7lozkW88c4bzDhpBn336pt2SWZmZnlzaGhkW2q2MGnuJP769l+5c+SdHN7t8LRLMjMza5DU72mQ1FPSnyQtk/SSpA98YIGkEZIqJC1Kvq5Po9Z81UYt1z11HfNfn8/koyczoueItEsyMzNrsGKYaagGroiIhZI6Ac9LejQilmXtNy8iTkuhvgaJCG577jZm/302lx9xOWf1OyvtkszMzHZJ6jMNEbE2IhYmy5VAOdAj3ap23cyXZnLfsvsY238s4waMS7scMzOzXZZ6aKhLUh9gELCgns1HS1os6XeSPlrQwvL0m1d+w3ee/w5j+ozhax//mj/t0czM9gjF8PYEAJI6Ar8EJkbEhqzNC4HeEbFR0qnAr4F+2znPBcAFAL169WrEiuv3xOonmDx/MkM/PJQpw6bQQkWVy8zMzBqsKH6jSWpFJjD8JCIeyt4eERsiYmOyPBtoJWmf+s4VEXdHxOCIGNy1a9dGrTvb4n8u5oq5V3Bwl4O5c+SdtG7ZuqDXNzMza0yphwZl5u5/CJRHxHe2s89+yX5IGkKm7rcKV+XOrfy/lVz82MV0a9+Nu068iw6tOqRdkpmZ2W5VDG9PHAucByyRtCgZuxboBRARZcAngQmSqoH3gHMjItIotj7r3lnHhXMupEQllI0uY+92e6ddkpmZ2W6XemiIiCeBHd4pGBHTgGmFqSg/FZsrmDBnApVVlfzo5B/Rs1PPtEsyMzNrFKmHhqZsU/UmLn38UlZtWMX3Rn2P/nv3T7skMzOzRuPQ0EDVtdV89YmvsujNRdw6/FaO+vBRaZdkZmbWqFK/EbIpighufPpG5r42l2uPupaT+5ycdklmZmaNzqGhAaYtmsZDLz/EhQMv5NxDzk27HDMzs4JwaMjTA+UPcPeLd3NOv3O4+PCL0y7HzMysYBwa8vD7V3/Pzc/czMieI7lu6HX+eGgzM2tWHBpy9PTap7lm3jUM6jaIW46/hZIWvofUzMyaF4eGHJS/Vc7EP02kT+c+TD1hKm1L2qZdkpmZWcE5NOzEaxteY8KcCXRu3ZmyUWWUtilNuyQzM7NUODTswPr31nPhnAupiRrKRpexb4d90y7JzMwsNX5jfjs2Vm3kojkXsf699fzgpB9wYOmBaZdkZmaWKoeGelTVVDFx7kRW/GsF3z3huwzsOjDtkszMzFLn0JClNmq59slrWbB2AVOGTeG4/Y9LuyQzM7Oi4HsaslRWVbJqwyquOPIKTv/I6WmXY2ZmVjQ805CltE0p9596P21atkm7FDMzs6LimYZ6ODCYmZl9kEODmZmZ5cShwczMzHLi0GBmZmY5KYrQIGmMpOWSXpF0dT3b20j632T7Akl9Cl+lmZlZ85Z6aJDUEpgOnAIcCnxG0qFZu40D/hURfYE7gG8XtkozMzNLPTQAQ4BXImJlRFQBPwPOyNrnDODeZPkXwImSVMAazczMmr1iCA09gNfqrK9OxurdJyKqgQpg74JUZ2ZmZsAe+OFOki4ALkhWN0paniyXkgkb5Li+D7C+kcrMvtbuPGZH+21vW33jOxtzv/Ibc7/yH6u77n4Vrl+9G3icNQcRkeoXcDTwhzrr1wDXZO3zB+DoZLmEzF8G5Xmdu/Ncf64RX/PdjXXMjvbb3rb6xnc25n65X43Zr3r6534VSb/81by/iuHtiWeBfpIOkNQaOBeYlbXPLOALyfIngccjIvK8zm/zXG9MDblWrsfsaL/tbatvfGdj7ld+Y+5X/mOF6pn7ZZYj5f+7txGKkE4F7gRaAvdExBRJN5BJy7MktQXuAwYBbwPnRsTKRq7puYgY3JjX2JO4X/lxv/LjfuXH/bLGUhT3NETEbGB21tj1dZY3Af9Z4LLuLvD1mjr3Kz/uV37cr/y4X9YoimKmwczMzIpfMdzTYGZmZk2AQ4OZmZnlxKHBzMzMcuLQkANJB0r6oaRfpF1LUyHpTEkzkgeNnZR2PcVOUn9JZZJ+IWlC2vU0BZI6SHpO0mlp11LsJI2QNC/5HhuRdj3WdDXb0CDpHklvSlqaNf6BJ25G5rkY49KptHjk2bNfR8SXgfHAp9OoN2159qs8IsYDnwKOTaPetOXTr8RVwIOFrbJ45NmvADYCbcl8VL9ZgzTb0ADMBMbUHcjxiZvN2Uzy79l1yfbmaCZ59EvS6cAjZP3342ZkJjn2S9JoYBnwZqGLLCIzyf37a15EnEImaH2rwHXaHqTZhoaIeILMB0XVlcsTN5utfHqmjG8Dv4uIhYWutRjk+z0WEbOSH+xjC1tpccizXyOAocBngS9LanY/y/LpV0TUJtv/BbQpYJm2hymKD3cqIvU9cfMoSXsDU4BBkq6JiP9OpbriVG/PgEuBUUCppL4RUZZGcUVoe99jI4CzyfxAb64zDfWpt18RcQmApC8C6+v8Umzutvf9dTZwMvAhYFoahdmewaEhBxHxFpn35i1HETEVmJp2HU1FRMwF5qZcRpMTETPTrqEpiIiHgIfSrsOavmY3pbcTa4Ceddb3T8Zs+9yz/Lhf+XG/8uN+WaNyaHi/XJ64ae/nnuXH/cqP+5Uf98saVbMNDZJ+CvwFOFjSaknjIqIauAT4A1AOPBgRL6VZZzFxz/LjfuXH/cqP+2Vp8AOrzMzMLCfNdqbBzMzM8uPQYGZmZjlxaDAzM7OcODSYmZlZThwazMzMLCcODWZmZpYThwYzQNJcSYMLcJ3LJJVL+knW+GBJU5PlEZKO2Y3X7CPps/Vdy8wsH372hNkuklSSfKhOLi4CRkXE6rqDEfEc8FyyOgLYCMzfTTX0IfM0yAfquZaZWc4802BNRvIv5nJJMyS9JOmPktol27bNFEjaR9KryfIXJf1a0qOSXpV0iaSvSHpB0tOSutS5xHmSFklaKmlIcnwHSfdIeiY55ow6550l6XHgsXpq/UpynqWSJiZjZcCBwO8kTcraf4SkhyX1IfNwtElJLcdJ6irpl5KeTb6OTY75pqT7JD0F3Jf0Z56khcnX1tmKm4HjkvNN2nqt5Bxdkv68mPRjYJ1z35P0daWky+r04xFJi5PX9uld+1M1s6bEMw3W1PQDPhMRX5b0IHAOcP9OjhkADALaAq8AV0XEIEl3AJ8H7kz2ax8Rh0s6HrgnOe7rwOMR8SVJHwKekTQn2f8IYGBEvF33YpKOBM4n84hwAQsk/TkixksaA4yMiPX1FRoRrybhYmNE3Jac7wHgjoh4UlIvMh8R3D855FBgWES8J6k9MDoiNknqB/wUGAxcDVwZEacl5xtR55LfAl6IiDMlnQD8GDg82XYIMBLoBCyX9D1gDPB6RHwiOVfpTnpvZnsQhwZrav4eEYuS5efJTL3vzJ8iohKolFQB/DYZXwIMrLPfTwEi4glJnZOQcBJwuqQrk33aAr2S5UezA0NiGPCriHgHQNJDwHHAC7m8wHqMAg6VtHW9s6SOyfKsiHgvWW4FTJN0OFADHJTDuYeRCV5ExOOS9pbUOdn2SERsBjZLehPYl0zPbpf0beDhiJjXwNdkZk2QQ4M1NZvrLNcA7ZLlav79dlvbHRxTW2e9lvf/Hch+EEuQmSk4JyKW190g6Sjgnbwqb7gWwNCI2JRVA1k1TALeAD6WHPO+/Rsgu9clEbFC0hHAqcBNkh6LiBt28Tpm1kT4ngbbU7wKHJksf7KB5/g0gKRhQEVEVJB5K+BSJb+hJQ3K4TzzgDMltZfUATgrGctVJZm3BLb6I3Dp1pVkJqE+pcDaiKgFzgNabud82bWOTc47AlgfERu2V5ik7sC7EXE/cCuZt2jMrJlwaLA9xW3ABEkvAPs08BybkuPLgHHJ2I1kpv1flPRSsr5DEbEQmAk8AywAfhAR+bw18VvgrK03QgKXAYOTmxWXkblRsj53AV+QtJjM/QhbZyFeBGqSmxcnZR3zTeBISS+SuWHyCzup7TAy93UsAiYDN+XxusysifOjsc3MzCwnnmkwMzOznDg0mJmZWU4cGszMzCwnDg1mZmaWE4cGMzMzy4lDg5mZmeXEocHMzMxy4tBgZmZmOfl/4jAKNKxO3YcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Collapse multiple runs into one\n",
    "plot_results = []\n",
    "for num_iters in iter_sweep:\n",
    "    runs = [tn_test_multiple_runs[(num_iters, j)] for j in range(num_sims)]\n",
    "    p_tn, likelihood = sorted(runs, key = lambda x:x[1])[-1]\n",
    "    plot_results.append(p_tn)\n",
    "    \n",
    "# Collapse multiple runs into one\n",
    "plot_results = []\n",
    "for num_iters in iter_sweep:\n",
    "    runs = [tn_test_multiple_runs[(num_iters, j)] for j in range(num_sims) \n",
    "            if not np.isinf(tn_test_multiple_runs[(num_iters, j)][1])]\n",
    "    p_tn, likelihood = sorted(runs, key = lambda x:x[1])[-1]\n",
    "    plot_results.append(p_tn)\n",
    "\n",
    "for i in range(5):\n",
    "    curve = [-np.log10(result[i]) for result in plot_results]\n",
    "    plt.plot(iter_sweep[2:], curve[2:], label=genes_list[i])\n",
    "plt.xscale('log')\n",
    "plt.xlabel('number of iterations')\n",
    "plt.ylabel(r'$-\\log_{10}(p)$')\n",
    "plt.legend(bbox_to_anchor=(1.4, 1))\n",
    "plt.savefig('figures/pbmc_%sv%s_num_iters.pdf'%(ident0, ident1), format='pdf', dpi=500, bbox_inches='tight')\n",
    "plt.show()    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Testing other clusters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_and_split_data(ident0, ident1, save=True):\n",
    "    cellsamp1 = pd.read_csv(os.path.join(data_dir, 'pbmc_seurat_%sv%s_cellsamp1.txt'%(ident0, ident1)), delimiter=' ')\n",
    "    pdict = {j:i for i, j in enumerate(data_R.columns)}\n",
    "    cellsamp_inds = np.array([pdict[i] for i in np.array(cellsamp1).reshape(-1)])\n",
    "    cellsamp1labels = np.array(pd.read_csv(os.path.join(data_dir, 'pbmc_seurat_%sv%s_cellsamplabels1.txt'%(ident0, ident1)), delimiter=' ')).reshape(-1)\n",
    "\n",
    "    # get hyperplane using first half of dataset\n",
    "    svm = SVC(kernel='linear', C=100)\n",
    "    svm.fit(data[cellsamp_inds], cellsamp1labels)\n",
    "    a = svm.coef_.reshape(-1)\n",
    "    b = svm.intercept_[0]\n",
    "    labels1hat = svm.predict(data[cellsamp_inds])\n",
    "    print('Consistency of new labels with old: %.3f'\\\n",
    "          %(np.sum(labels1hat == cellsamp1labels)/float(len(cellsamp1labels))))\n",
    "\n",
    "    # assign points in second half of dataset\n",
    "    inds = np.ones(len(data)).astype(bool)\n",
    "    inds[cellsamp_inds] = 0\n",
    "    for i in np.unique(labels):\n",
    "        if i != ident0 and i != ident1:\n",
    "            inds[labels == i] = 0\n",
    "    labels2 = svm.predict(data[inds])\n",
    "    y, z = data[inds][labels2 == 0], data[inds][labels2 == 1]\n",
    "\n",
    "    if save:\n",
    "        np.savetxt('pbmc_seurat_%sv%s_cellsamp2.txt'%(ident0, ident1), data_R.columns[inds], fmt='%s')\n",
    "        np.savetxt('pbmc_seurat_%sv%s_cellsamplabels2.txt'%(ident0, ident1), labels2, fmt='%s')\n",
    "    \n",
    "    return y, z, a, b\n",
    "\n",
    "\n",
    "def compare_clusters_multiple_runs_less_genes(ident0, ident1, num_runs=10):\n",
    "    \"\"\"Set up and run TN test data splitting framework\n",
    "        run TN test multiple times to account for variance in optimization\n",
    "        use less genes to save computation\n",
    "    \"\"\"\n",
    "    \n",
    "    picklefile = 'pbmc_seurat_%sv%s_de_tntest_multiple_runs_max_withseed.pickle'%(ident0, ident1)\n",
    "    if os.path.isfile(picklefile):\n",
    "        tn_test_multiple_runs = pickle.load(open(picklefile, 'rb'))\n",
    "        p_tn, likelihood, seed = sorted(tn_test_multiple_runs, key = lambda x:x[1])[-1]\n",
    "        return p_tn\n",
    "    \n",
    "    y, z, a, b = load_and_split_data(ident0, ident1)\n",
    "\n",
    "    # Run TN test on genes that actually change across the two clusters\n",
    "    keep_inds = np.var(np.vstack((y, z)), 0) > 0\n",
    "    np.savetxt('pbmc_seurat_%sv%s_filtered_genes.txt'%(ident0, ident1), keep_inds)\n",
    "    y = y[:, keep_inds]\n",
    "    z = z[:, keep_inds]\n",
    "    a = a[keep_inds]\n",
    "\n",
    "    genes_list = grab_interesting_de_genes(ident0, ident1)\n",
    "    genes_to_test = [np.where(genes[keep_inds] == gene)[0][0] for gene in genes_list]\n",
    "\n",
    "    # Run TN test multiple times to evaluate stability\n",
    "    tn_test_multiple_runs = []\n",
    "    num_runs = 10\n",
    "    for _ in range(num_runs):\n",
    "        seed = np.random.get_state()\n",
    "        p_tn, likelihood = tn.tn_test(y, z, a=a, b=b, verbose=True, learning_rate=0.5,\n",
    "                                      eps=1e-1, genes_to_test=genes_to_test,\n",
    "                                      return_likelihood=True)\n",
    "        tn_test_multiple_runs.append((p_tn, likelihood[-1], seed))\n",
    "    pickle.dump(tn_test_multiple_runs, open(picklefile, 'wb'))\n",
    "    \n",
    "    p_tn, likelihood, seed = sorted(tn_test_multiple_runs, key = lambda x:x[1])[-1]\n",
