10.convert-to-matrices.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Convert STRING PPI scores from table to matrix for each evidence channel\n",
    "\n",
    "Information on STRING datasets at\n",
    "\n",
    "- https://string-db.org/cgi/help.pl\n",
    "- https://string-db.org/cgi/download.pl?species_text=Homo+sapiens"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import scipy.sparse\n",
    "from IPython.display import display\n",
    "\n",
    "from utils import get_string_dataset_path, get_protein_info_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Download and examine STRING datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>index</th>\n",
       "      <th>protein_external_id</th>\n",
       "      <th>preferred_name</th>\n",
       "      <th>protein_size</th>\n",
       "      <th>annotation</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>19564</th>\n",
       "      <td>19564</td>\n",
       "      <td>9606.ENSP00000485675</td>\n",
       "      <td>ENSG00000280116</td>\n",
       "      <td>84</td>\n",
       "      <td>annotation not available</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19565</th>\n",
       "      <td>19565</td>\n",
       "      <td>9606.ENSP00000485678</td>\n",
       "      <td>OR6Q1</td>\n",
       "      <td>317</td>\n",
       "      <td>Olfactory receptor 6Q1; Odorant receptor; Olfa...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       index   protein_external_id   preferred_name  protein_size  \\\n",
       "19564  19564  9606.ENSP00000485675  ENSG00000280116            84   \n",
       "19565  19565  9606.ENSP00000485678            OR6Q1           317   \n",
       "\n",
       "                                              annotation  \n",
       "19564                           annotation not available  \n",
       "19565  Olfactory receptor 6Q1; Odorant receptor; Olfa...  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Show the head of protein info\n",
    "gene_df = get_protein_info_df()\n",
    "protein_to_index = gene_df.set_index('protein_external_id')['index'].to_dict()\n",
    "gene_df.tail(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>protein1</th>\n",
       "      <th>protein2</th>\n",
       "      <th>neighborhood</th>\n",
       "      <th>neighborhood_transferred</th>\n",
       "      <th>fusion</th>\n",
       "      <th>cooccurence</th>\n",
       "      <th>homology</th>\n",
       "      <th>coexpression</th>\n",
       "      <th>coexpression_transferred</th>\n",
       "      <th>experiments</th>\n",
       "      <th>experiments_transferred</th>\n",
       "      <th>database</th>\n",
       "      <th>database_transferred</th>\n",
       "      <th>textmining</th>\n",
       "      <th>textmining_transferred</th>\n",
       "      <th>combined_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>9606.ENSP00000000233</td>\n",
       "      <td>9606.ENSP00000272298</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>332</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>62</td>\n",
       "      <td>0</td>\n",
       "      <td>181</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>125</td>\n",
       "      <td>490</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9606.ENSP00000000233</td>\n",
       "      <td>9606.ENSP00000253401</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>186</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>56</td>\n",
       "      <td>198</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               protein1              protein2  neighborhood  \\\n",
       "0  9606.ENSP00000000233  9606.ENSP00000272298             0   \n",
       "1  9606.ENSP00000000233  9606.ENSP00000253401             0   \n",
       "\n",
       "   neighborhood_transferred  fusion  cooccurence  homology  coexpression  \\\n",
       "0                         0       0          332         0             0   \n",
       "1                         0       0            0         0             0   \n",
       "\n",
       "   coexpression_transferred  experiments  experiments_transferred  database  \\\n",
       "0                        62            0                      181         0   \n",
       "1                         0            0                      186         0   \n",
       "\n",
       "   database_transferred  textmining  textmining_transferred  combined_score  \n",
       "0                     0           0                     125             490  \n",
       "1                     0           0                      56             198  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Show the head of protein links\n",
    "link_path = get_string_dataset_path(\"protein.links.full\")\n",
    "link_head_df = pd.read_csv(link_path, sep=' ', nrows=2)\n",
    "link_head_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Convert protein links to matrix format\n",
    "\n",
    "Each matrix contains a single evidence channel.\n",
    "\n",
    "Construct matrices use dense numpy array for speed. Then convert to [`scipy.sparse.csr_matrix`](https://docs.scipy.org/doc/scipy/reference/generated/scipy.sparse.csr_matrix.html), so all matrices fit in memory."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def generate_link_df_itertuples():\n",
    "    link_df_generator = pd.read_csv(link_path, sep=' ', chunksize=5_000)\n",
    "    for link_df in link_df_generator:\n",
    "        yield from link_df.itertuples()\n",
    "\n",
    "def get_empty_score_array() -> np.ndarray:\n",
    "    shape = len(gene_df), len(gene_df)\n",
    "    score_array = np.zeros(shape, dtype=np.uint16)\n",
    "    return score_array\n",
    "\n",
    "def load_score_array(channel=\"combined_score\") -> np.ndarray:\n",
    "    array = get_empty_score_array()\n",
    "    for row in generate_link_df_itertuples():\n",
    "        index_1 = protein_to_index[row.protein1]\n",
    "        index_2 = protein_to_index[row.protein2]\n",
    "        array[index_1, index_2] = getattr(row, channel)\n",
    "    return array\n",
    "\n",
    "def array_to_df(array: np.ndarray):\n",
    "    return pd.DataFrame(\n",
    "        data=array,\n",