    "    return p_tn\n",
    "\n",
    "\n",
    "def compare_clusters_tn_test(ident0, ident1):\n",
    "    \"\"\"Set up and run TN test data splitting framework\"\"\"\n",
    "\n",
    "    picklefile = 'pbmc_seurat_%sv%s_de_tntest_all_genes.pickle'%(ident0, ident1)\n",
    "    if os.path.isfile(picklefile):\n",
    "        return pickle.load(open(picklefile, 'rb'))\n",
    "\n",
    "    y, z, a, b = load_and_split_data(ident0, ident1)\n",
    "\n",
    "    # Run TN test on genes that actually change across the two clusters\n",
    "    keep_inds = np.var(np.vstack((y, z)), 0) > 0\n",
    "    np.savetxt('pbmc_seurat_%sv%s_filtered_genes.txt'%(ident0, ident1), keep_inds)\n",
    "    y = y[:, keep_inds]\n",
    "    z = z[:, keep_inds]\n",
    "    a = a[keep_inds]\n",
    "    \n",
    "    # Get seed based on previous experiment\n",
    "    seedfile = 'pbmc_seurat_%sv%s_de_tntest_multiple_runs_max_withseed.pickle'%(ident0, ident1)\n",
    "    tn_test_multiple_runs = pickle.load(open(seedfile, 'rb'))\n",
    "    _, _, seed = sorted(tn_test_multiple_runs, key = lambda x:x[1])[-1]\n",
    "    np.random.set_state(seed)\n",
    "\n",
    "    # Run TN test multiple times to evaluate stability\n",
    "    p_tn, likelihood = tn.tn_test(y, z, a=a, b=b, verbose=True, learning_rate=0.5,\n",
    "                                  eps=1e-1, return_likelihood=True, num_iters=10000, num_cores=64)\n",
    "    pickle.dump(p_tn, open(picklefile, 'wb'))\n",
    "\n",
    "    return p_tn, likelihood[-1]\n",
    "\n",
    "\n",
    "def plot_bar_p(gene_names, p_correct, p_seurat=None, label=None, width=0.35,\n",
    "               set_ylabel=True, p_tn_se=None, mark_red=[]):\n",
    "    fig, ax = plt.subplots(figsize=(3, 1.5))\n",
    "    ind = np.arange(len(p_correct))\n",
    "    if p_seurat is not None:\n",
    "        rects = ax.bar(ind-0.2,\n",
    "                       p_seurat, width,\n",
    "                       align='center',\n",
    "                       label=label)\n",
    "        rects = ax.bar(ind-0.2+0.3,\n",
    "                       p_correct, width,\n",
    "                       yerr=p_tn_se,\n",
    "                       align='center',\n",
    "                       label='TN test')\n",
    "    else:\n",
    "        rects = ax.bar(ind,\n",
    "                       p_correct, width,\n",
    "                       yerr=p_tn_se,\n",
    "                       align='center',\n",
    "                       label='TN test',\n",
    "                       color='#ff7f0e')\n",
    "    if set_ylabel: ax.set_ylabel('-log($p$)')\n",
    "    xticks_pos = [0.65*patch.get_width()+patch.get_xy()[0]-0.1 for patch in rects]\n",
    "    plt.xticks(np.array(xticks_pos), gene_names, rotation=90)\n",
    "    for ticklabel in plt.gca().get_xticklabels():\n",
    "        if ticklabel.get_text() in mark_red:\n",
    "            ticklabel.set_color('r')\n",
    "    plt.legend()     \n",
    "\n",
    "        \n",
    "def get_de_df(ident0, ident1):\n",
    "    genes_list = grab_interesting_de_genes(ident0, ident1)\n",
    "    df_dict = {'TN': list(-np.log10(np.max(tn_test_multiple_runs, axis=0)))}\n",
    "    for i, test in enumerate(tests):\n",
    "        seurat_de = -np.log10(pd.read_csv(os.path.join(data_dir, 'pbmc_seurat_%sv%s_de_%s.txt'%(ident0, ident1, test)), delimiter=' '))\n",
    "        df_dict[test] = list(seurat_de.loc[genes_list]['p_val'])\n",
    "    de_df = pd.DataFrame.from_dict(df_dict).set_index(genes_list).round(2)\n",
    "    interesting_genes = genes_list[np.argmin(np.array(de_df), 1) == 0]\n",
    "    return de_df, interesting_genes\n",
    "\n",
    "\n",
    "def view_genes(gene_names, ident0, ident1, marked_genes=None):\n",
    "    \n",
    "    labels1 = np.array(pd.read_csv('./pbmc_seurat_%sv%s_cellsamplabels1.txt'%(ident0, ident1), delimiter=' ')).reshape(-1)\n",
    "    labels2 = np.loadtxt('./pbmc_seurat_%sv%s_cellsamplabels2.txt'%(ident0, ident1))\n",
    "\n",
    "    cellsamp1 = pd.read_csv(os.path.join(data_dir, 'pbmc_seurat_%sv%s_cellsamp1.txt'%(ident0, ident1)), delimiter=' ')\n",
    "    pdict = {j:i for i, j in enumerate(data_R.columns)}\n",
    "    inds1 = np.array([pdict[i] for i in np.array(cellsamp1).reshape(-1)])\n",
    "    cellsamp2 = np.loadtxt('./pbmc_seurat_%sv%s_cellsamp2.txt'%(ident0, ident1), dtype='str')\n",
    "    inds2 = np.array([pdict[i] for i in np.array(cellsamp2).reshape(-1)])\n",
    "\n",
    "    data1 = data[inds1]\n",
    "    data2 = data[inds2]\n",
    "\n",
    "    y1, z1 = data1[labels1 == 0], data1[labels1 == 1]\n",
    "    y2, z2 = data2[labels2 == 0], data2[labels2 == 1]\n",
    "    \n",
    "    for i, gene_name in enumerate(gene_names):\n",
    "        if i%4 == 0:\n",
    "            if i > 0:\n",
    "                plt.tight_layout()\n",
    "                plt.show()\n",
    "            plt.figure(figsize=(2*4, 1.8))\n",
    "        plt.subplot(1, 4, i%4+1)\n",
    "        g_ind = np.where(genes == gene_name)[0][0]\n",
    "        utils.plot_stacked_hist(np.vstack((y1, y2))[:, g_ind], np.vstack((z1, z2))[:, g_ind])\n",
    "        plt.title(gene_name+('*' if gene_name in marked_genes else ''))\n",
    "        plt.axis('off')\n",
    "        print('%10s | type %s with mean %.2f'%(gene_name, ident0, np.mean(data[labels == ident0, g_ind])))\n",
    "        print('%10s | type %s with mean %.2f'%(gene_name, ident1, np.mean(data[labels == ident1, g_ind])))\n",
    "\n",
    "    plt.tight_layout()\n",
    "    \n",
    "    \n",
    "def create_and_save_visualization(p_tn, ident0, ident1, genes_to_view, genes_to_mark_red=[], ngenes=5):\n",
    "    plt.savefig('figures/pbmc_%sv%s_multiple_runs_gene_exp.pdf'\\\n",
    "                %(ident0, ident1), format='pdf', dpi=500, bbox_inches='tight')\n",
    "    \n",
    "    \n",
    "    filename = 'pbmc_seurat_%sv%s_filtered_genes.txt'%(ident0, ident1)\n",
    "    genes_tn = np.array(data_R.index)[np.loadtxt(filename).astype(bool)]\n",
    "    \n",
    "    if genes_to_view is None:\n",
    "        genes_to_view = grab_interesting_de_genes(ident0, ident1)\n",
    "    \n",
    "    view_genes(genes_to_view, ident0, ident1, marked_genes=[])\n",
    "    plt.savefig('figures/pbmc_%sv%s_gene_exp.pdf'%(ident0, ident1), format='pdf', dpi=500, bbox_inches='tight')\n",
    "\n",
    "    for i, test in enumerate(tests):\n",
    "        seurat_de = -np.log10(pd.read_csv(os.path.join(data_dir, 'pbmc_seurat_%sv%s_de_%s.txt' \\\n",
    "                                                       %(ident0, ident1, test)), delimiter=' '))\n",
    "        del seurat_de['avg_logFC']\n",
    "        del seurat_de['pct.1']\n",
    "        del seurat_de['pct.2']\n",
    "        del seurat_de['p_val_adj']\n",
    "        gene_names = seurat_de.index[:ngenes]\n",
    "        p_seurat = seurat_de['p_val'][:ngenes]\n",
    "        p_correct = [-np.log10(p_tn)[genes_tn == gene][0] for gene in gene_names][:ngenes]\n",
    "        if test == \"t\": \n",
    "            test = 'Welch\\'s $t$'\n",
    "        plot_bar_p(gene_names, p_correct, p_seurat=p_seurat, label=test,\n",
    "                   set_ylabel = True if i == 0 or i == 4 else False, mark_red=genes_to_mark_red)\n",
    "        plt.savefig('figures/pbmc_de_%sv%s_%s.pdf'%(ident0, ident1, test), format='pdf', dpi=500, bbox_inches='tight')\n",
    "        \n",
    "        print(test)\n",
    "        for j in range(ngenes):\n",
    "            print('%10s\\tpDE: %.2f\\tpTN: %.2f'%(gene_names[j], p_seurat[j],  p_correct[j]))\n",
    "        print(' ')\n",
    "\n",
    "    inds = np.argsort(p_tn)[:ngenes]\n",
    "    p_correct = -np.log10(p_tn)[inds]\n",
    "    gene_names = genes_tn[inds]\n",
    "    plot_bar_p(gene_names, p_correct, set_ylabel=False)\n",
    "    plt.savefig('figures/pbmc_de_%sv%s_tn.pdf'%(ident0, ident1), format='pdf', dpi=500, bbox_inches='tight')\n",
    "    plt.show()\n",
    "    \n",
    "    \n",
    "def tsne_select2(ident0, ident1):\n",
    "    labels_ = np.array(labels).astype(str)\n",
    "    labels_[~((labels == ident0) | (labels == ident1))] = 'Other'\n",
    "    plt.scatter(tsne[labels_ == str(ident0), 0], tsne[labels_ == str(ident0), 1],\n",
    "                label=r'c%s ($n$ = %s)'%(ident0, np.sum(labels_== str(ident0))), c='g', s=10)\n",
    "    plt.scatter(tsne[labels_ == str(ident1), 0], tsne[labels_ == str(ident1), 1],\n",
    "                label=r'c%s ($n$ = %s)'%(ident1, np.sum(labels_== str(ident1))), c='r', s=10)\n",
    "    plt.scatter(tsne[labels_ == 'Other', 0], tsne[labels_ == 'Other', 1],\n",
    "                label=r'Other ($n$ = %s)'%(np.sum(labels_=='Other')), c='k', s=10)\n",
    "    plt.legend(bbox_to_anchor=(1.4, 1))\n",
    "    plt.axis('equal')\n",
    "    plt.axis('off')\n",
    "    plt.savefig('figures/pbmc_tsne_%sv%s.pdf'%(ident0, ident1),\n",
    "                    format='pdf', dpi=500, bbox_inches='tight')\n",
    "    plt.show()    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    S100A4 | type 0 with mean 2.63\n",
      "    S100A4 | type 1 with mean 1.73\n",
      "   S100A11 | type 0 with mean 1.11\n",
      "   S100A11 | type 1 with mean 0.55\n",
      "       B2M | type 0 with mean 5.33\n",
      "       B2M | type 1 with mean 5.06\n",
      "     HLA-A | type 0 with mean 3.38\n",
      "     HLA-A | type 1 with mean 3.07\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.6/site-packages/ipykernel_launcher.py:189: RuntimeWarning: divide by zero encountered in log10\n",
      "/opt/conda/lib/python3.6/site-packages/ipykernel_launcher.py:189: RuntimeWarning: invalid value encountered in log10\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Welch's $t$\n",
      "    S100A4\tpDE: 43.37\tpTN: 70.27\n",
      "       B2M\tpDE: 24.36\tpTN: 12.43\n",
      "      IL32\tpDE: 23.39\tpTN: 21.98\n",
      "     CLIC1\tpDE: 16.21\tpTN: 17.24\n",
      "    LGALS1\tpDE: 15.68\tpTN: 23.11\n",
      "  SH3BGRL3\tpDE: 15.12\tpTN: 23.56\n",
      "     ANXA2\tpDE: 14.12\tpTN: 16.73\n",
      "   S100A11\tpDE: 13.61\tpTN: 26.50\n",
      "      CCR7\tpDE: 13.32\tpTN: 15.54\n",
      "   S100A10\tpDE: 13.17\tpTN: 22.38\n",
      " \n",
      "wilcox\n",
      "    S100A4\tpDE: 40.84\tpTN: 70.27\n",
      "       B2M\tpDE: 29.07\tpTN: 12.43\n",
      "      IL32\tpDE: 22.11\tpTN: 21.98\n",
      "  SH3BGRL3\tpDE: 17.16\tpTN: 23.56\n",
      "     HLA-A\tpDE: 16.42\tpTN: 7.14\n",
      "     CLIC1\tpDE: 15.98\tpTN: 17.24\n",
      "    LGALS1\tpDE: 15.29\tpTN: 23.11\n",
      "     ANXA2\tpDE: 13.94\tpTN: 16.73\n",
      "   S100A11\tpDE: 13.75\tpTN: 26.50\n",
      "     RPS3A\tpDE: 13.42\tpTN: 14.30\n",
      " \n",
      "bimod\n",
      "    S100A4\tpDE: 44.58\tpTN: 70.27\n",
      "       B2M\tpDE: 22.48\tpTN: 12.43\n",
      "      IL32\tpDE: 22.19\tpTN: 21.98\n",
      "    LGALS1\tpDE: 20.50\tpTN: 23.11\n",
      "  SH3BGRL3\tpDE: 15.65\tpTN: 23.56\n",
      "     HLA-A\tpDE: 15.16\tpTN: 7.14\n",
      "     CLIC1\tpDE: 14.59\tpTN: 17.24\n",
      "   S100A11\tpDE: 14.01\tpTN: 26.50\n",
      "     ANXA2\tpDE: 13.78\tpTN: 16.73\n",
      "      CCR7\tpDE: 13.04\tpTN: 15.54\n",
      " \n",
      "tobit\n",
      "    S100A4\tpDE: 44.35\tpTN: 70.27\n",
      "       B2M\tpDE: 24.34\tpTN: 12.43\n",
      "      IL32\tpDE: 23.43\tpTN: 21.98\n",
      "    LGALS1\tpDE: 16.80\tpTN: 23.11\n",
      "  SH3BGRL3\tpDE: 16.14\tpTN: 23.56\n",
      "     CLIC1\tpDE: 16.00\tpTN: 17.24\n",
      "     ANXA2\tpDE: 15.05\tpTN: 16.73\n",
      "   S100A11\tpDE: 13.99\tpTN: 26.50\n",
      "     HLA-A\tpDE: 13.86\tpTN: 7.14\n",
      "      CCR7\tpDE: 13.30\tpTN: 15.54\n",
      " \n",
      "poisson\n",
      "       B2M\tpDE: 135.02\tpTN: 12.43\n",
      "    S100A4\tpDE: 116.01\tpTN: 70.27\n",
      "       LTB\tpDE: 65.09\tpTN: 7.37\n",
      "      IL32\tpDE: 52.52\tpTN: 21.98\n",
      "     RPS3A\tpDE: 38.07\tpTN: 14.30\n",
      "     HLA-A\tpDE: 37.28\tpTN: 7.14\n",
      "      ACTB\tpDE: 32.48\tpTN: 13.25\n",
      "  SH3BGRL3\tpDE: 32.01\tpTN: 23.56\n",
      "     HLA-C\tpDE: 31.92\tpTN: 8.49\n",
      "     ISG15\tpDE: 27.34\tpTN: 2.70\n",
      " \n",
      "negbinom\n",
      "    S100A4\tpDE: 42.22\tpTN: 70.27\n",
      "      IL32\tpDE: 20.50\tpTN: 21.98\n",
      "    LGALS1\tpDE: 16.99\tpTN: 23.11\n",
      "       B2M\tpDE: 16.88\tpTN: 12.43\n",
      "     HLA-A\tpDE: 16.38\tpTN: 7.14\n",
      "   S100A11\tpDE: 14.99\tpTN: 26.50\n",
      "       LTB\tpDE: 13.93\tpTN: 7.37\n",
      "     ANXA2\tpDE: 13.75\tpTN: 16.73\n",
      "  SH3BGRL3\tpDE: 12.73\tpTN: 23.56\n",
      "     HLA-C\tpDE: 12.65\tpTN: 8.49\n",
      " \n",
      "MAST\n",
      "    S100A4\tpDE: 45.11\tpTN: 70.27\n",
      "       B2M\tpDE: 22.67\tpTN: 12.43\n",
      "      IL32\tpDE: 22.52\tpTN: 21.98\n",
      "    LGALS1\tpDE: 16.27\tpTN: 23.11\n",
      "  SH3BGRL3\tpDE: 15.88\tpTN: 23.56\n",
      "     HLA-A\tpDE: 15.73\tpTN: 7.14\n",
      "     CLIC1\tpDE: 15.22\tpTN: 17.24\n",
      "     ANXA2\tpDE: 13.75\tpTN: 16.73\n",
      "   S100A11\tpDE: 13.41\tpTN: 26.50\n",
      "      CCR7\tpDE: 12.43\tpTN: 15.54\n",
      " \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x129.6 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAMwAAACfCAYAAAC1ODarAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAGkdJREFUeJztnXt4VcW1wH8r4REqjwhaSo00UFGRAkEIpTxU5BZBfPFSwIpaW4wKRdBPqddWaNXiFQsS+aS0iOgV8UUvraK2KigWEQkGkYdKJEAsIqQiUKSFZt0/Zic5OTk5e+/z3Anz+7795ezZe82s7HPWnpk1a2ZEVbFYLN7ISLcCFkt9whqMxeIDazAWiw+swVgsPrAGY7H4wBqMxeIDazAWiw+swVgsPrAGY7H4oFEqCzvllFM0Nzc3lUVaLJ4oKirar6qnut2XUoPJzc1l/fr1qSzSYvGEiOz0cp9tklksPrAGY7H4IKVNMt9Mb1VH+lep1cNicQi2wZzgHDt2jLKyMo4ePZpuVRoMWVlZ5OTk0Lhx45jkrcEEmLKyMlq0aEFubi4ikm516j2qSnl5OWVlZXTo0CGmPGwfJsAcPXqUNm3aWGNJECJCmzZt4qqxrcEEHGssiSXe52kNxmLxge3D1CNyp72U0PxKZw5LaH7RWLVqFbNmzeLFF1+sde3iiy9myZIlZGdnp0yfWLEGY0k7K1asSLcKnvHUJBORbBF5XkS2ichWEfmBiLQWkb+KyCfO35OTrawl9ZSWltK5c2d++tOf0qVLFwYPHszXX39NSUkJQ4YMoWfPngwYMIBt27YBUFJSQp8+fejatSt33303zZs3r8rr4MGDDBs2jLPOOouCggIqKioAEzK1f//+OssCKC4upk+fPnTr1o3hw4fz5ZdfAnDBBRcwZcoUevXqRefOnXnvvfcYMWIEnTp14u6770748/Dah3kYeEVVzwa6A1uBacDrqtoJeN05tzRAPvnkE2655RY2b95MdnY2L7zwAhMmTKCwsJCioiJmzZrFzTffDMDkyZOZPHkymzZtIicnp0Y+69ato7CwkC1btlBSUsKyZcs8lQUwfvx4HnjgAT744AO6du3KjBkzqmSaNGnC+vXrKSgo4PLLL2fevHl8+OGHPP7445SXlyf0WbgajIi0As4DFgKo6r9V9QBwObDYuW0xcEVCNbMEhg4dOpCXlwdAz549KS0tZc2aNYwePZq8vDxuvPFG9uzZA8A777zD6NGjARg3blyNfHr37k3Hjh3JzMxk7NixvP32257K+uqrrzhw4ADnn38+ANdeey1vvfVWlcxll10GQNeuXenSpQvt2rWjadOmdOzYkd27dyf0WXjpw3QA9gGLRKQ7UARMBtqq6h7nns+BtpGERWQCMAGgffv2cStsST1Nmzat+pyZmcnevXvJzs6muLjYVz7hLt1ILt7wsiqbZF70y8jIqCGfkZHB8ePHfenohpcmWSPgXOBRVe0B/JOw5pea5TMjLqGpqgtUtZeq9jr1VNfpBpZ6QMuWLenQoQPPPfccYEbQN27cCECfPn2qmlFLly6tIbdu3Tp27NhBRUUFzzzzDP379/dUXqtWrTj55JNZvXo1AE8++WRVbZNqvNQwZUCZqr7rnD+PMZi9ItJOVfeISDvgi2QpaTGk0g3sxlNPPcVNN93Evffey7FjxxgzZgzdu3dnzpw5/OhHP+K+++5jyJAhtGpVHUCbn5/PxIkT2b59OwMHDmT48OGey1u8eDEFBQUcOXKEjh07smjRomT8W66Il7WVRWQ18BNV/UhEpgMnOZfKVXWmiEwDWqvqHdHy6dWrl/qaQHaCRytv3bqVzp07p1sNXxw5coRmzZohIixdupSnn36a5cuXp1utGkR6riJSpKq93GS9jsNMAp4SkSbAp8D1mObcsyJyA7ATuNKX1pYGSVFRERMnTkRVyc7O5rHHHku3SgnFk8GoajEQyfoGJVYdS31nwIABVf2ZhoiNJbNYfGANxmLxgTUYi8UH1mAsFh/YaOX6RF1u9pjzi+6eLy8vZ9Ag49f5/PPPyczMpHLweePGjUydOpWHHnoIgFmzZnH48GGmT59eI49Vq1bRpEkT+vbt61u9yhCc8BCbdGJrGEudtGnThuLiYoqLiykoKGDKlClV502bNmXZsmXs378/ah6rVq1izZo1MZVfWlrKkiVLYpJNFtZgLDHRqFEjJkyYwOzZs+u8p7S0lPnz5zN79mzy8vJYvXo1+/btY+TIkeTn55Ofn8/f/vY3AN58803y8vLIy8ujR48eHDp0iGnTprF69Wry8vKilpNKbJPMEjO33HIL3bp14447Igd45ObmUlBQQPPmzbn99tsBE8E8ZcoU+vfvz65du7jooovYunUrs2bNYt68efTr14/Dhw+TlZXFzJkz65ylmS6swVhipmXLlowfP565c+fSrFkzTzKvvfYaW7ZsqTo/ePAghw8fpl+/fkydOpWrr76aESNG1JpLExSswVji4tZbb+Xcc8/l+uuv93R/RUUFa9euJSsrq0b6tGnTGDZsGCtWrKBfv368+uqryVA3bmwfxhIXrVu35sorr2ThwoURr7do0YJDhw5VnQ8ePJjCwsKq88o5NSUlJXTt2pU777yT/Px8tm3bVks2CNgapj4R0Cjt2267jUceeSTitUsvvZRRo0axfPlyCgsLmTt3blXf5/jx45x33nnMnz+fOXPmsHLlSjIyMujSpQtDhw4lIyODzMxMunfvznXXXceUKVNS/J/VxlN4f6Kw4f3+qI/h/fWBeML7PTfJRCRTRN4XkRed8w4i8q6IbBeRZ5zQf4ulQeOnSTYZs1pMS+f8AWC2qi4VkfnADcCjsSoSaZG60qwIN1osacTrumQ5wDDgD865ABdipiuDXTUmaaSyyXwiEO/z9NokmwPcAVQ4522AA6pauSRHGXBaJEERmSAi60Vk/b59++JS9kQjKyuL8vJyazQJonK7i3CXth9cm2QicgnwhaoWicgFfgtQ1QXAAjCdft8ansDk5ORQVlaGfdEkjsoNlWLFSx+mH3CZiFwMZGH6MA8D2SLSyKllcoDPYtbCEpHGjRvHvPGPJTm4NslU9eeqmqOqucAY4A1VvRpYCYxybrsWCNbSIBZLEohnpP9OYKqIbMf0aSIP9VosDQhfI/2qugpY5Xz+FOideJUsluBiY8ksFh9Yg7FYfGANxmLxgTUYi8UH1mAsFh9Yg7FYfGANxmLxgTUYi8UH1mAsFh/YOf0NlRN8eneysDWMxeIDazAWiw+swVgsPvAy4/J04AmgLaDAAlV9WERaA88AuUApcKWqfpk8VdNApH6A7QMknwD3v7x0+o8Dt6nqBhFpARSJyF+B64DXQ7Ydn4aZIxMc7A/ekmC8zLjco6obnM+HMEstnQZcjlktBuyqMZYTBF9uZRHJBXoA7wJtVXWPc+lzTJMtkswEYAJA+/btY9XTFbuumSUV+Fn5sjnwAnCrqh4MvaZmHaCIK8Ko6gJV7aWqvSq3e7NY6iteF/JrjDGWp1R1mZO8V0TaOdfbAV8kR0WLJTi4GoyzyuVCYKuq/jbk0p8wq8WAXTXGcoLgdV2ya4BNIlLspN0FzASeFZEbgJ3AlclRMTXE2weKKD9zWMrkA0cD9VC6Goyqvg1IHZcHJVYdiyXY2ODLBoD1EKYOazCWiAYH3puEJ5LBWoOxpJ1wgwuysVmDSRKlWeNgeoQLDaDjGyTirR39YqOVLRYfWIOxWHxgDcZi8YHtw1gaHMnsP1qDsUQk4o/OOiyswQQV62ULJrYPY7H4wBqMxeIDazAWiw/iMhgRGSIiH4nIdmchDIulQROzwYhIJjAPGAqcA4wVkXMSpZjFEkTiqWF6A9tV9VNV/TewFLOSjMXSYInHYE4DdoeclzlpFkuDRcyCLzEIiowChqjqT5zza4Dvq+rEsPuqllkCzgI+iqG4U4D9MSmafvn6rHu65VNZ9ndU1XVZo3gGLj8DTg85z3HSaqCqC4AFcZSDiKxX1V71Ub4+655u+XTrHol4mmTvAZ1EpIOINAHGYFaSsVgaLDHXMKp6XEQmAq8CmcBjqro5YZpZLAEkrlgyVV0BrEiQLtGIq0mXZvn6rHu65dOtey1i7vRbLCciNjTGYvGBNRiLxQfWYCwWH9QLgxGRJ9KtA4CI/DCNhUfcf6chISK9RSTf+XyOiEwVkYvTrVcogev0i0j4WI4AA4E3AFT1Mhf5rsDvMWE6LwN3Vu69KSLrVLV3HLrtUtW6d4USOUTNfXLEOTd/VVv6LDAbGAmMAzqj+m2/OldnJS+r6lAP9/UDDqjqZhEZAZwHlAC/c2IGo8lehBnAfl1VS0PSf6yqj7nI3oMJ5G0E/BX4PrAS+CHwqqre5yI/FfhKVReGpd8AtFDVOdHkvRJEg9kAbAH+QPWP7WnMwCiq+qaL/NvAvcBa4CfA9cBlqloiIu+rag8X+boGXwW4UFVPiiL8f8C3gGXAUlR3RSurjjyaYYJYx2F2e2uB2Q7xLVQrXHQ/N4ruL6pqOxf5QqAX0ATYhAkteQXoCxxX1fFRZO8H+gMbgEuBOapa6FzboKp16VYpvwnIA5pidrTLUdWDYp7Hu6razUW+COijqsfC0psA693kPaOqgTowzcQpmLdMnpP2qQ/5jWHnA4FPgD7ABg/yXwLDgPPDjguAva46QCuF6xVeVXhT4WaF1p70hyUKuxUWKvxQIVNhh4///T+YmnhlhONrD/KbMcbVDCgHGjnpAmxykd0Ucn82ZnxutnP+voey34/02Tkv9vu9h+uWqN9n4BbBUPMWnS0izzl/9+J/