    "        columns=gene_df.preferred_name,\n",
    "        index=gene_df.preferred_name,\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "14"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "channels = list(link_head_df.columns[2:])\n",
    "len(channels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "neighborhood\n",
      "neighborhood_transferred\n",
      "fusion\n",
      "cooccurence\n",
      "homology\n",
      "coexpression\n",
      "coexpression_transferred\n",
      "experiments\n",
      "experiments_transferred\n",
      "database\n",
      "database_transferred\n",
      "textmining\n",
      "textmining_transferred\n",
      "combined_score\n",
      "CPU times: user 26min 47s, sys: 11.6 s, total: 26min 59s\n",
      "Wall time: 26min 59s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "score_arrays = {}\n",
    "for channel in channels:\n",
    "    print(channel)\n",
    "    score_array = load_score_array(channel)\n",
    "    score_array_sparse = scipy.sparse.csr_matrix(score_array)\n",
    "    score_arrays[channel] = score_array_sparse"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Information on scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def describe_array(channel: str, array: scipy.sparse.csr_matrix):\n",
    "    info = {\"channel\": channel}\n",
    "    info[\"nrows\"], info[\"ncols\"] = array.shape\n",
    "    info[\"size\"] = np.prod(array.shape)\n",
    "    info[\"n_nonzero\"] = array.nnz\n",
    "    info[\"density\"] = \"{:.3%}\".format(info[\"n_nonzero\"] / info[\"size\"])\n",
    "    info[\"n_non_symmetric\"] = (array - array.transpose()).nnz\n",
    "    if info[\"n_nonzero\"] > 0:\n",
    "        info[\"nonzero_mean\"] = array.data.mean()\n",
    "        info[\"nonzero_median\"] = np.median(array.data)\n",
    "    return info"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>name</th>\n",
       "      <th>nrows</th>\n",
       "      <th>ncols</th>\n",
       "      <th>size</th>\n",
       "      <th>n_nonzero</th>\n",
       "      <th>density</th>\n",
       "      <th>n_non_symmetric</th>\n",
       "      <th>nonzero_mean</th>\n",
       "      <th>nonzero_median</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>neighborhood</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000%</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>neighborhood_transferred</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>404222</td>\n",
       "      <td>0.106%</td>\n",
       "      <td>0</td>\n",
       "      <td>93.567302</td>\n",
       "      <td>70.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>fusion</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>23388</td>\n",
       "      <td>0.006%</td>\n",
       "      <td>0</td>\n",
       "      <td>82.749102</td>\n",
       "      <td>6.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>cooccurence</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>60404</td>\n",
       "      <td>0.016%</td>\n",
       "      <td>0</td>\n",
       "      <td>346.419409</td>\n",