L85WqvqVk99KERmJ2UGttQfxtcARjVCTiYh74KgpdxEiizG14lwgC/htVDnDORiD3QpsRfU/iPhpAmwFblTVTyLovjvC/eH8S80v7GsR2amqx8FsySgix1xkG4Xcf0BELgUWON9jEw9l/1tEvqGqR4CeIXq3AqLWrA4ZItJWVfeGJkqi+36JsrxkHZi3/f0+7h+HqZrD09sDv0+6ztBXoVChWOERhQE+5c9WmKGwTeFthX1qNuD18r+PAs6q49oVHuTLgJ8Bk0M+V57vdpF9ETg/Qvq9QIWHspvWkX4K8D0P8uOB9ZjWQAvnuAAT83htor7fwPVhIiEiJwEjgDGqmtJtuUSkDabju0tVi1xuLgUOYCbTvQEcr3Hd2b7dR+E9MS+A0UAZqn19yftERH4d7bqq/iKKbDPnnq8jXDtNVWtFsvvQK7qzpfq+ocA04HuY/u9mYKaqvhxr2bXKCKrBOJ21YZgfzEWYJtUyVf2zi9y3gHsw1fgvgUkYT9NWYLJWb5Vel/yLwDRV/dDZ7HYD5s31XWCBRvO2iKyi2ktW6bCg6lz1wmhlR1MKGIDqWzHJmyyuV9VFLvfcpKqPxlpGhPy+i/n+xqhqlzjy2a2qp7vfmQKS3kTxeQCDgUWYuTX/i/G4lPqQfwVjJNOAD4A7MfN2JgHLPchvDvl8F/CE87kF8EHKn4kHR4XH57LLwz1xlwV8G+O0eQ84inl5dU227sn6f8KPwHX6MT/41UB/Vd0BICIP+5Bvq9XuzJtV9QEnvdDxybsR2rkdhBnTQVUPiYh751OkJXAqqiVh6d1Q/cBD+bVy9HyjSF35C5DUgU9nZu1YzPjXs8ANmBfUDI/yU+u6BDSPVa0Y5eokiAZzLsa79JqIfIrpD2T6kA+NXgiPEPCSz24RmYTp9J6LMeDKNnrjqJIiVwJzgC8QaQxch+p7ztXHnfz8EnnHoMi0xTRfvwzXDFjjQb6biPwjQrpgnGXRvIyPAO8A41R1PYD48/C1iHLNzwszFD/Pzhspb2L4q1L7AoXA3zGj9hM8yPwKaB4h/QzgeQ/y3wTmA8uBwSHpA4Hbo8obz1g753Nvx9M13Dl3HYuIkm9/hXkedF+IqZkjXVviQf59zEsl4uEi2wYoAN7ErNvwa1w8az5+ByfFIdsfD8/Oc36JyiiZB6bWGIyZ1RlPPiPjlJ8V9Z7wATJop1Ck8DPffRHoofCgQqnCSoVJKXjOsRt1zXxygNswzpKteBwWwDTnegFNnPNvAvcDf/dZfg/gQaAUM2ibsGeX1C8gjgfeCrgKmOocVwHZCcg3ps6jZ3lYo/DdsLQWCq+rGRSMXgacqXBPyBjMJIWdcep8EnAN8JKHe38R5dqtMZZ/ZrR8Q/MH9mGadRswYU3lwGwqa233cu4BtgFvY5w8cT27iOUkOsO4FTIDUCXAo8DdzjHfSRsfZ95xNRFc5aG7QqcI6Y0VrnYtAyrUhNOcEZLmOSwoRM8mwHDgOeAgxut4aZz/e9SXhVOjZERIbwMs9JD/FpwQIswg81Ggpw/9Kpzm4Bkhab6fnWs5ic4wboVM+7dWbQKcDHyczC/duad1HUcboMx3uXCKOuNdHu69QmGpmniy3ysMUn+xZHG55F3ydhvpXwAUA/1C0m52mkWutRNhTVaixIbVIX8FxkG0G+PZHISPZ+f1CNzApYh8DOSrEwsWkt4KE3XayUV+EzVD7KsuAWeqalMX+R3UHnSsQlU7RBHuA8wE/oHp9D6JCe3IAMaj+kq0skPyOQkTsTwWuBDj7fsjqn9x0b0C45K/Tqtd8p+qakdP5UbP23W0XUT6Ytbb/hA4GxP0epu6DBY7sl9gfvCVjAk9V9WfedQz4rNTl2fnlSAazLWYEfq/UL0UbXvMvIhfq+rjLvLfiXZdVXcmQM26Cl+PGexshXnjDkV1LSJnA0/jMrWgjjxPxoTGXIXqoOi3Sh7mhzYaqHTJ/1JVoz6TEPnw+TxVl4Bmqhp1GELMGNSDwBDMS+Jq9Rid4HzvdaKqi73kE5Zn1bNTl2fnOc+gGQxU/aMXUb1W82eYSUTh4wup0sdbiIdIMap5zuetqHYOufZ+TAZTLb8LD/FUITr3xbxlRwIbMW/ZhC87FFLeNcAM4HfAQ0B3TG3zMcYd/0Wyyvagm6dYNC8EcoqyYxgrQw+vxiIih0TkYITjkIgc9KqDiHxbRKaIyHuYIL4MnElsUQiNBAgPQkzpm0lV16jqJIyLdzZmPlAyGQUMVNUHVPW4mkDVH2C8XmvdhEWkv4iMDzl/XkTecI7YYvCSQOBqGKdZMR/TrCnDNAdyMFHAN6vfiF//5YeHeDyLCfGou+9SLfwf4J9UT8I6UnkFyEI1eqRA9Lw91TBOX28ItWvnAzGXHScicqqq7nO553XMeMkW53wTcB3GLX6Xqg6Jo/yE1TBBDI15HDMJ6t3QRDEd6kWYqj6ZxB7ioeonhKc2ccZTOW/oezD9v8pw+oHA/SIyQ1WTtphIaO0QAcU4QKLRstJYHD5xailE5Dceyk9GLFotgmgwJ4UbC4CqrnU8IMmmHaaj+JAzVeBZ3GLIEke88VT/jRm7qFGbOH3Cd6kdW5dI8utIvwxT27kZTHboiaqOCDn1EjiajFi0WgSxSTYXM/fkCaq9ZKdjBjR3aNj+M0nWJQcTZTAW0zT4o6relary/RKvSz6BeghwNWZqxRbgPnWJ1BaRPwPzVfWlsPRLgJs0xRMH6yJwBgNVM+cup2Y7/E9qFj9Pl06dgLGq+qskFjI36nWXsYh4XfLxIiKNMP2O2zEd/d+oqqcNtETkDEx08RpMaAyYuf19gUtU9WMX+ajPzus4jqueQTSYdOKMJbRVZyEJERmN6cCD6TzvrVM4/sJDxyJmYPoj1XgYi0iXS15EbsHM/X8deEBD1iXzkUdTTM1U6brfjImyPupBNuqzi2UcJ2I59clgRGSBqk5wvzO+MoA1lW9jEdmOmVrQDLM2V0Eyyw9RJOZxG2ellCqDSaqRV5dZAXyBCaAM/VEJZhEM384aETkFKFefP1LxsP5crASu0y8idU1SEiAVy4bmAzeGnB9yxjMQs0hgqvD9JqvLJS8iqXDJR3K7C6b/+XM3YYkSViQi49VrWJEhabVA4AwG84baSfgCEub8mykov1HYG+2akM/Z4TcHjMdJk0s+NORIRHpQvdrNDswCJm48QnVY0RvAUMczasKKnJmv6SaIBvMpMEgjLLMq3haji5cKEfmWqn4OoKofOmWfhrcF5WKnZizXN6iOTPC6NnPaXPIicibGmzgWs3PxM5gm/0CPWTSqDJAUkV+p6loAVd1mnG6u5dd4dhL27NTvutZ1KZmITBLMHEwof6R1if8nBeU/CPxZRG7DTNkFMxd/lnMkD9VoYwleeFlEXiKySz7Zb+htmEjpS1R1O4CITPEhH1dYkcb/7DwRuE6/mO0Odle+4Z0R5JGYZtp0VY20SEOidRiCaR50IUkLwiWLdLnkReQKTKxdP4xxLgX+4CmkyMhHDSvSeMKKEkgQDWYD8F+q+g8ROQ/z4CdhVnbvrKqj0qjbrZqgbRMaKsmej5JugmgwGytdkCIyD9inqtOd82KtDJ9Pj24JC+JLNalwyUco09d8FBHJALMgvZiVT7+HmTGa9FaFV4IY3p/pjBiDmWb6Rsi1dPe5Er4wXCIRkdZ1HG1IjUu+Bqr6paou8GgsVwB7gM9E5HJMf+hB4AMxOwEEgnT/ACPxNPCmiOzHdP5WQ1XoxFfRBFNAsKrj2qTbJR8P92Dc3s0wE97yVfUjZwbtC0DUNbVTReAMRlXvc+ZGtAP+EjImkoHpyyQVt2m6yS4/TtLtko+LEEfPrsoYNFXdWdlUCwKBMxgw4wYR0qIG3yWw7JS4J5NEul3ycSEiGWo21PpxSFom3jZkSgmBNBhLzKwDquLGwl3yadLJKxMwhnFUVdeFpJ+OCZkJBIHzklliJ8gu+YaCrWEaFpkhLtirMBtAvQC8ICLFadTLFal7PTkANFG7IMeJNZiGRaaIVG7OOgjTzKkk6N/1Jc5fwUwkS7kb3AtBf4gWfwTZJR+VsGjnf2kyF1yMA9uHaWA4ofyVLvl/OmlnYvbMSeoSVYlCRDaoaiybTyUdW8M0MNLpko8HEQk1kGZh5wTF2G0NYwkEIrKSmovA1/hhaqw7UCcYazCWQCAivTHTOvY459dixpBKSdG0Di8EJuTAcsIzH/gXgDOG9BtgMcZZkbRF1P1i+zCWoFAvxpBsDWMJCkGe1lFFYBSxnPDUizEk2+m3BIb6MIZkDcZi8YHtw1gsPrAGY7H4wBqMxeIDazAWiw+swVgsPvh/IPsmb9V5EUQAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "np.random.seed(0)\n",
    "ident0, ident1 = 0, 1\n",
    "_ = compare_clusters_multiple_runs_less_genes(ident0, ident1)\n",
    "p_tn = compare_clusters_tn_test(ident0, ident1)\n",
    "\n",
    "ngenes = 10\n",
    "genes_to_view = ['S100A4', 'S100A11', 'B2M', 'HLA-A']\n",
    "genes_to_mark_red = ['B2M', 'HLA-A']\n",
    "tsne_select2(ident0, ident1)\n",
    "create_and_save_visualization(p_tn, ident0, ident1, genes_to_view,\n",
    "                              genes_to_mark_red=genes_to_mark_red, ngenes=ngenes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "      ACTB | type 1 with mean 3.51\n",
      "      ACTB | type 3 with mean 4.13\n",
      "       B2M | type 1 with mean 5.06\n",
      "       B2M | type 3 with mean 5.60\n",
      "      CCL5 | type 1 with mean 0.34\n",
      "      CCL5 | type 3 with mean 3.58\n",
      "      CCR7 | type 1 with mean 0.86\n",
      "      CCR7 | type 3 with mean 0.08\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x129.6 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "      CST7 | type 1 with mean 0.07\n",
      "      CST7 | type 3 with mean 1.89\n",
      "      CTSW | type 1 with mean 0.55\n",
      "      CTSW | type 3 with mean 1.97\n",
      "      GZMA | type 1 with mean 0.10\n",
      "      GZMA | type 3 with mean 1.93\n",
      "      GZMH | type 1 with mean 0.03\n",
      "      GZMH | type 3 with mean 1.11\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjgAAAB6CAYAAACyeim4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAACXxJREFUeJzt3XvIZHUdx/H3Vy0zr5hoZZj+UbjYhVIhdb10gVrEtOiiRrtF/RF2MdLMSEIDxbIrlAWBlaXkBSvC1sIsXXUVn5AK2k2tTC23SNdal8128dsf54yO4zz3eWae+c77BcM+8ztnzmXnO7/zOef8nnkiM5EkSapkp1FvgCRJ0qAZcCRJUjkGHEmSVI4BR5IklWPAkSRJ5RhwJElSOQYcSZJUjgEHiIjTI2IqIh6PiIcjYm1ErIyIfSLisojYFBFbIuKeiDg3Ig5q5+08MiK2dj0/tl1G9zz/i4jfj3pftTjT1MqNPe/z9q7na9vXfSAiNrZ19I+I+FlE7BkRR7VtO3et49vTtH1rFPusxYmIUyPizraP+Gf78xkR8dKePqLz2BERN7WvPb/tX87sWeaZbfv5Pe2HRMSTEfHNIe6iBmwYNRMRJ0TEQ33W/euI+OBQdnSJTXzAiYhPAF8FLgIOAA4CLgVOBr4C7AGsAPYG3grcl5kPZOYenUe7qFd3ta3LzFU989wOXDPk3dMAzVArd3e9zxcBV3W996si4vi2/bTM3JOmnq5qFztF8zl8bdeqjgUe6mk7Drhl6fZOSyEizgK+BlwCvJCmbj4EHANs6u4j2vo5GthGUy8d9wCrexa9pm3vtRrYDLw7InYd6M5oKEZQM2VNdMCJiL2BzwEfzszrMnNrZm7PzJ9m5ieBI4ErM3NzZj6ZmRsz89oFrOdgmoPW5YPcfg3PHGplJkcC6zPzboDMfDQzv5eZWzJzO3AHTYAhIvYHngtc3dP2cgw4Y6WrZs7IzGvb9zsz8+7MfE9mPtEz/17AtcDnM/PGrkl3Ac+PiMPa+Q4Dnte2d78+aA5q5wHbgZOWat+0NIZdM9VNdMABjqJ50380zfQ7gAsj4v0R8bJFrGc1sC4z71/EMjRas9XKTO4E3hwRF0TEMX3OrG+hDTPtv7e2j+62v2Tmsy4na1k7CtgV+Mkc5/8OcC9wYZ9p3+fpM/I17fNeK4GXAD+kCchr5rOxWhaGXTOlTXrAeQHwr8zcMc30jwJXAB8B/hAR90XEqgWsZzXw3YVtopaJ2WplWpm5Dng7zS2n64FHIuLLXWNsbgZWtmfgxwLrgPXA67rabh7APmi49qOnZiLi9oh4LCK2RcRxXe1nAYcD783+fyDwB8BpEfEc4NT2ea81wNrM3AxcCbylvfqn8THsmnlxu+ynHjRBuYRJDziPAPtFxC79Jmbmtsy8KDMPpznAXQ1cExH7znUFEbGS5j7qvG9taVmZsVZmk5lrM/MkYF+a8V3vAzoD+e6gGev1CpqrNesy83Hgwa42b0+Nn2fVTGYenZn7tNN2gqf6iAuAd2Tmo/0WlJkPAPfRjLO4NzMf7J4eEbsB76Q5ISMz1wMPAKcPeqe0pIZWM62/Z+Y+3Q+aq8clTHrAWQ88AZwy24yZ+R+aQtkdOGQe61gDXNcesDS+5lwrM2nHcv0SuIkmvJCZ/6W5N34S8KLM3NjOvq5texUGnHHUqZmTp5shIg6gGXB+dmZOzbK8y4Gz6D+W723AXsCl0fzW5ybgQLxNNW6GWTPlLehstIrM/HdEfBb4RkTsAH5BMzjvTcDrgS3ADcBvacLgmcBjwB/nsvz2rOpdNJ2PxthstZKZ50z32og4GdgN+DlN/RwJHA98vGu2W2jq67autltpfkvr4cz80wB3R0OQmY9FxAU0oSNo3v+tNIF1d2BnmvEyN2XmXL4C4Cqa3667rc+0NcBlwGe62g4E7oqIV2amX1ExBoZcM+VNdMAByMwvtWc759Fc3t0C/IZm0NYbaAZxHQTsAH4HnDiPqzGn0BzQfjXo7dbwzVIrM9kMfAz4Os0AwoeBSzLziq55bgY+zTMvD98K7E/ToWkMZeYXIuJvwDk0Z9FbgT8Dn6IJyCcA2yKi9yTor5l5WM+ytgE39sxHRBwIvBF4TWZu6pq0KSJuoAk/Zw9mj7TUhlEzkyL6j02SJEkaX5M+BkeSJBVkwJEkSeUYcCRJUjkGHEmSVI4BR5IklWPAkSRJ5RhwJElSOQYcSZJUztC+yfjgc6+fArj/4hOPGNY6Nd46NQPWjebGmtFCeHyqySs4kiSpHAOOJEkqx4AjSZLKMeBIkqRyDDiSJKkcA44kSSrHgCNJksox4EiSpHIMOJIkqRwDjiRJKseAI0mSyjHgSJKkcgw4kiSpHAOOJEkqx4AjSZLKMeBIkqRyDDiSJKkcA44kSSrHgCNJksox4EiSpHIMOJIkqRwDjiRJKseAI0mSyjHgSJKkcgw4kiSpHAOOJEkqx4AjSZLKMeBIkqRyDDiSJKkcA44kSSpnl1FvgCRJy8GGQ1dMdX5esXHDEaPcFi2eV3AkSVI5BhxJklSOAUeSJJVjwJEkSeUYcCRJUjkGHEmSVI4BR5IklWPAkSRJ5RhwJElSOQYcSZJUjgFHkiSVY8CRJEnlGHAkSVI5BhxJklSOAUeSJJVjwJEkSeUYcCRJUjkGHEmSVI4BR5IklWPAkSRJ5ewy6g2Qqthw6Iqpzs8rNm44YpTbIkmTzis4GgsbDl0x1R0gJEmaiQFHkiSVY8CRJEnlOAZHUkmOiZImm1dwJElSOUO/guNZlcZNp2atV0kaH96imgcPdBoEQ74kLT1vUUmSpHK8gqOJd/C5108B3H/xiQO/muLVGkkaDQOO1Ecn9ACsHcL6DEJaiLneNre+NIkMOIvUr+OwM5E0Ko4VlBoTFXAMHhoUDyIaBPskaelMVMAZJTsySYthqNYwVThmGXCmYWciadJUOKhJHRMfcEb5gbYzWV6e8dfKT/ninF4z7MHI0qDY/2ghxunkf+IDjjRXhhkNg8FDw1a15gw40jI2TmdLkgZrPsFjlF8ZsFz7qYkIOJ0zb8+69VQt/PjspxvneDtqQeuZpW0hlmtnomd7xlW/rppb1dbcuPRJ/WrOOlz+Jv09GtuAM2vHMcsBbKYDUHf7KG5LTHpRDsqwwsxiDDr0wNJ+H5O12ejX/yz2/2QxfdJ8+q6lPOGbKQj1tqvRr5+aLQCPw+dwOWxjZOao1i1JkrQk/GObkiSpHAOOJEkqx4AjSZLKMeBIkqRyDDiSJKkcA44kSSrHgCNJksox4EiSpHIMOJIkqRwDjiRJKseAI0mSyjHgSJKkcgw4kiSpHAOOJEkqx4AjSZLKMeBIkqRyDDiSJKkcA44kSSrHgCNJksox4EiSpHIMOJIkqRwDjiRJKseAI0mSyvk/8f3/PEzGXPAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 576x129.6 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     HLA-C | type 1 with mean 3.16\n",
      "     HLA-C | type 3 with mean 3.93\n",
      "      NKG7 | type 1 with mean 0.21\n",
      "      NKG7 | type 3 with mean 3.30\n",
      "     RPL13 | type 1 with mean 5.08\n",
      "     RPL13 | type 3 with mean 4.71\n",
      "    RPL13A | type 1 with mean 5.03\n",
      "    RPL13A | type 3 with mean 4.72\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x129.6 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     RPL32 | type 1 with mean 4.71\n",
      "     RPL32 | type 3 with mean 4.15\n",
      "     RPS12 | type 1 with mean 4.67\n",
      "     RPS12 | type 3 with mean 4.21\n",
      "     RPS27 | type 1 with mean 4.60\n",
      "     RPS27 | type 3 with mean 4.19\n",
      "     RPS3A | type 1 with mean 4.32\n",
      "     RPS3A | type 3 with mean 3.86\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x129.6 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "      RPS6 | type 1 with mean 4.90\n",
      "      RPS6 | type 3 with mean 4.50\n",
      "    S100A4 | type 1 with mean 1.73\n",
      "    S100A4 | type 3 with mean 2.97\n",
      "      SRGN | type 1 with mean 0.66\n",
      "      SRGN | type 3 with mean 1.61\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.6/site-packages/ipykernel_launcher.py:189: RuntimeWarning: divide by zero encountered in log10\n",
      "/opt/conda/lib/python3.6/site-packages/ipykernel_launcher.py:189: RuntimeWarning: invalid value encountered in log10\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Welch's $t$\n",
      "      NKG7\tpDE: 74.84\tpTN: 268.90\n",
      "      CCL5\tpDE: 73.15\tpTN: 157.59\n",
      "      CST7\tpDE: 43.15\tpTN: 177.53\n",
      "      GZMA\tpDE: 41.45\tpTN: 132.24\n",
      "      CTSW\tpDE: 35.67\tpTN: 41.93\n",
      "     HLA-C\tpDE: 32.53\tpTN: 33.71\n",
      "       B2M\tpDE: 29.33\tpTN: 34.44\n",
      "      CCR7\tpDE: 26.86\tpTN: 18.63\n",
      "    S100A4\tpDE: 25.75\tpTN: 25.19\n",
      "     RPS12\tpDE: 25.61\tpTN: 28.52\n",
      " \n",
      "wilcox\n",
      "      NKG7\tpDE: 65.70\tpTN: 268.90\n",
      "      CST7\tpDE: 54.08\tpTN: 177.53\n",
      "      CCL5\tpDE: 50.13\tpTN: 157.59\n",
      "      GZMA\tpDE: 47.73\tpTN: 132.24\n",
      "     HLA-C\tpDE: 37.74\tpTN: 33.71\n",
      "      CTSW\tpDE: 36.15\tpTN: 41.93\n",
      "       B2M\tpDE: 35.06\tpTN: 34.44\n",
      "     RPS12\tpDE: 34.62\tpTN: 28.52\n",
      "     RPL32\tpDE: 32.74\tpTN: 46.57\n",
      "      GZMH\tpDE: 31.50\tpTN: 59.02\n",
      " \n",
      "bimod\n",
      "      NKG7\tpDE: 76.86\tpTN: 268.90\n",
      "      CCL5\tpDE: 60.51\tpTN: 157.59\n",
      "      CST7\tpDE: 53.77\tpTN: 177.53\n",
      "      GZMA\tpDE: 47.08\tpTN: 132.24\n",
      "     RPL32\tpDE: 40.99\tpTN: 46.57\n",
      "     HLA-C\tpDE: 40.89\tpTN: 33.71\n",
      "     RPS12\tpDE: 40.69\tpTN: 28.52\n",
      "      CTSW\tpDE: 39.62\tpTN: 41.93\n",
      "      SRGN\tpDE: 36.89\tpTN: 26.93\n",
      "      GZMH\tpDE: 33.49\tpTN: 59.02\n",
      " \n",
      "tobit\n",
      "      NKG7\tpDE: 81.68\tpTN: 268.90\n",
      "      CST7\tpDE: 59.31\tpTN: 177.53\n",
      "      CCL5\tpDE: 57.87\tpTN: 157.59\n",
      "      GZMA\tpDE: 51.17\tpTN: 132.24\n",
      "      CTSW\tpDE: 37.36\tpTN: 41.93\n",
      "     RPS12\tpDE: 36.52\tpTN: 28.52\n",
      "     RPL32\tpDE: 34.58\tpTN: 46.57\n",
      "     HLA-C\tpDE: 34.53\tpTN: 33.71\n",
      "      GZMH\tpDE: 34.03\tpTN: 59.02\n",
      "       B2M\tpDE: 30.33\tpTN: 34.44\n",
      " \n",
      "poisson\n",
      "      NKG7\tpDE: 290.56\tpTN: 268.90\n",
      "      CCL5\tpDE: 268.39\tpTN: 157.59\n",
      "       B2M\tpDE: 140.79\tpTN: 34.44\n",
      "     RPL13\tpDE: 139.94\tpTN: 31.41\n",
      "     RPS12\tpDE: 126.84\tpTN: 28.52\n",
      "     RPL32\tpDE: 121.80\tpTN: 46.57\n",
      "      ACTB\tpDE: 100.03\tpTN: 27.75\n",
      "     RPS3A\tpDE: 98.96\tpTN: 10.68\n",
      "      RPS6\tpDE: 98.13\tpTN: 22.05\n",
      "    RPL13A\tpDE: 96.76\tpTN: 17.18\n",
      " \n",
      "negbinom\n",
      "      NKG7\tpDE: 107.37\tpTN: 268.90\n",
      "      GZMA\tpDE: 52.32\tpTN: 132.24\n",
      "      CST7\tpDE: 48.78\tpTN: 177.53\n",
      "      CTSW\tpDE: 37.75\tpTN: 41.93\n",
      "     RPS27\tpDE: 31.09\tpTN: 26.88\n",
      "     HLA-C\tpDE: 30.53\tpTN: 33.71\n",
      "      CCL5\tpDE: 28.93\tpTN: 157.59\n",
      "      SRGN\tpDE: 28.79\tpTN: 26.93\n",
      "     RPL32\tpDE: 28.25\tpTN: 46.57\n",
      "     RPS12\tpDE: 27.45\tpTN: 28.52\n",
      " \n",
      "MAST\n",
      "      NKG7\tpDE: 77.39\tpTN: 268.90\n",
      "      CCL5\tpDE: 57.61\tpTN: 157.59\n",
      "      CST7\tpDE: 54.71\tpTN: 177.53\n",
      "      GZMA\tpDE: 48.06\tpTN: 132.24\n",
      "     HLA-C\tpDE: 40.22\tpTN: 33.71\n",
      "      CTSW\tpDE: 37.61\tpTN: 41.93\n",
      "     RPS12\tpDE: 34.56\tpTN: 28.52\n",
      "      GZMH\tpDE: 33.31\tpTN: 59.02\n",
      "     RPL32\tpDE: 32.72\tpTN: 46.57\n",
      "      SRGN\tpDE: 31.01\tpTN: 26.93\n",
      " \n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x129.6 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAANMAAACKCAYAAAAwhn/mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAF69JREFUeJztnXuYFMW1wH+HFVgSnqLxhXEhokFcWB5r9gqIBq8iRBEUI2rQxIircIOgV8iVGLhqop+LEFYil8QHGgH1SsQoiYlRFEFUHgvIw8vDVTZBhVVeQQyEc/+onmV2mJnumel9zHJ+39ffTlf1qaru7dN1+tSpalFVDMPInEZ13QDDaCiYMhlGSJgyGUZImDIZRkiYMhlGSJgyGUZImDIZRkiYMhlGSJgyGUZIHFPXDQA47rjjNC8vr66bYRhHsHz58h2qenyQY+uFMuXl5bFs2bK6boZhHIGIfBT0WDPzDCMkTJkMIyTqhZmXMhNbJUjfVbvtMIwoslOZjnIOHDhARUUF+/fvr+umNBhyc3Np164djRs3TrsMU6YspKKighYtWpCXl4eI1HVzsh5VpbKykoqKCtq3b592OfbOlIXs37+ftm3bmiKFhIjQtm3bjHt6U6YsxRQpXMK4nqZMhhES9s7UAMgb/3Ko5ZXfPzDU8pKxcOFCSkpKeOmll47IGzBgALNnz6Z169a11p5MMGUy6i0LFiyo6yakhJl5RlqUl5fTqVMnbrrpJjp37sxFF13El19+yebNm+nfvz89evSgT58+bNiwAYDNmzdTVFREfn4+EyZMoHnz5lVl7d69m4EDB3LmmWdSXFzMoUOHABdmtmPHjoR1AZSVlVFUVESXLl0YPHgwX3zxBQDnn38+Y8aMoWfPnnTq1In33nuPIUOG0LFjRyZMmFAj18SUyUibjRs3MnLkSNauXUvr1q15/vnnGTFiBKWlpSxfvpySkhJuvfVWAEaPHs3o0aNZs2YN7dq1q1bOu+++S2lpKevWrWPz5s3MmzcvUF0Aw4cP54EHHmD16tXk5+czadKkKpkmTZqwbNkyiouLGTRoENOnT+f999/niSeeoLKyMvTrYcpkpE379u0pKCgAoEePHpSXl7NkyRKGDh1KQUEBN998M9u2bQPg7bffZujQoQBcc8011co555xz6NChAzk5OQwbNoy33norUF27du1i586d9O3bF4Drr7+eN998s0rmsssuAyA/P5/OnTtz0kkn0bRpUzp06MDWrVtDvhr2zmRkQNOmTat+5+Tk8Omnn9K6dWvKyspSKifWLR3PTR1bV8TMC9K+Ro0aVZNv1KgRBw8eTKmNQbCeyQiNli1b0r59e5577jnARRasWrUKgKKioirTbO7cudXk3n33XT788EMOHTrEM888Q+/evQPV16pVK9q0acOiRYsAeOqpp6p6qbrAeqYGQG26sv14+umnueWWW7j33ns5cOAAV199NV27dmXq1Klcd9113HffffTv359WrQ4HKxcWFjJq1Cg2bdrEBRdcwODBgwPXN2vWLIqLi9m3bx8dOnTg8ccfr4nTCoTUh7XGe/bsqSlNDjzKo8bXr19Pp06d6roZKbFv3z6aNWuGiDB37lzmzJnD/Pnz67pZ1Yh3XUVkuar2DCJvPZNRKyxfvpxRo0ahqrRu3ZrHHnusrpsUOr7KJCKnAk8CJwAKzFTVX4nIscAzQB5QDlylql+Ie3v8FTAA2AfcoKoraqb5RrbQp0+fqvenhkoQB8RB4HZVPQsoAkaKyFnAeOCvqtoR+Ku3D3AJ0NHbRgCPhN5qw6iH+CqTqm6L9CyqugdYD5wCDAJmeYfNAi73fg8CnlTHUqC1iJwUessNo56RkmtcRPKAbsA7wAmqus3L+gRnBoJTtOgRsQovzTAaNIGVSUSaA88Dt6nq7ug8dS7BlNyCIjJCRJaJyLLt27enImoY9ZJA3jwRaYxTpKdVNRI49amInKSq2zwz7jMv/W/AqVHi7by0aqjqTGAmONd4mu03IPFQQdrlJR9iqKyspF+/fgB88skn5OTkcPzxbp3GVatWMXbsWCZPngxASUkJe/fuZeLEidXKWLhwIU2aNOHcc89NuXmRsKXYsKS6xrdn8rxzjwLrVfWhqKwXgeu939cD86PSh4ujCNgVZQ4aDYC2bdtSVlZGWVkZxcXFjBkzpmq/adOmzJs3jx07diQtY+HChSxZsiSt+svLy5k9e3ZasjVJEDOvF/AD4LsiUuZtA4D7gX8XkY3Ahd4+wAJgC7AJ+A1wa/jNNuorxxxzDCNGjGDKlCkJjykvL2fGjBlMmTKFgoICFi1axPbt27niiisoLCyksLCQxYsXA/DGG29QUFBAQUEB3bp1Y8+ePYwfP55FixZRUFCQtJ7axtfMU9W3gEQT5PvFOV6BkRm2y8hiRo4cSZcuXbjzzjvj5ufl5VFcXEzz5s254447ABdJPmbMGHr37s3HH3/MxRdfzPr16ykpKWH69On06tWLvXv3kpuby/33359wdm5dYhEQRui0bNmS4cOHM23aNJo1axZI5tVXX2XdunVV+7t372bv3r306tWLsWPHcu211zJkyJAj5kLVJ0yZjBrhtttuo3v37vzwhz8MdPyhQ4dYunQpubm51dLHjx/PwIEDWbBgAb169eKVV16pieaGgk3BSIeJrY7cjGoce+yxXHXVVTz66KNx81u0aMGePXuq9i+66CJKS0ur9iNzojZv3kx+fj7jxo2jsLCQDRs2HCFbX7CeqSFQT6Plb7/9dh5++OG4eZdeeilXXnkl8+fPp7S0lGnTplW9ax08eJDzzjuPGTNmMHXqVF5//XUaNWpE586dueSSS2jUqBE5OTl07dqVG264gTFjxtTymcXHpmCkQ7z6a/GGzsYpGNlAplMwzMwzjJAwZTKMkLB3pixFVetmvfG/r4yffnK32m1HyITxumM9UxaSm5tLZWVlKDeAcfiTMrFu+VSxnikLadeuHRUVFdRJtP3Oz+Kn71pfu+0ImcjHzjLBlCkLady4cUYf5cqIiUUJ0uune742MTPPMELClMkwQsKUyTBCwpTJMELi6HRA1HU4ktEgsZ7JMELClMkwQsKUyTBCwpTJMEIiyFJfj4nIZyLyflTasSLyFxHZ6P1t46WLiEwTkU0islpEutdk4w2jPhGkZ3oC6B+TZov2G0YMQZb6etNbYzyaQcD53u9ZwEJgHFGL9gNLRaR1ZNXXTBqZN/7lavvlmQX3GkaNkO47ky3abxgxZOyASGfRfrCF+42GR7rK9Gnkm0vpLNoPbuF+Ve2pqj0ji74bRjaTrjLZov2GEUOQb9rOwTkbjhORCuDnuEX6nxWRG4GPgKu8wxfgvmW7Cfc922DLeRpGAyCIN29YgixbtN8worAICMMICVMmwwgJUybDCAlTJsMICVMmwwgJUybDCAlTJsMICVMmwwgJUybDCIkGv9RX7FwosPlQRs3Q4JUpU0wZ6xl1/AnUZJiZZxghYT1TXVCPn65G+pgy1TBmJh49mDIdbdg66zWGKVM9J27Pdv/AtOWtV6w5TJmyjPLca2BinIwa6FkyNVHjy19z5IEJ2h52/XHrTlJ/qpg3zzBCwpTJMELClMkwQqJGlElE+ovIB94C/uP9JQwj+wldmUQkB5iOW8T/LGCYiJwVdj2GUd+oiZ7pHGCTqm5R1X8Cc3EL+htGg6YmlMkW7zeOSsStGxligSJXAv1V9cfe/g+A76jqqJjjRuC+4QRwJvBBGtUdB+zIoLlHs3w2tz1T+VRkT1PVQIvh18SgbaDF+1V1JjAzk4pEZJmq9jT57Kq7ruUzrTsRNWHmvQd0FJH2ItIEuBq3oL9hNGhC75lU9aCIjAJeAXKAx1R1bdj1GEZ9o0Zi81R1Ae6LGDVNRmbiUS6fzW3PVD7TuuMSugPCMI5WLJzIMELClMkwQsKUyTBCImuUSUS61ECZ3xKRn4lIjXobRWSqiFwlIhYJ0oDJGmUCVorIRhG5J5PAWRE5WUTGiMh7wFrcNbjaR+YhEemVbp24b/xeDiwWkXIRmS0io0Skm4gE+h+IyOci8lsR6ScikkrlIjLW+/5wbPqNInJbAPkuUb8bi8gEEXlRRH4hIl/zkR2ebAvY/nwRWSoiW0Vkpoi0icp710f2VBGZKyKLROS/RKRxVN4LQeoPStZ480RkJfADYBjwfeAfwBxgrqqWB5Af4cmeAjzrbfNVtX0A2e24D2EfDzwDzFHVlWmex8nAud52GfANVW0ZQO4DoNQ7hzzgf712LA0guxwoUtUDMelNgGWqmrTXF5EVqtrd+z0ZaAs8jntAtFXVhEohIqUJsi4DTlHVIB8pfwu4F1gK/Bj34fHLVHWziKxU1W5JZP8CPO/J3gj0AC5V1Uo/2ZRR1azYgBUx++cAD+ECaZcEkP8n8AbQMyptS8C6V3p/zwB+huvRNuC+PH9GwDIE6AIUA08By4E/AT9P9fyBbwJ3AiuALcAvfGRXJclbE/T8vd9lQOOoc1qdwv9QgOuANbiHUpeAcqti9i8ANgJFsfdFHNmymP3rvP/ft/xkU75HwyysJrfof2icf1DfAPJtvRv5DVxQ7T3A1oB1H3HRPcX4JW66iZ/8X4B3gP/BPVU7hXj+3/ZTSO/mPSFO+gkBlWkLMBi4Algfk5dQUaOOOQbXo2wAngDOTPHcVwGt4lz/jUClj+xaIDcm7UKc6b0tzHs0m96ZHoyXqI43/IRVtVJVZ6hqX6AfsBP4VETWi8gvfMSPeEdR1dWq+lNVPT1A27cAh4CO3na6iBwXQC6a1+MlquoGVZ3kI/sg8LKI9BWRFt52PvASUBKg7jdwZtn3gKUicgKAiJyIT/S1iIwE1uHMq/6qeoOqpjpD4AGgU3SCqq7G/R/n+cj+FvhOjOyrwFDg/RTbkZSseWeqKUSkIzBMVf87yTFnq2rGF15EWuJMk3O9v8cD76vq9ZmWHaDuS4DxwNmA4p7Y96vqHzMs9zuq+k6S/EPAZ8B2r96qLNyzMHQvbV2RNevmicggoJ2qTvf238HdjADjVPU5H/khSbL9FOVJoHvQtibhK2Af8KX3ux3QJKiwiFzsyfxVo5wuIvIjVX0smaynNBkpTgKew73DJcLXweOH1wP+HNe73w38B57JCYxW1W1plnt3sodoyuVlS88kIouBq1V1q7dfhuvmvw48rqr9fOQP4V6eyyJJUdmqqj9KIpuR10dEpuB6o47ASuBtYDHwtqruDFjGL4FeOKfDpcBUVS318qq8bSm0KWWZBOVsVdVT/Y/MqI4/AS/j/tfXAE8Ds3HexAtVNa1lEUTkY1VN9iBIiazpmYAmEUXyeEtVK4FKEfl6APkhuPGkLsB8nFt5U8C6TxGRaYkyVfUnPvIfAr/DeZb+FbDOWL4HdFM3xWUiMFtEOqjqGOK80wUgHZl4JH0ae+Nbx6rqg97+34AWXv3/qaozAtRxQtSD41ZVfcBLL403fhZT/+5EWUCzAHUHJpuUqU30jlafBu87rVhVXwBe8BRvEDBZRNoCdwVwYHyJc2Wny3xgZ0SRROQC3FP1I+BhdQvP+HGMqh4EUNWdInIpMFNEniMFUzGKI9ceToCI/IH4SiM4L2kyioH+UfufqeopIpKLm/MWRJmiHWVPxuTl+MjuBApV9dPYDBHZGuf4tMkmZXpHRG5S1d9EJ4rIzUDSUfAY9gO7gN3AaUCQ1asrVXVWCnXE8izOtbxLRApw7xm/BLoCv8a5jf3YLCJ9I4rvKeaNInIv7v0hJVR1goj0xjlfRvocnszj5+cNFM+CiPCcV/9+EQnaM8wXkeaquldVJ1QVLHI6/muHPIn7Px+hTDhTMTSy6Z3pG8ALuBf3FV5yD6ApcHm8J0+M/HdxZt45wKu4yIllAeteqqpFGbR9dcRrJSIlwCFVvdMLJSoL4tGK3Hiq+mWcvFNU9Yh1NhKU0w333jEUZ37Oi5hQSWSOB45X1XUx6WcB21V1exLZTfGGD7xz36SqHYK0O0n5V6jq85mUERZZo0wRPKXo7O2uVdXXRKSJn6nkOSBWA2/hTJZqJ57svUdE8oAvVHWXt5+SmSYia1Q13/u9Avipqr7i7a8OqEyn4UzFlNsgImfgwpCG4caFngHuUNXT/Or15OcCv1bVN2PS+wC3qGqCz0uAiPwa+Dy6R/HS7wWOU9XiIG1IUn5SJ0Im1y3ltmSLMiVyY3pjNy+q6vk+8jeQ5GU5mRnnueEHq+rfPTPtVZyZ1gU4oN6yZknkpwEnAttwg59nqOoBETkJ+IMGWCknkzZ4D5JFwI0Rp4uIbAnaK0iS1XxE5H1VPTuJ7NdxA6eFuEgGcObtMuAmVd0TpA1Jyk/qTcz0f5cK2fTO1FtE7lPVuyIJ3kj8K/iPggM8q6r74mWIiN9YSDNV/bv3+zrcIjGTI2ZagLpH44JzTwJ66+GA0xOBuxJKhdeGiCfzdc/NPJfUvHktkuQ1TpKHqv4Dt0R2Bw5bFOtUdXMK9Setwic/0/9dCi0JMTapJjeco+Al4CFvvyMuvqo4oPwBYBLQKE6eX7DkmuhjgYuj9n0DPYE/h3D+GbXBOy4yTvMHXNT9I8BFAeReBgbESb8E+GOa53MG8Jug544z0WO3NcBXNX3dgm5Z0zOp8/4MBp4RkTm4QdDbVPX3AYvYgosUXiwi16jqh1F5fk/p10TkWZyZ1gZ4DcAz04LY3IFWBK3hNqCul5iNG6Nqg3NCjAP+7CN6Gy627yoODxH0BP4NN/6VEHFzoUqAk3EOpOnAw7h4uclB2u1Xhw8ZX7egZNM701jvZ2Pc9INFQNULsao+5CO/QlW7i8h1OJv5LlV9MjoviWxHDs+helY9z5mInAfkqxfilER+C3BHonxV9TVTRUQ4bCpGt6Ebbk7UK35lJCjXNwrAc0GfiLMGIu9Ha4H/w0VeJzTZvHeWR3BRH5cAPwVmAXer6v4U2nk5cDqupwl8rjV13eKRNT0T1e32aXHSAqGqvxM32ewpERkA3BxAbArOA7cmJv0L3IBkUmUCWuGervF6QCXAO5+qqojsxw1Sno235LSmOUkxRabizv/x6EQRyffyLk0i21RVn/B+fyAiP1HVO1Op3PMIdgaWAPeIyDmqek8Q2dq8blmjTOo/zcCPqhtZVctFpC9uot9K/MNKToijSKjqGs9t7sdHmiT2Lwgi8gjue1cp31AhkMn553q9QOT6fxW9r6orEkoe5jygq6r+S9w0+UW4+Wi+1OZ1yxplEpG7k2RrgAtULXxGVQ8Bkzzvlp+itk6SF2QU/0wR6aWqi6MTxa0r8UkyMymKPqR/Q41NlAU0D1BEJue/DfduFFGmT6j+rvTdAPX/U71QLFXd55luQUn7uqVKNk0O/EecDdy8/nEB5JeJm6gGOFvee5eZAzwaQPam2EQR+THBYvbewYUvxbIbZyYFodoNRequ7Xhbc+BXAeQzOf9xwLWqeoGqXoCbabsXN+3lyoDt/7aIrPa2NVH7a0RktY9sJtctJbLGARGNiLTAjd3ciIt7m6yqn/nIpD2FwxvP+j3O+xPtzWqCGxD8xKfu91S1MEFeVXSETxn7cEMB4G6Ib3n7NT7JLpPz9yI+LlTVzz2HzVzcfKQC3PR9X4XyohgSoqofJZGtteuWNWYegIgcC4wFrsV5hLqr6hcBxdOewqEu7u9cLxQl4s16WVVfC1h3myR5QYM9O/kfEh9JMn0E/KeQZHj+Oar6uff7+8BMdbF0z3sPNF8SKYs38DoMFxqUiLSvW6pkjTKJyIO4kfyZOHf03hSLyGgKhyfzOgnWYvDhPYkf8R7UTMz0hoquYxJu1mrKpHn+OSISmT7Sj8Nfi4SA958XMjYSt0zbi7gFakYBt+NClJ5O0uZMrltKZI2Z58WXfQUcJP5aAknXnhORp4GFcW7om4HzVXVYyE2OriMjM9ErI+kNpQFnm0rYa8X513cXMAAXYPtNnDWh3tjVLFX1XdxTRObjhiHexinkN3D/99GqmrR3C+u6BSFrlClTJMMpHCG1IdpMWpuCmZjRDRVTTijT1VNBRIpwg6Z/9qIwIpHszYO4xqV61H0OzkP4zSCDvmFdtyAcNcoUQeJM4ajL9gQlkxsqppxaV6ZMiW1zKucQ1nULQta8M4WFpzxZoUAxVC1t7I2ZVAS9IURkD4dN46/J4XURApnI9YCuMW1u5u0HaX/a1y1VjrqeKVsRkX9xeGwtshhIZNwkGxSiTqjN62bKZBghkU0REIZRrzFlMoyQMGUyjJAwZTKMkDBlMoyQ+H+BrZIB29ekQwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "np.random.seed(0)\n",
    "ident0, ident1 = 1, 3\n",
    "_ = compare_clusters_multiple_runs_less_genes(ident0, ident1)\n",
    "p_tn = compare_clusters_tn_test(ident0, ident1)\n",
    "\n",
    "create_and_save_visualization(p_tn, ident0, ident1, None, ngenes=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     ALDH2 | type 2 with mean 0.87\n",
      "     ALDH2 | type 5 with mean 0.30\n",
      "       B2M | type 2 with mean 4.73\n",
      "       B2M | type 5 with mean 5.13\n",
      "     BLVRB | type 2 with mean 1.11\n",
      "     BLVRB | type 5 with mean 0.49\n",
      "     CD79B | type 2 with mean 0.11\n",
      "     CD79B | type 5 with mean 0.91\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjgAAAB6CAYAAACyeim4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAACwpJREFUeJzt3XusHGUZx/HvwyUCAbmHIiJVQalKMIom3CGoSLBQ/ymGShEkRpFgFEiKJoKGKALeETHxQkFACSKItUpixLSJqIVIBFtFoTEEK5R7y0Uuj3/MLEyHPae7p+ec3X3P95NMcvad6+68+85v3pk5G5mJJElSSTYb9AZIkiRNNgOOJEkqjgFHkiQVx4AjSZKKY8CRJEnFMeBIkqTiGHAkSVJxZkTAiYhbI+LRiHhVo+yKiLhgjOkzItZHxLqIeDgifhsRJ3RZ5mmtsiMi4v7G60si4p6IeDIiVkXEwsl+bxqMiFgdEU/XdeTRiFgSEXvW486JiLvq/X5fRJzTmjcj4sGI2KJRtmVd5j+mKsBG6kfXtqduK17RRkTEWRFxW/338oh4pl7uQxFxfUTs1pj2goh4rh6/LiL+FhHzpvK9ampExIkRsaLej/+JiKURcUhEnF/v4yfr4R8RcWlE7N6Yd0GjDqyLiKfqdued9fgdImJx3eY8GBHnt9bdPAaujYhrI2KHaf4INlnxASciZgOHAgkc18es+2fmtsCbgSuASyPivD5Xvx6YC2wPnAx8MyIO6nMZGl5z6zqyO/Bf4Nt1eQALgR2B9wNnRMSHWvM+ChzTeH1MXaZyjFU/xnIlVb1pOwlY3Hj98Xq5b6KqYxe1pr86M7etpzkbuDYidpnIG9BgRMRngG8AXwJ2A14HXAYcX0/y08zcDtgJ+CAwC7i9E3Iy86U6UNeD04F7gTvq+b8ObAPMBt4NnBQRp7Q2o3MMfANVPTt/Ct7qlCo+4FA1GLdRhZST+505M9dm5lXAJ4BzI2LnPuY9LzNXZeaLmflHYBlwYL/boOGWmc8A1wNvqV9flJl3ZObzmfl34Cbg4NZsV7HhwWwh1QFOhWnXj3FcBRwREa/tFETEfsAc4CddlvsoVd16+zjr/hXwNNVBSiMgIrYHvgh8MjNvyMz1mflcZt6cmRv0BtfldwMnAA8BZ42x2JOBK/Plny6YC1yUmU9l5mrgB8Cp3WbMzCeAX7Dx+jt0ZkrAuboejm525/bpJmALqrTbt4jYGngXcPcE168hFRHbUDUwt3UZF1Q9iO39fiNwWN1VvGM9zU1Tva2afuPVj6b6QLMM+HCjeCFwcx1m2svdhers/Z9jrDci4jiqHsVVE9p4DcKBwFbAz3udITNfoGo/Dm2Pi4i9gMN45QlUtP5+W7dl1+3TPDZSf4dR0QEnIg4B9gKuy8zbgX8BJ05kWZn5HLCWqkuw41sR8VhnAH45ziIuB+4EfjOR9Wso3Vjv98eB9wIXd5nmfKrv2Y9a5c8AN1Md+E6gOkN6Zsq2VIPQS/1oW0x1SYqI2JyqvVrcmuayiHic6oz91cCnWuNPrNe7nuogeUF9Fq7RsDOwNjOf73O+B9jw+NSxEFiWmfc1yn4NLIqI7SJib6rem21a891R16O1VJfIvtfn9gxc0QGHqlvulsxcW7++hglcpoLqJlBgV+CRRvGZmblDZwA+MMa8F1Ol4/mNLkKNvnn1ft8KOAP4fUTM6oyMiDOoGpdjM/PZLvN37rnw8lSZxq0fY7ge2CsiDgCOArYElramOT0zt6e6NLUrsEdr/DV1m7QNsA9wWkR8dBPfi6bPw8AuzYcQerQHGx6fOhbyypB8JtWly3uoen6uBe5vTfOORv39LrAsIrbqc5sGqtiAU18Smg8cHhFrImIN8Glg/4jYfwKLPB54HvhTn9vxBaobSN/nWVSZMvOFzLwBeAE4BCAiTgUWAUdlZrvh6FhGdQPqbsDy6dhWTb9u9WOcadcBN1AdlE6iCitdz+Qz807gy8Cl4yzvXqqz9bkT23oNwB+AZ6kuC/UkIjaj2sfLWuUHA6+hCs4vycxHMnNBZs7KzLdSZYGux7b66sX3gdczxmWsYdVvQhwl86galP2A/zXKr+Plmzs3byXSFzOzOS0RsRNVQPka8JXMfLjXDYiIc6m6mA/tZz6Nlvo+m+OonjRYGRELqJ5+OLI+wHSVmRkRcxt/T8v2anq16wdVT+94bc9iqpuKt6bLPRUtPwTOi4hjM3NJl3XvCRxNH/dzaLAy8/GI+DzwnYh4HrgFeA54D3Ak8FRn2rqXZx+qS+GzqI5TTScDP8vMJ5uFEfFG4LF6eB/wMeDwbttTXyo9harHZ8z2bChlZpED1VnLV7uUzwfWAD+menS8OSyvp0mq69frqLr8fgec2FrOrcBprbIjgPsbr5Mqia9rDJ8d9GfjMCn1azXVF34d8CRwF7CgHncfVYPU3O+Xt+rF3l2WuXf1lRz8+3OY0vpxxVhtTz1+M+DfwF+7LHc58JFW2eeA2+q/L2jVvQeoHi/eetCfiUPfdWgBsKI+Fq0BlgAHUYWZzj5eT3WZ6TJgj9b8W1EFmKO6LHt+XTeeAv4CHN0a3zwGPgH8uT3NKAxRvxlJkqRiFHsPjiRJmrkMOJIkqTgGHEmSVBwDjiRJKo4BR5IkFceAI0mSimPAkSRJxTHgSJKk4kz7TzWs3HfOCoA5q1YeMN3rlmYiv3OVzucAfhbqzexFS16qM6svPNY6M2LswZEkScUx4EiSpOIYcCRJUnEMOJIkqTgGHEmSVBwDjiRJKo4BR5IkFceAI0mSimPAkSRJxTHgSJKk4hhwJElScQw4kiSpOAYcSZJUHAOOJEkqjgFHkiQVx4AjFWD2oiUrZi9asmLQ2yFJw8KAI0mSimPAkSRJxTHgSDPQyn3nrFi57xwvaUkqlgFHkiQVZ4tBb4CkqdG56XjpoDdEkgbAHhxJklQcA44kSSrOtFyiav5/DrvLJUnSVLMHR5IkFceAI0mSimPAkSRJxTHgSJKk4hhwJElScQw40ojxl8Ol6efPm4weA44kSSqOP9UgqWj+ZIU0M9mDI0mSimPAkSRJxTHgSJKk4hhwJElScQw4kiSpOAYcSZJUHAOOJEkqjgFHkiQVx4AjSZKK438ylgqywW/lzLtkgFsiSYNlD44kSSqOAUeSJBXHgCNJkopjwJEkScUx4EiSpOIYcCRJUnEMOJIkqTgGHEmSVBwDjiRJKo4BRxpisxctWTF70ZIVG59SktRkwJEkScUx4EiSpOIYcCRJUnEMOJIkqThbDHoDJE2P5s3KSwe5IZI0DQw40ohaue+cl5+umnfJpC1vzqqVB2zywiRpwLxEJUmSimPAkSRJxTHgSJKk4hhwJElScbzJWJKGWPPpt9UXHusN4FKPigk4nUbABkDSMBvvaTXDjDR5igk4vejWeNigaFiMVxcn+5FwjTZP6GYu/51D70Y64IzX6BtcJA1as42a6gNSt9BjENJMNtIBZ6rYKEjTe3CeCTrtiv9FWsOspB6i4gLOS42y3fiShtigepk9gZuZup2wlH4SM5IBZxBnQjYKkqZKtxOzXss2lZfzh1evvSnt6UoPLr0ayYDTr+lqKKReDctBpdsPcHopZdMN8jOcjBvSh6V+qjLe97RZNt68m1IXewlawxiqZkTA6VUvoae5E4+pp1t649mMVTYsO7oUo3p9uN8eQAP46BnEk269rnMyQ8/GbmK2t3vT9Bpcxpt3Y/MNW+jptqzJaOsjMyc6ryRJ0lDypxokSVJxDDiSJKk4BhxJklQcA44kSSqOAUeSJBXHgCNJkopjwJEkScUx4EiSpOIYcCRJUnEMOJIkqTgGHEmSVBwDjiRJKo4BR5IkFceAI0mSimPAkSRJxTHgSJKk4hhwJElScQw4kiSpOAYcSZJUHAOOJEkqjgFHkiQVx4AjSZKKY8CRJEnF+T8u9Be0pMYEQQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 576x129.6 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    FCGR3A | type 2 with mean 0.16\n",