       "      <td>361.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>homology</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>217710</td>\n",
       "      <td>0.057%</td>\n",
       "      <td>0</td>\n",
       "      <td>676.650379</td>\n",
       "      <td>631.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>coexpression</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>2513002</td>\n",
       "      <td>0.656%</td>\n",
       "      <td>0</td>\n",
       "      <td>131.588230</td>\n",
       "      <td>98.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>coexpression_transferred</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>5090144</td>\n",
       "      <td>1.330%</td>\n",
       "      <td>0</td>\n",
       "      <td>89.584010</td>\n",
       "      <td>62.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>experiments</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>445816</td>\n",
       "      <td>0.116%</td>\n",
       "      <td>0</td>\n",
       "      <td>341.087009</td>\n",
       "      <td>305.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>experiments_transferred</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>4296274</td>\n",
       "      <td>1.122%</td>\n",
       "      <td>0</td>\n",
       "      <td>158.216116</td>\n",
       "      <td>157.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>database</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>706440</td>\n",
       "      <td>0.185%</td>\n",
       "      <td>0</td>\n",
       "      <td>859.364192</td>\n",
       "      <td>900.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>database_transferred</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000%</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>textmining</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>6689886</td>\n",
       "      <td>1.747%</td>\n",
       "      <td>0</td>\n",
       "      <td>204.672557</td>\n",
       "      <td>168.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>textmining_transferred</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>7214278</td>\n",
       "      <td>1.884%</td>\n",
       "      <td>0</td>\n",
       "      <td>127.804802</td>\n",
       "      <td>95.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>combined_score</td>\n",
       "      <td>19566</td>\n",
       "      <td>19566</td>\n",
       "      <td>382828356</td>\n",
       "      <td>11759454</td>\n",
       "      <td>3.072%</td>\n",
       "      <td>0</td>\n",
       "      <td>295.806642</td>\n",
       "      <td>217.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        name  nrows  ncols       size  n_nonzero density  \\\n",
       "0               neighborhood  19566  19566  382828356          0  0.000%   \n",
       "1   neighborhood_transferred  19566  19566  382828356     404222  0.106%   \n",
       "2                     fusion  19566  19566  382828356      23388  0.006%   \n",
       "3                cooccurence  19566  19566  382828356      60404  0.016%   \n",