      "    FCGR3A | type 5 with mean 2.49\n",
      "      FCN1 | type 2 with mean 3.02\n",
      "      FCN1 | type 5 with mean 2.18\n",
      "      FTH1 | type 2 with mean 5.31\n",
      "      FTH1 | type 5 with mean 5.64\n",
      "      GPX1 | type 2 with mean 2.67\n",
      "      GPX1 | type 5 with mean 1.30\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x129.6 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     GSTP1 | type 2 with mean 2.67\n",
      "     GSTP1 | type 5 with mean 1.59\n",
      "    IFITM2 | type 2 with mean 1.43\n",
      "    IFITM2 | type 5 with mean 3.25\n",
      "      LST1 | type 2 with mean 2.85\n",
      "      LST1 | type 5 with mean 3.74\n",
      "       LYZ | type 2 with mean 5.14\n",
      "       LYZ | type 5 with mean 3.39\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x129.6 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    MALAT1 | type 2 with mean 4.51\n",
      "    MALAT1 | type 5 with mean 4.87\n",
      "    MS4A6A | type 2 with mean 1.57\n",
      "    MS4A6A | type 5 with mean 0.41\n",
      "      RHOC | type 2 with mean 0.22\n",
      "      RHOC | type 5 with mean 1.65\n",
      "     RPS19 | type 2 with mean 3.56\n",
      "     RPS19 | type 5 with mean 4.42\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x129.6 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    S100A8 | type 2 with mean 4.07\n",
      "    S100A8 | type 5 with mean 1.30\n",
      "    S100A9 | type 2 with mean 4.67\n",
      "    S100A9 | type 5 with mean 2.37\n",
      "    TMSB4X | type 2 with mean 5.45\n",
      "    TMSB4X | type 5 with mean 5.65\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.6/site-packages/ipykernel_launcher.py:189: RuntimeWarning: divide by zero encountered in log10\n",
      "/opt/conda/lib/python3.6/site-packages/ipykernel_launcher.py:189: RuntimeWarning: invalid value encountered in log10\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Welch's $t$\n",
      "       LYZ\tpDE: 24.21\tpTN: 76.24\n",
      "    FCGR3A\tpDE: 22.10\tpTN: 152.40\n",
      "    S100A9\tpDE: 19.38\tpTN: 70.59\n",
      "    IFITM2\tpDE: 18.60\tpTN: 43.60\n",
      "      GPX1\tpDE: 17.89\tpTN: 37.33\n",
      "    S100A8\tpDE: 15.53\tpTN: 82.35\n",
      "     BLVRB\tpDE: 14.68\tpTN: 12.26\n",
      "    MS4A6A\tpDE: 14.19\tpTN: 24.11\n",
      "     ALDH2\tpDE: 13.94\tpTN: 4.90\n",
      "     GSTP1\tpDE: 13.35\tpTN: 28.54\n",
      " \n",
      "wilcox\n",
      "       LYZ\tpDE: 26.51\tpTN: 76.24\n",
      "    FCGR3A\tpDE: 26.34\tpTN: 152.40\n",
      "    IFITM2\tpDE: 20.34\tpTN: 43.60\n",
      "    S100A9\tpDE: 20.07\tpTN: 70.59\n",
      "      GPX1\tpDE: 18.87\tpTN: 37.33\n",
      "     RPS19\tpDE: 18.15\tpTN: 30.40\n",
      "      RHOC\tpDE: 15.33\tpTN: 47.52\n",
      "    S100A8\tpDE: 14.57\tpTN: 82.35\n",
      "     CD79B\tpDE: 14.49\tpTN: 22.88\n",
      "     GSTP1\tpDE: 14.22\tpTN: 28.54\n",
      " \n",
      "bimod\n",
      "       LYZ\tpDE: 47.07\tpTN: 76.24\n",
      "    FCGR3A\tpDE: 29.80\tpTN: 152.40\n",
      "    IFITM2\tpDE: 24.82\tpTN: 43.60\n",
      "    S100A9\tpDE: 23.44\tpTN: 70.59\n",
      "      GPX1\tpDE: 21.24\tpTN: 37.33\n",
      "    S100A8\tpDE: 20.74\tpTN: 82.35\n",
      "      RHOC\tpDE: 16.02\tpTN: 47.52\n",
      "     GSTP1\tpDE: 15.99\tpTN: 28.54\n",
      "     RPS19\tpDE: 15.95\tpTN: 30.40\n",
      "     CD79B\tpDE: 15.27\tpTN: 22.88\n",
      " \n",
      "tobit\n",
      "       LYZ\tpDE: 39.93\tpTN: 76.24\n",
      "    FCGR3A\tpDE: 29.52\tpTN: 152.40\n",
      "    S100A9\tpDE: 25.66\tpTN: 70.59\n",
      "      GPX1\tpDE: 24.09\tpTN: 37.33\n",
      "    S100A8\tpDE: 21.52\tpTN: 82.35\n",
      "    IFITM2\tpDE: 20.24\tpTN: 43.60\n",
      "     GSTP1\tpDE: 17.95\tpTN: 28.54\n",
      "      RHOC\tpDE: 16.05\tpTN: 47.52\n",
      "      FCN1\tpDE: 14.87\tpTN: 17.39\n",
      "     CD79B\tpDE: 14.63\tpTN: 22.88\n",
      " \n",
      "poisson\n",
      "    S100A9\tpDE: inf\tpTN: 70.59\n",
      "       LYZ\tpDE: inf\tpTN: 76.24\n",
      "    S100A8\tpDE: 160.66\tpTN: 82.35\n",
      "     RPS19\tpDE: 159.50\tpTN: 30.40\n",
      "    TMSB4X\tpDE: 155.75\tpTN: 5.41\n",
      "      FTH1\tpDE: 149.48\tpTN: 8.39\n",
      "    MALAT1\tpDE: 139.13\tpTN: 11.26\n",
      "    FCGR3A\tpDE: 122.49\tpTN: 152.40\n",
      "    IFITM2\tpDE: 121.18\tpTN: 43.60\n",
      "       B2M\tpDE: 110.10\tpTN: 13.92\n",
      " \n",
      "negbinom\n",
      "       LYZ\tpDE: 54.74\tpTN: 76.24\n",
      "    S100A9\tpDE: 39.13\tpTN: 70.59\n",
      "    FCGR3A\tpDE: 29.25\tpTN: 152.40\n",
      "      GPX1\tpDE: 22.06\tpTN: 37.33\n",
      "    S100A8\tpDE: 21.87\tpTN: 82.35\n",
      "    IFITM2\tpDE: 19.30\tpTN: 43.60\n",
      "     RPS19\tpDE: 18.46\tpTN: 30.40\n",
      "      RHOC\tpDE: 16.39\tpTN: 47.52\n",
      "      LST1\tpDE: 15.65\tpTN: 27.51\n",
      "     GSTP1\tpDE: 14.96\tpTN: 28.54\n",
      " \n",
      "MAST\n",
      "       LYZ\tpDE: 38.64\tpTN: 76.24\n",
      "    FCGR3A\tpDE: 30.02\tpTN: 152.40\n",
      "    IFITM2\tpDE: 24.74\tpTN: 43.60\n",
      "    S100A9\tpDE: 23.87\tpTN: 70.59\n",
      "      GPX1\tpDE: 21.06\tpTN: 37.33\n",
      "    S100A8\tpDE: 19.17\tpTN: 82.35\n",
      "     RPS19\tpDE: 16.24\tpTN: 30.40\n",
      "     GSTP1\tpDE: 15.66\tpTN: 28.54\n",
      "      RHOC\tpDE: 15.46\tpTN: 47.52\n",
      "    MS4A6A\tpDE: 14.03\tpTN: 24.11\n",
      " \n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x129.6 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAOUAAACTCAYAAABxst7KAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAHUNJREFUeJztnXmYVMXVh9/DIoiAgChBkQyuEeMICoYEcN9jXFCIC+JujLhiYtDkE0yi0bgGcAmCKy5oQNwVQRYxuIDAgGIUBWHQAIIgI2qA+X1/1B3oaXq5vUzPpafe5+mn7617T1Xd7j5dVadOnTJJeDye6FCvtivg8Xiq45XS44kYXik9nojhldLjiRheKT2eiOGV0uOJGF4pPZ6I4ZXS44kYXik9nojRoLYrkE9at26tkpKS2q6Gx5OQmTNnfiVpx3T3FZVSlpSUMGPGjNquhseTEDP7PMx9vvvq8UQMr5QeT8Qoqu5rlCgZ+FLC9EW3/LLANfFsbRS9Uq5fv57y8nK+//77gpb7wIltE6bPnz8/dB7lX3+3RVq7lttmXadkNG7cmHbt2tGwYcO85+3JnIIppZk9CJwALJf00yBtMHARsCK47XpJLwfXrgMuADYCV0h6LZtyy8vLadasGSUlJZhZjk8RnvXlqxOm79OuRU55ZCIfBkmsXLmS8vJyOnTokNe8PdlRyJbyYWAY8Ghc+l2Sbo9NMLOOwOnAvsDOwAQz20vSxkwL/f777wuukFGhLMkfQ2mMYpsZO+ywAytWrEh4r6fwFEwpJU01s5KQt58EPCXpB2ChmS0ADgKmZ1N2XVTITNjqPp/B2ydIW1P4etQQUbC+XmZmZWb2oJm1DNJ2AZbE3FMepG2BmV1sZjPMbIb/t/cUA7Vt6LkP+Aug4P0O4PxMMpA0HBgO0KVLl7QBh5JZRbOlJqypF154IQMGDIDmO+c970zwFuTaoVaVUtKyqmMzewB4MThdCuwac2u7IK1OMGLECCD5mNBT3NRq99XMYucNTgHmBcfPA6ebWSMz6wDsCbxb6Prli6VLFnPSoQdx3eUXsc8++3Daaaexbt06Jk6cSOfOndlvv/04//zz+eGHHwA49NBDmTFjBhs3buT/rr6UXkf8nFOP/AWPPXAvAEOGDKFjx46UlpZy+umnA7Bq1SpOPvlkSktL6datG2VlZQDcd+ct3HDNZVzQ+wSO796Jxx/8Z+18CJ7QFEwpzexJnKFmbzMrN7MLgL+b2VwzKwMOA64GkPQB8DTwIfAq0D8by2uUWPTpJ/TpdwHz58+nefPm3HnnnZx77rmMHj2auXPnsmHDBu67775qMv/5YC7Ll33J2InTGTPh35zU5ywAbrnlFmbNmkVZWRn3338/AIMGDaJz586UlZVx8803069fv5iyP+a+UWN4/IWJ/POuW1m/fn3hHtyTMQVTSklnSGorqaGkdpJGSjpb0n6SSiWdKOnLmPtvkrS7pL0lvVKoetYUP9p5Fzp37QZA3759mThxIh06dGCvvfYC4JxzzmHq1KnVZNq1L6H880X87f+u5a1JE2jarBkApaWlnHXWWYwaNYoGDdwIZNq0aZx99tkAHH744axcuZKKtd8A0PPwo9mmUSNattqBVq13ZNmyZXiiS20beuoM8dMOLVq0YOXKlSllmrdowTPj3+TfU97gmVEP8dqL4xg3ehQvvfQSU6dO5YUXXuCmm25i7ty5KfPZZptGm47r1avHhg0bsn+QDElkLPKGotREYUqkTvDl0nLmzHTD4ieeeIIuXbqwaNEiFixYAMBjjz3GIYccUk3m61Urqays5MjjT6T/7//IR/PmUFlZyZIlSzjssMO49dZbWbNmDRUVFfTs2ZPHH38cgMmTJ9O6dWuaNmte2If05IU611LW1r90ye578tQjI7h54JV07NiRIUOG0K1bN3r37s2GDRvo2rUrl1xySTWZ5f/9ghuuuQxVVgJwxcAb2LhxI3379mXNmjVI4oorrqBFixYMHjyY888/n9LSUpo0acIjjzxSG4/pyQN1Tilri/r16/O3IcOrubgdccQRzJo1a4t7J0+eDMA25asZ/cqUatcaNmzItGnTtpBp1aoV48aNq5ZWVr6a3w4YWC1t7MTplOTZf9aTX3z31eOJGF4pC8Auu7Zn7MSs3HY9dRCvlB5PxPBK6fFEDG/o8USahPOcjXOTh2jPlfqW0uOJGHWvpUy0QDan/FIvrl399SouPv0kAL5asZxGDRuw444uHu+cOXMYMGAAd9xxBwC33347FRUVDB48uFoe702fRsOGDenU5WcZV2/pksXMmfEOx5/SO2NZT+2QcUtpZtuZWf2aqEwx0qJlK55+7U2efu1Nevc9j6uvvprZs2cze/ZsGjVqxNixY/nqq69S5jFj+jTmzMhukcwX5Yt5+bl/ZSXrqR3SKqWZ1TOzM83sJTNbDnwEfGlmH5rZbWa2R81Xszhp0KABF198MXfddVfSe5YuWcwzox7isRH30eeYnrz55pusWLGCU089la5du9K1a1feeustAKZMmUKnTp3o1KkTnTt35tuKtfzjbzcy693p9Dmm56alX55oE6b7OgmYAFwHzJNUCWBmrXDLrW41s2cljaq5ahYv/fv3p7S0lGuvvTbh9V12bU/vvufRpMl2nHPJ5ZS2a8GZZ57J1VdfTY8ePVi8eDHHHHMM8+fP5/bbb+eee+6he/fuVFRU8PFX33PldYN45J9DGfbw6AI/mSdbwijlkZK2WIAnaRUwBhhjZj5gaJY0b96cfv36MWTIELbdNlxM1wkTJvDhhx9uOv/mm2+oqKige/fuDBgwgLPOOotevXrRoEHTmqp2naE2rLdpu6+JFDKbe4LAWMvNbF5MWisze93MPgneWwbpZmZDzGxBEFTrgHT5b81cddVVjBw5km+//TbU/ZWVlbz99tubxqZLly6ladOmDBw4kBEjRvDdd9/RvXt3Fi74uIZr7qkJQht6zOw4M3vHzP5jZk+b2c8zLOth4Ni4tIHAREl7AhODc4DjcCFA9gQuxgXYKlpatWpFnz59GDlyZMLrTbZryrffVmw6P/rooxk6dOim89mzZwPw6aefst9++/GHP/yBrl27snDBJ2y3XVPWVVRskacnumQyJXIv0BcXouNA4DYzu0fSk2GEk8R9PQk4NDh+BJgM/CFIf1SSgLfNrIWZtY2NTJA1EY0Pes011zBs2LCE1w456lh+95tzmDz+ZR64/16GDBmyaSy6YcMGDj74YO6//37uvvtuJk2aRL169dh3333pcdiRWL161Ktfn95H9+DE3mdy9kWXFvjJPJmSiVIul/RWcDzBzKYD7wChlDIJbWIU7b9Am+A4WdzX3JWyFvntgIHVlm5VxLRgbdq0Yd26dQnlSnbbg3+97j76KvnRo7c03MS2nrA5Gt6I0c/nVnFPQclknnKhmf3VzLYJztcDeYsrEbSKaeO2xuODMXuKjUyUshIXBnKJmU0DFgCTzWzPHMpfVhVmMnhfHqSHjvsqabikLpK6VHnKeDxbM6GVUtKZkvYF2gNXAoMBAx4ws8VZlv88cE5wfA7wXEx6v8AK2w1Yk8t40jXCnmT4zydapB1Tmpkp5lsLNt2ZGbyq7gnjGfQkzqjT2szKgUHALcDTQQzYz4E+we0vA8fjWuN1wHkhn2cLGjduzMqVK9lhhx22vo1sCkDVVniNG2ew9MJTo4Ty6DGzMcBzkja1iMHYsgeuhZuEm/JIiqQzklw6IsG9AvqHqFta2rVrR3l5ecG3eluWYMNXgPlrw2/6miiPXOUT5VG1aawnGoRRymNxm+48aWa7AV8D2+K6vuOBuyVtGf0pIjRs2LBWNkM9Lg+eIInyyFU+0zw8hSetUkr6HjdHeW/gTtca+E6S333G46kBQs9TmtmAuHOANcBMSbPzXC+Pp86SifNAl+D1QnB+AlAGXGJmz0j6e74rV2wsanyms1nHE1EvI0/tkIlStgMOkFQBYGaDgJeAg3GWWK+UHk8eyMR5YCfgh5jz9Tg3ue/i0j0eTw5k0lI+DrxjZs/hnAZOAJ4ws+1wTurFT7L4Pr776ckjoZVS0l/M7BWge5B0iaQZwfFZea+Zx1NHyTSa3XqcD6yCY4/Hk2cymRK5ErgIFwLEgFFmNlzS0NSSEcJ3Pz1bAZm0lBcAP5P0LYCZ3QpMB7YepfR4tgIysb4asDHmfGOQ5vF48kgmLeVDOOvrs8H5yUDioDIejydrMrG+3mlmU9hsfT0vyo7oHs/WSkbWV0nV1lF6PJ78E2aR81oSx84x3NLH5rlWwswWAWtx49QNkroEEdhHAyXAIqCPpK9zLcvjiTphgjE3k9Q89gXsXZWex7ocJqmTpC7BebKYsB5PUZPt/pSJV8/ml5NwsWAJ3k8uQJkeT62TrVLmeypEwHgzm2lmFwdpyWLCejxFTbabxj6Q11pAD0lLzWwn4HUz+yj2oiSZWcKQa4ESXwzQvn37PFfL4yk8WbWUkvK60aGkpcH7cuBZ4CCSx4SNl/VxXz1FRbbd17wR7AzdrOoYOBqYR/KYsB5PUZNt9zWftAGeDWL+NACekPSqmb1H4piwHk9ORD0sS60rpaTPgP0TpK8kQUxYj6fYqfXuq8fjqY5XSo8nYnil9HgihldKjydi1Lqhx5MhiUKaRMRq6MkPvqX0eCKGV0qPJ2J4pfR4IoZXSo8nYnil9HgihldKjydieKX0eCKGn6f0eDKkpleZeKX0FJSoL5uKAr776vFEjMi3lGZ2LPAPoD4wQtIttVylOo1v6WqeSCulmdUH7gGOAsqB98zseUl1Y+foGsArVfSJevf1IGCBpM8k/Q94ChcP1uMpWqKulLsAS2LOy4M0j6doMSlhONVIYGanAcdKujA4Pxu3ce1lMfdsivsK7A38J0TWrYGvcqhabctHoQ7+GTKX/7GktHFQIz2mBJYCu8actwvSNiFpODA8k0zNbEbMniUZU9vyUaiDf4b8fAaJiHr39T1gTzPrYGbbAKfj4sF6PEVLpFtKSRvM7DLgNdyUyIOSPqjlank8NUqklRJA0svAy3nONqPubgTlo1AH/wz5+Qy2INKGHo+nLhL1MaXHU+fwSunxRAyvlEWOmSXdtNPMehayLp5weKWsYcxsGzPrZ2ZHBudnmtkwM+tvZg2zyK+lmTXPQGSymV0b+BFX5dHGzEYBd2Vafk1gZkflKF9Uu3zXCaU0s93M7Hdm9g8zu9PMLgn7wzazN83skCTXxofI4iHgl8CVZvYY0Bt4B+gKjAhZh53N7FEzW4PzIJlnZovNbHAIxT4Q2B2YbWaHm9mVwLvAdJxvcU6YWT4skCOzKLeFmV1gZhOBWSFljglkSuLSzw8hu9bMvol5rY19z7T+KcsqduurmV0BnABMBY7HfYGrgVOASyVNTiP/BfANMA74k6QNMddmSeqcRr5MUqmZNcB5I+0saaO5DTnnSCoN8QxvAH+WNNnMegE9gT8B1wE7Sbo4ZQYujytxLeMXQDdJ5elkYmRbJbsUPEO7EHkkc/ow4HBJ24XIY1vcgoQzgc5AM+BkYKqkyjSyNwM9gPeBXwF3SxoaXHtf0gFp5McBPwLGAk9JWpyuvlkjqahfwFygfnDcBJgcHLcHZoWQfx9oCjyM8zDaI/ZaCPl5wDZAS2At0CpIbwzMD/kMc+LOZ8Ycf5RGtgXwT2A2bpfsu4PP5PAMPsONwGfAwphX1fn/QubxNa7HcEjc61BgWQj5J3CLE0bilvLVBxZm+DtoEPOZvAzcFZyn/R0E920PnIdzZpkCXFr1febzFXnngTzRAPfDaoRTMCQtDjumk1QBnGtmvYFJZnajpBG4f/l0jAQ+wv2I/gg8Y2afAd2A0SHrv8LM+gKTgF7AIoCgtU03BHkfuBfoL9fKjzezTsC9Zva5pDNClP8ZcIQStA5mtiTB/Yl4G1gnaUqCPMIsIuiIU+z5uD+zjWaWSTevQfD8SFptZr8ChpvZM7g/zbRIWgM8ZGaP4Fw+h+D+XO/MoB6hCirqF3AlUAY8gFOO84L0HXHdnnTy78ed74pTjrHAxyHrsDOu2wruX/o04KAMnqE98DSu1R0FtA3SdwBOTSPbLsW1i0KW3x/YP8m1ywv4Xf4EuDH4HqcBK4A2IWVfBA5JkP5XoDJkHr8AhuJ6HcOAnjXxnEU/pgQws32BfYB5kj7KUPZ1SUfFpRkwEBgsqVEW9dkO1+KdLumXmcrnipldKuneApc5XtLReczvQNzYsjdQLukXae7fFkDSdwmu7SJp6ZZS1e5ZhLNFPAW8AWyIvS7p/Uzqn7KsuqCU8WSiFGbWVdJ7Sa6ZQn6AwSqXX+J+SMcAY4Cxkl7IqPJb5nuDpD+nuD4gPglnILoZQFLarldgXEqKpLEh8khrFMuG4A+yp6Spae5LachJp1RmNhmo+q5F9aGLJB2evrbhqCtjymRKcX8I0RFmNg24TlI103cYhTSzo4EzcEaWScCjQFdJ52X2BEm5EEiqlLju3svAB2z+IdXHWS7D8qu449g/EuG68unYPpVyp1NsM2sCXBaUNxQ3puuFG2P+JUT5M3Dd/6pFydWUCkipVJIODVFGXij6ljKBUowGhkoqCSlfH7gaF91gkKQnMyy/EngTOFfSwiDtM0m7ZZBHsnkwA7aVlPTPNfDouQNnrLlR0rpMy4/LL6sWz8xWAs+R2DgmSSnnCs3saZz1dVtchIn5uO/yROBHks5OI38Vbiy/BtcFfVbOgJfJMzQHdpT0aVx6qaSyTPJKSaEG6bX1Aipx5usOMWmfZZFPR9wXuhpYhbMErgoh1wm4BfgUeB24APg8w7IXk8SgASwJmcdJwFu4H2bGzx+TT9ppoHzKxcjPDt4N+C+bGxQDyjLIZzfgepwDx9NAp5ByfXBzvLNxvY6u+Xq2+Fdd8Og5AOe9MsHMXjezC3Ddt9CY2Tm4f/lBOKvtjrj4LGnjrUiaLWmgpN0D+U5AQzN7xVx8oTA8Cvw4ybUnwmQg6Tlcb+FnuABkhSbM9FFa5LTg5eC96jx0d0/SZ7jvcjzOo2mvkKLXAwdK6oSbq3zMzE4JruXl2aoo+u5rLGb2C1xX9lRgDq4Lk9JNzMzeBL4ErpL0RZ7qUQ84EmdoSuviVduY2Qts/uEfjPOO2oSkE0Pk8VNJ8xKk9wDOkNQ/jfwI3HdQEZe+O/CIpB5p5HfDjUNPwnWDnwJeUgJrbBL5uZL2izlvi5tmeQQ3NElpSMqEOqWUVWSiFGZ2rKRXk1y7StLdIcrbHjiWzeExlwKvSVodsr4/BlbLTV5jZofh3Ms+B4bJxcRNJrsnzmlhFW6S+wGcm96nwIVKYlmOyyOh728VSuAQkCa/zmyezliIs0IPzSSPTAnG9mW4VvIb4lpXpbFCm9m/gbMVM540s2Y498seymJqLGlZdUEpc1WKFPkulpR0aVRwTz9ct3U8myPxtcO5it0o6dEQ5bwDnCLpi8AbZwLwN6AUWK8gBGcS2Wm47m9znMHqKpz1tCfwV0k/C1H+w5LOTXdfmjz2wvVSzsBZQEcDv5OUrFseNt+jgGsVN5ec4L5Bqa5LujGN/P44j6RP4tIbAn0kPR6yyunJ5wA1ii+gH65VuA/nxP0n3FTIp0C/HPNOa2TBxaFtkSC9JeE9gspijm8H/h4c1yONkYPAQBIcL0h2LU0eORsy2Gxwi/UdDm1wwk1ZfAxU4Lya9sNNc8wEeoWQv6wGflutCRq2fL7qwjzlH3ED9Gqtopm1xFng0rZUKQjTzbAk91US3kAQe9/huMl/JFW6ufOUxK6eiJ9aSbmyIoYmQZczYWEK583SCzemm2Rmr+LGdJkYSO7ATUtNB44L3gdKGhZS/nyca1xWmFk3nBV9FW5e9DGcUtYzs35KMsTJhrqglDkphZmtTSJvuDmzdNwEvG9u7WWV83Z7XPc1zKQ3wBvBPN2XuBb2jaBubYGk48mAn5hZWVDf3YPjqvqHnavcBacUCecYSTPxDiBpHDAu8KY6CdeN3snM7sMZ3NKtTZU2L7MbZ2ZLM1DIfDAMZ4HdHvf5HyfpbTP7CfAkkDelLPoxZTCdcQNuTLeFUkh6uAB1aInzIoof034dUt6AXwNtgacV+GkGrddOkl5LIZtyzCbp8xDl15SLXEucsefXko5Ic+9nwO9ikm4Dfl91ovQeQRuAdYkuOXGlXPRuZrPlpkMws/mS9om5ltfPp+iVEnJXijzVoU1s+ZKWZSh/MrAHMDeVEobIZwfctMZiSTNDytSIUsbkH8Zg9lCKy1J6K3pOz2AxC6EtblF0/Hmu1IXuK5K+NrNJVFeKgihkYC29H9ftKcf9M7czs9W4yAdpx2NBF68j8G/gL2Z2kKRQXV8zexE39poXdHffxxlIdjez4QoxpYMzktUqyp+vcLbsH7g7GrBtjOuj4dZU5o98W46i9sJ50LyN85V8HTed8FGQdkAByp+N2yksPr0bcREFUuQxj+rRE2ZmUP4HMcfXA48Gx80I6Z5GjPUVGFMDn9HiEPe0w80HVp0PwA1LbiDGoptC/vpC//ayfdWFlvJh4DeS3olNDKxpDwH713D528WXDSBnJEgblybgf5I2BnLrLITJNYb1McdH4JwHkLQ2mFAPQ2x52Tqyxy8hi827aYgsbgNi5wJ/g9s2oAluJcxZaeTbmtmQZBclXRGiDgWhLihlPpQiF14xs5dwUy9VhqZdcfOnYS12VRZUqG5FrTJSpAq+tcTMLsd1nQ+oKtPcot+wIS6V5DgTUi0V+0cI+b0lvRhzvk7SHbDJFTIdM0LcEwmK3tAT/DvuTmKlWKiYDWhrsA7H4aYBYg1Nz8ttXhRGPmsLqpnthFtv2Ra4R8HUQ+Cqd6Ck20OUvxH4ls3TQFVWzFCWyxD5J11IHnPPh5I6xpy3krQqOK5mDU0ivylGT9QpeqWE3JUiqgQ+vGcony5eBcLMOrLZ7W610my+Grgani3p47j0n+DGySlj2MZZT4dKujynB6hB6kL3FUmvAK/Udj3iCayfYWK2NscFr9oFt2nu67hV+NfgVrskVUpLHm8VCLfCI1+YC4JcpYjrccvRukhaFEJ8EPCimd2EsyCDCzR9PS44WtriY467h6tx7VAnWspkhFWKHMvIRyDj53CLqqfjjDU7BfJXSpqdRnYFrtv+JM6tsJqRSBmu8MgWM5uOc4p/ChfM+BMzWyipQwZ5/BS4Ftg3SJoH3KYES8ISyCadZ4waRd9SplGK4wtQhRW4JVbxMWEMp1xh2E3BWr5gXeGXQHtJ34eQ/RHOe+kM3HKpl4AnVfgdsZfhWvo2uMXhn5CB0cjMGuOCNveLS9/RzBqH+CxSuRumM5YVlKJXSvKjFLmQj0DGm6Y15IIQl4dUSIKplFeBV82sEU45J5sLKF0w31FJJwdL6HoBg4N1ni0CR4h3Q2QxBPcc8e50PXARFX6bRj6lIShKFH331cw+IYVSSNq1hsvvD0yTNCfBtcsVYnFvjPUTqltAw/ptNsJF8jsDKMGNSx9UmlinNUngdtgHt3KkfbrvwcxmSjowybUPJO2b6NrWSF1oKe/GraxItCHL3wtQ/ru4rhuwadHzqbjWe3CYDCRlFFMoFjN7FPgpLszkjWHGX4VAzvd3KDA03ZRPQJMU19LGmkqz2ifnaZ18Uhdayq64xcj/Dc6rKUXVXFcNlv8+cKSkVWZ2MM7QcTnO/W8fSafVcPmVbG5lY7/sgv4Yc7UCm9kU4PfxXd3g+71D0sG51zIa1AWlrG2lmCNp/+D4HmCFpMHB+ablQDVYfo2u8MigHjlZgc3sIFxIyIdx0QYAuuCcQE5P5LW1tVIXuq/1Y1rDXwPDJY0BxphZyumEfJUf401yBG71fBWF+Pyj8q+bkxVY0ruBYvYHzg2SP8A5+y/Pf3VrjzqhlLWsFE8CU8zsK+A7XLR0zGwPXHDnmmanFM7gKMReIvkgVyuwmbUPjHUpA2AVA3VBKWtVKSTdZG4L8LbAeG0eL9TDdaNrmvq4VRh5DRicDQmswEOAZ0OKj8M51GNmYySdWhN1jAJFP6aETcu0qpTi2yBtL6Cp8riFWRSJivdKnBX4qUytwLFj46iMk2uKOqGUdZmo/IBztQJvTW5yueKVssiJXeK0NVPTy8eihFdKjydi1IVdtzyerQqvlB5PxPBK6fFEDK+UHk/E8Erp8USM/wfFvTi80OsGlAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x108 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "np.random.seed(0)\n",
    "ident0, ident1 = 2, 5\n",
    "_ = compare_clusters_multiple_runs_less_genes(ident0, ident1)\n",
    "p_tn = compare_clusters_tn_test(ident0, ident1)\n",
    "\n",
    "create_and_save_visualization(p_tn, ident0, ident1, None, ngenes=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ident0 = 0\n",
    "ident1 = 1\n",
    "ident2 = 2\n",
    "ident3 = 3\n",
    "ident4 = 5\n",
    "\n",
    "labels_ = np.array(labels).astype(str)\n",
    "labels_[~((labels == ident0) | (labels == ident1) | (labels == ident2) \n",
    "          | (labels == ident3) | (labels == ident4))] = 'Other'\n",
    "plt.scatter(tsne[labels_ == str(ident0), 0], tsne[labels_ == str(ident0), 1],\n",
    "            label=r'c%s ($n$ = %s)'%(ident0, np.sum(labels_== str(ident0))), c='g', s=10)\n",
    "plt.scatter(tsne[labels_ == str(ident1), 0], tsne[labels_ == str(ident1), 1],\n",
    "            label=r'c%s ($n$ = %s)'%(ident1, np.sum(labels_== str(ident1))), c='r', s=10)\n",
    "plt.scatter(tsne[labels_ == str(ident2), 0], tsne[labels_ == str(ident2), 1],\n",
    "            label=r'c%s ($n$ = %s)'%(ident2, np.sum(labels_== str(ident2))), c='teal', s=10)   \n",
    "plt.scatter(tsne[labels_ == str(ident3), 0], tsne[labels_ == str(ident3), 1],\n",
    "            label=r'c%s ($n$ = %s)'%(ident3, np.sum(labels_== str(ident3))), c='purple', s=10)    \n",
    "plt.scatter(tsne[labels_ == str(ident4), 0], tsne[labels_ == str(ident4), 1],\n",
    "            label=r'c%s ($n$ = %s)'%(ident4, np.sum(labels_== str(ident4))), c='brown', s=10)    \n",
    "plt.scatter(tsne[labels_ == 'Other', 0], tsne[labels_ == 'Other', 1],\n",
    "            label=r'Other ($n$ = %s)'%(np.sum(labels_=='Other')), c='k', s=10)\n",
    "plt.legend(bbox_to_anchor=(1.4, 1))\n",
    "plt.axis('equal')\n",
    "plt.axis('off')\n",
    "plt.savefig('figures/pbmc_tsne_supp.pdf',\n",
    "            format='pdf', dpi=500, bbox_inches='tight')\n",
    "plt.show()    "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}