       "4                   homology  19566  19566  382828356     217710  0.057%   \n",
       "5               coexpression  19566  19566  382828356    2513002  0.656%   \n",
       "6   coexpression_transferred  19566  19566  382828356    5090144  1.330%   \n",
       "7                experiments  19566  19566  382828356     445816  0.116%   \n",
       "8    experiments_transferred  19566  19566  382828356    4296274  1.122%   \n",
       "9                   database  19566  19566  382828356     706440  0.185%   \n",
       "10      database_transferred  19566  19566  382828356          0  0.000%   \n",
       "11                textmining  19566  19566  382828356    6689886  1.747%   \n",
       "12    textmining_transferred  19566  19566  382828356    7214278  1.884%   \n",
       "13            combined_score  19566  19566  382828356   11759454  3.072%   \n",
       "\n",
       "    n_non_symmetric  nonzero_mean  nonzero_median  \n",
       "0                 0           NaN             NaN  \n",
       "1                 0     93.567302            70.0  \n",
       "2                 0     82.749102             6.0  \n",
       "3                 0    346.419409           361.0  \n",
       "4                 0    676.650379           631.0  \n",
       "5                 0    131.588230            98.0  \n",
       "6                 0     89.584010            62.0  \n",
       "7                 0    341.087009           305.0  \n",
       "8                 0    158.216116           157.0  \n",
       "9                 0    859.364192           900.0  \n",
       "10                0           NaN             NaN  \n",
       "11                0    204.672557           168.0  \n",
       "12                0    127.804802            95.0  \n",
       "13                0    295.806642           217.0  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rows = (describe_array(*args) for args in score_arrays.items())\n",
    "channel_info_df = pd.DataFrame(rows)\n",
    "channel_info_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Score distributions\n",
    "\n",
    "Plot nonzero scores for each evidence channel."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_score_distribution(array: scipy.sparse.csr_matrix, channel: str=\"\"):\n",
    "    plt.figure(figsize=(8, 2))\n",
    "    plt.hist(array.data, bins = np.arange(0, 1001, 10)) \n",
    "    plt.title(f\"score histogram: {channel}\") \n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfkAAACcCAYAAAB8+NK5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAVBUlEQVR4nO3dfZRdVX3G8e8jgRACkoTElLyDRDRSFaQQLdUUMIRAG7S+IUuSEAgIKlqXGC01FVDjqm9l1WIR0hAVhCJIBDTGYGoRgyRKIbyZAQJJyBt5AQFbQH/9Y+/Bk8udmTtzZ+beOfN81jprztn73LP32WfP/d2zz547igjMzMysfF7R6AqYmZlZz3CQNzMzKykHeTMzs5JykDczMyspB3kzM7OScpA3MzMrKQd5Kx1JKySd2UbeOEnPSNqjt+vVV0n6pqR/rHHfRZIuaSc/JB3STfWaJen27jhWDWV1W717iqRLJD0paXODym/32ltjDGh0Bcx6U0Q8Duzb0X6SZgFnRsQxPV6pJhcR5zS6Dn2BpEXAhoi4sAFljwM+AYyPiK29Xb41L9/JW1NTUqp+WsZz6m2S+twNSg/XeRywvSsBvlp/7Ivta9X5jcbqJulTkjZK+p2khyQdl9P3kPQZSQ/nvNWSxua8t0q6S9JT+edbC8dbIenzkn4BPAccLOm1kpZJ2pHLeG8H1Rov6Re53J9IGp6PPSEPvQ7I27MkPZL3e1TSaZJeB3wTeEse2t+V991f0mJJ2yQ9JunC1jfHfK5fycOlj0r6cEU51c5ptqQHctmPSDq70AZTJG2QdIGkrZI2STpF0nRJv83t8JlOXKNFkr4h6ZZc3p2SXl3Ib7N9K4dhc502SXpC0plVhrKHtlVONj2f75OS/rnQhq/IbfpYPufFkvavuG5zJD0O3Faoz5cl7cztfmIhfZSkJfmcWiSdVcgbKOnr+RyeyOsDC/mfLJzjGTW071zgNOCC3Gd+mNPXKf1+3AM8K2mApHn60+/E/ZLeWTjOLEm3t3NO1frr8cAyYFQue1Hed7KkOyTtkvQ/kqYUjlOtP4ak8yStBdbm/U6WdHc+xh2S3lA4xuGSfp3rci2wd0ftZA0QEV68dHkBDgXWA6Py9gTg1Xn9k8C9eR8BbwQOAIYBO4EPkh4ZnZq3D8ivWwE8Drw+5++fy5idtw8HngQmtVGnFcDDwGuAQXl7QaF+kY8zGHgaODTnHQi8Pq/PAm6vOO5i4CZgv3yc3wJzct45wP3AGGAo8NPWcto4pz2Bk4BX57Z5O+nN9oi8/xTgReCzed+zgG3A1bn81wO/Bw7K+x8D7GrnOi0CtgNH5fK/C3wv5w1ur33zay/J69OAzbn8fYDv5PM8pKNycn4AP8t9YFxuwzNz3hlAC3Aw6ZHKDcC3K67b4lzfQfkavZDbZg/gQ8ATgPJrfg78Gyn4vCm337E57yJgJfAqYARwB3Bx4Ry3AIflsq4unmMHbXxJRdo64G5gLDAop70HGEW6yXof8CxwYKHfVT0n2u+vU0iPClrLHZ2vw/Rczjvy9oh2+mOQPiwMy+17OLAVODrXZWY+n4HAXsBjwMfza9+d631Je23kpQHv0Y2ugJe+vQCH5DeC44E9K/IeAmZUec0HgV9VpP0SmJXXVwAXFfLeB/x3xf7/Dsxvo04rgAsL2+cCP87rE9g9yO8C/q71DbjwmlkUgnx+k3uewgcL4GxgRV6/DTi7kHc8Lw/yF1Wrb+E1PwDOz+tTSEF8j7y9Xz7e0YX9VwOn1HidFgFXFLanAw/W0r7sHuQXAl+suP6VQb5qOXk7gGkV12Z5Xl8OnFvIOzQHjgGF63ZwxTVqKWzvk/f5M1JQ/QOwXyH/i8CivP4wML2QdwKwrnCOCwp5r6G+IH9GB6+7m/x70sE5tddfp7B7kP8U+QNSIW0pMLOt/pjLObawfRn5g0/F7/TbgbdR+ECV8+6oPH8vjV88XG91iYgW4GPAPwFbJX1P0qicPZb0ZlppFOkuoOgx0t1Hq/WF9fHA0XnIcJfS8PlppDe+thRnGD9Hlcl2EfEsKcCdA2zKQ8yvbeN4w0l3LMV6F+s8qqLOxfWqaZJOlLQyDyfvIgXE4YVdtkfEH/L67/PPLYX831c7r3a01Sadad9azrOjti++5rF8zNZjV7bvAGBkO+W9VFZEPJdX983H2hERv6s4XvF6VZZVrEdlHetRed1PLwyB7yKNGBSve9Vz6mR/HQ+8p+KaHkO6+69arypp44FPVBxjLKl9RgEbI0f3rN52sh7gIG91i4irI81CH0+6G/hSzlpPGo6u9ETet2gcsLF42ML6euC/ImJIYdk3Ij7UDXVfGhHvIL35PQh8q0r5kIavX6iod7HOm0hD9a3GViuudSU///0+8GVgZEQMAW4lDcv2ts60by3n2ZHia8aR+gO8vF+MIz2yKH6wqfXfZj4BDJO0X8XxNhbyK8tqrcemKnWsRVt1K1738aQ+9mHS46khwBpqvO7t9NdK60l38sVrOjgiFnRQ38rfu89XHGOfiLiG1EajJRXrXWs7WS9ykLe6SDpU0rE5aP0v6e7yjzn7CuBiSROVvEHSAaRg9hpJH8gTkd4HTAJubqOYm/P+H5S0Z17+QmmCXD11HylphqTBwP8BzxTqvgUYI2kvgHxHfR3weUn75Tfrvyc9kybnnS9ptKQhpOHS9uxFera5DXgxT66aWs/51KEz7XsdMFvS6yTtA9T09/MVPilpqNIkzPOBa3P6NcDHJR0kaV/gC8C1EfFiZwuIiPWk4eMvSto7Txibw5+u1zXAhZJGKE3K/Cy7X8tZkiblc5xfY7FbSPMJ2jOYFEi3AUiaTbqT71AH/bXSd4C/kXSC0qTQvZUmc45pY/9qvgWcI+no/Ps7WNJJ+YPTL0kfwD6a+8u7SPMwrMk4yFu9BgILSHe6m0kTmT6d875KesP8CWnC0JWkZ4nbgZNJf9e7HbgAODkinqxWQB5ynQq8n3S3tZk0WjCw2v6d8ApSoH4C2EF61th693obcB+wWVJrvT5CmiT1CHA7aULWwpz3rXye9wC/IX2QeZH0XLitc/ooqX12Ah8AlnT1RCT9laRnuvLazrRvRPwIuJQ0ea6FNHkNUtCp1U2k+QR3A7eQ+gWktvw2acLco6QPjR/p3Nns5lTSs/wngBtJcwx+mvMuAVaRrte9wK9zWus5fp3UB1oozOTvwJXApDy0/YNqO0TE/cBXSEFyC/DnwC9qPH57/bWynPXADOAzpA8U60kTYWt+z4+IVaQJgP9K6qMtpDkDRMTzwLvy9g7SY4Qbaj229Z7WWahm1o3ynfk3I6LysUSp5Lv9NcDArtxxm1nP8p28WTeQNEjpb9gHSBpNGuK9sdH16gmS3qn0d+ZDSXf8P3SAN2tODvJm3UPA50jDmr8BHiA95y2js0l/Nvkw6XFE3RMg+wpJ9yl94Uzlclqj62ZWjYfrzczMSsp38mZmZiXlIG9mZlZSpftPQ8OHD48JEyY0uhpmZma9YvXq1U9GxIhqeaUL8hMmTGDVqlWNroaZmVmvkNTmVwp7uN7MzKykHOTNzMxKykHezMyspEr3TL7ZTJh3y27b6xac1KCamJlZf+M7eTMzs5JykDczMyspB3kzM7OScpA3MzMrKQd5MzOzknKQNzMzKykHeTMzs5JykDczMyspB3kzM7OScpA3MzMrKQd5MzOzknKQNzMzK6kOg7ykhZK2SlpTSBsmaZmktfnn0JwuSZdKapF0j6QjCq+ZmfdfK2lmIf3Nku7Nr7lUktorw8zMzGpTy538ImBaRdo8YHlETASW522AE4GJeZkLXAYpYAPzgaOBo4D5haB9GXBW4XXTOijDzMzMatBhkI+InwM7KpJnAFfl9auAUwrpiyNZCQyRdCBwArAsInZExE5gGTAt570yIlZGRACLK45VrQwzMzOrQVefyY+MiE15fTMwMq+PBtYX9tuQ09pL31Alvb0yzMzMrAZ1T7zLd+DRDXXpchmS5kpaJWnVtm3berIqZmZmfUZXg/yWPNRO/rk1p28Exhb2G5PT2ksfUyW9vTJeJiIuj4gjI+LIESNGdPGUzMzMyqWrQX4J0DpDfiZwUyH99DzLfjLwVB5yXwpMlTQ0T7ibCizNeU9Lmpxn1Z9ecaxqZZiZmVkNBnS0g6RrgCnAcEkbSLPkFwDXSZoDPAa8N+9+KzAdaAGeA2YDRMQOSRcDd+X9LoqI1sl855Jm8A8CfpQX2inDzMzMatBhkI+IU9vIOq7KvgGc18ZxFgILq6SvAg6rkr69WhlmZmZWG3/jnZmZWUk5yJuZmZWUg7yZmVlJOcibmZmVlIO8mZlZSTnIm5mZlZSDvJmZWUk5yJuZmZWUg7yZmVlJOcibmZmVlIO8mZlZSTnIm5mZlZSDvJmZWUk5yJuZmZWUg7yZmVlJOcibmZmVlIO8mZlZSTnIm5mZlZSDvJmZWUk5yJuZmZWUg7yZmVlJOcibmZmVlIO8mZlZSTnIm5mZlZSDvJmZWUk5yJuZmZWUg7yZmVlJOcibmZmVlIO8mZlZSTnIm5mZlZSDvJmZWUkNaHQF+psJ827ZbXvdgpMaVBMzMys7B/km5A8CZmbWHTxcb2ZmVlJ1BXlJ6yTdK+luSaty2jBJyyStzT+H5nRJulRSi6R7JB1ROM7MvP9aSTML6W/Ox2/Jr1U99TUzM+tPuuNO/q8j4k0RcWTengcsj4iJwPK8DXAiMDEvc4HLIH0oAOYDRwNHAfNbPxjkfc4qvG5aN9TXzMysX+iJ4foZwFV5/SrglEL64khWAkMkHQicACyLiB0RsRNYBkzLea+MiJUREcDiwrHMzMysA/UG+QB+Imm1pLk5bWREbMrrm4GReX00sL7w2g05rb30DVXSzczMrAb1zq4/JiI2SnoVsEzSg8XMiAhJUWcZHcofMOYCjBs3rqeLMzMz6xPqupOPiI3551bgRtIz9S15qJ38c2vefSMwtvDyMTmtvfQxVdKr1ePyiDgyIo4cMWJEPadkZmZWGl0O8pIGS9qvdR2YCqwBlgCtM+RnAjfl9SXA6XmW/WTgqTysvxSYKmlonnA3FVia856WNDnPqj+9cCwzMzPrQD3D9SOBG/NftQ0Aro6IH0u6C7hO0hzgMeC9ef9bgelAC/AcMBsgInZIuhi4K+93UUTsyOvnAouAQcCP8mJmZmY16HKQj4hHgDdWSd8OHFclPYDz2jjWQmBhlfRVwGFdraOZmVl/5m+8MzMzKykHeTMzs5JykDczMyspB3kzM7OScpA3MzMrKQd5MzOzknKQNzMzKykHeTMzs5JykDczMyspB3kzM7OScpA3MzMrKQd5MzOzknKQNzMzKykHeTMzs5JykDczMyspB3kzM7OScpA3MzMrKQd5MzOzknKQNzMzK6kBja5Afzdh3i09fsx1C07q9jLMzKz5+U7ezMyspBzkzczMSsrD9dYhD/+bmfVNvpM3MzMrKQd5MzOzknKQNzMzKykHeTMzs5JykDczMyspz67vA7p7dnujZ8s3unwzs/7CQd5eprPfwuegbWbWnDxcb2ZmVlIO8mZmZiXl4XrrtJ74pzpmZtb9HOT7AQdlM7P+ycP1ZmZmJdX0d/KSpgH/AuwBXBERCxpcpdLp6Tv9ZhxJ8F8EmFl/0NRBXtIewDeAdwAbgLskLYmI+xtbs8Zq9gDVjEHdzKw/auogDxwFtETEIwCSvgfMAPp1kC+bZv/QYmbWVzV7kB8NrC9sbwCOblBdmlbZ7pyb4XzqrYM/qJhZM2j2IF8TSXOBuXnzGUkPdePhhwNPduPx+qOmb0N9qbmPRx9owz7C7Vg/t2H9ursNx7eV0exBfiMwtrA9JqftJiIuBy7viQpIWhURR/bEsfsLt2H93Ibdw+1YP7dh/XqzDZv9T+juAiZKOkjSXsD7gSUNrpOZmVmf0NR38hHxoqQPA0tJf0K3MCLua3C1zMzM+oSmDvIAEXErcGsDq9AjjwH6Gbdh/dyG3cPtWD+3Yf16rQ0VEb1VlpmZmfWiZn8mb2ZmZl3kIN8OSdMkPSSpRdK8RtenWUkaK+lnku6XdJ+k83P6MEnLJK3NP4fmdEm6NLfrPZKOaOwZNA9Je0j6jaSb8/ZBku7MbXVtnoCKpIF5uyXnT2hkvZuFpCGSrpf0oKQHJL3F/bBzJH08/x6vkXSNpL3dD9snaaGkrZLWFNI63e8kzcz7r5U0szvq5iDfhsJX6p4ITAJOlTSpsbVqWi8Cn4iIScBk4LzcVvOA5RExEVietyG16cS8zAUu6/0qN63zgQcK218CvhYRhwA7gTk5fQ6wM6d/Le9n6f9c/DgiXgu8kdSW7oc1kjQa+ChwZEQcRprw/H7cDzuyCJhWkdapfidpGDCf9IVvRwHzWz8Y1MNBvm0vfaVuRDwPtH6lrlWIiE0R8eu8/jvSG+toUntdlXe7Cjglr88AFkeyEhgi6cBernbTkTQGOAm4Im8LOBa4Pu9S2YatbXs9cFzev9+StD/wNuBKgIh4PiJ24X7YWQOAQZIGAPsAm3A/bFdE/BzYUZHc2X53ArAsInZExE5gGS//4NBpDvJtq/aVuqMbVJc+Iw/XHQ7cCYyMiE05azMwMq+7bav7OnAB8Me8fQCwKyJezNvFdnqpDXP+U3n//uwgYBvwH/mRxxWSBuN+WLOI2Ah8GXicFNyfAlbjftgVne13PdIfHeSt20jaF/g+8LGIeLqYF+nPOPynHG2QdDKwNSJWN7oufdgA4Ajgsog4HHiWPw2RAu6HHcnDwzNIH5hGAYPphrvJ/q6R/c5Bvm01faWuJZL2JAX470bEDTl5S+vwZ/65Nae7bV/uL4G/lbSO9GjoWNLz5SF52BR2b6eX2jDn7w9s780KN6ENwIaIuDNvX08K+u6HtTseeDQitkXEC8ANpL7pfth5ne13PdIfHeTb5q/UrVF+Bncl8EBEfLWQtQRonSE6E7ipkH56nmU6GXiqMKzVL0XEpyNiTERMIPW12yLiNOBnwLvzbpVt2Nq278779+s71IjYDKyXdGhOOo70b6ndD2v3ODBZ0j7597q1Dd0PO6+z/W4pMFXS0DyiMjWn1ScivLSxANOB3wIPA//Q6Po06wIcQxqKuge4Oy/TSc/mlgNrgZ8Cw/L+Iv3lwsPAvaSZvA0/j2ZZgCnAzXn9YOBXQAvwn8DAnL533m7J+Qc3ut7NsABvAlblvvgDYKj7Yafb8HPAg8Aa4NvAQPfDDtvsGtIchhdII0pzutLvgDNyW7YAs7ujbv7GOzMzs5LycL2ZmVlJOcibmZmVlIO8mZlZSTnIm5mZlZSDvJmZWUk5yJuZmZWUg7yZmVlJOcibmZmV1P8DtbdubYTFlK8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 576x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdoAAACcCAYAAADYkmpqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAUGElEQVR4nO3df9RVVZ3H8fdHQEBhRIRxAJFHTc0fK7NI1JzJpWSKlK01TUllmkzW9MtmLENz0mo16prGsslJyZRlpY4/soyazF9kZppQToGIoqCAID+ERLNJ8jt/7P3g4fr8uBee/dyH+3xea93FPWfvu8/37LOf+71nn3MvigjMzMysjB2aHYCZmVkrc6I1MzMryInWzMysICdaMzOzgpxozczMCnKiNTMzK8iJ1oqSNEfSP3ZStqek5yUN6O24rPe0+nGWNFTSjyT9QdKNTYphqaTJzdi2dc+J1pomIp6KiGER8Zeu6kk6TdK9vRWX9ax6j/O2aHKieRewO7BbRPxDk2KwPsyJ1uqmpKXGTCvu07aQNLDZMZRQeL8mAI9GxKZGX9hRXK16DPozv8G0GEmflbRC0kZJiyQdm9cPkHSupMdz2TxJ43PZkZIezFNfD0o6stLeHElflvRL4I/A3pJeK+l2Sc/mbby7m7AmSPpl3u7PJI3KbbdJivY3lnzm+kSut0TS+yQdAFwOHJGnHzfkurtIukbSGklPSjqvPWHmff0PSWtzOx+v2U5H+/RBSQvztp+Q9OFKHxwtabmksyWtlrRS0jslTZH0aO6Hcxs4RkNzfE/mPr9X0tBc9g5JCyRtyHEeUHndWEk3531eIumTef3IHN/b8/IwSYslfSAvz5J0eT5mGyX9XNKESrsh6WOSHgMey+umSnoox3GfpNfVMcYOkzRX0nOSnpF0SSfHeaykW3O/LZb0oUrbF0i6IR/bjbkvJnbTn98B9gR+lMfI2ZVtTpf0FHBXrnujpFW53++RdFClnVmSLpP047ztByTtk8sk6av5+D8n6feSDpb0BeDzwHvytqfn+qfn8bRe0m1d9XdlfH1W0irgakk7SJqh9Pe6LvfJyEobp+Txs07S57obc9ZkEeFHizyA/YFlwNi83Absk59/Bvh9riPgEGA3YCSwHjgFGAhMy8u75dfNAZ4CDsrlu+RtfDAvHwqsBQ7sJKY5wOPAfsDQvHxRJb7I7ewMPAfsn8vGAAfl56cB99a0ew3wQ2B4budRYHou+wjwMLAHsCtwR/t2OtmnQcCJwD65b95CSsBvyPWPBjaR3lAHAR8C1gDX5u0fBLwI7JXrHwVs6OI4XZZjGAcMAI4EBuc+egF4a97O2cBiYEfSh+J5OYYdgb2BJ4C35TaPA1YBfw18C7ipsr1ZwEbg7/J2Lq32Z+6b2/NYGJqP6WpgUo7vVGBpfm1XY+xXwCn5+TDg8NrjnJfvAf4LGAK8PvflMbnsAuBPwJS87QuB++sY+0uByZXl9m1eQxpbQ/P60/MxGwx8DXiopp/WAYflcfE94Ppc9rbc/yNIY+QAYEwl5u9W2jkpH7cDcjvnAfd10d9Hk8bXxTmuocCZwP2kMTwYuAK4Lr/+QOD5yvG8JL9+cnf95EeT3puLNQxX5T/W+XXWfzfpzXEBcG2zO2Z7fACvyX0+GRhUU7YIOKmD15wC/Lpm3a+A0/LzOcAXK2XvAX5RU/8K4PxOYpoDnFdZ/ijw0/x88xtwfjPcAPx9+5ti5TWnsWViGAD8mUpyBz4MzMnP7wI+XCmbzKsT7Rc7irfymh8AZ+bnR5MS6YC8PDy3N6lSfx7wzjqO0Q65rUM6KPtX4Iaauivy9icBT9XUPwe4urL8n6QPUyvIH5Ty+lnkhJGXhwF/Acbn5SAnurz8TeBLHYyft3Qzxu4BvgCMqllfPc7j87aHV8ovBGbl5xcAd1TKDgRerKNfl9Jxot27i9eMyHV2qfTTlZXyKcAj+fkxpA9zhwM71LRzAVsm2v8hf+irHMc/AhM66e+jSeN5SGXdQuDYyvIY4KXch5+vOZ4759c70fbRR8mp41nA8fVUlLQv6U3jzRFxEPCpgnG1rIhYTOq7C4DVkq6XNDYXjyedWdYaCzxZs+5J0tlWu2WV5xOASXlKcYPSVO77gL/pIrRVled/JL3R18b+AimJfwRYmafvXttJe6NIZ3zVuKsxj62Jufq8w3WSTpB0f57O3EB6kx1VqbIuXrmZ58X87zOV8hc72q9OYh9CHcciIl7OcY4j9fvYmn4/l3QTTruZwMGkpLWupu3N+xsRzwPP5u29qjxv66yabY0nncV2Ncamk87KH1G6BDG1k318NiI2VtbVjrfa8TJEW3/dcvN+KV1SuChPxz5HSs6w5XHucKxGxF3AN0izEaslzZT0V51scwJwaaXvniWdBXf2NwWwJiL+VNPGLZU2FpI+oOxOzfjOfzu1x9v6kGKJNiLuIQ2wzSTtI+mnStcHf1F5I/0QcFlErM+vXV0qrlYXEddGxFGkP9QgTUdB+sPcp4OXPJ3rVu1JOiva3Gzl+TLg5xExovIYFhH/1AOx3xYRbyV9en+ENAVau31IU9Uv1cRdjXklacqt3fiONtf+RNJg4GbgK8DuETEC+AnpzbGnrSVNjXZ7LCSJFPsKUr8vqen34RExJdcdQEq01wAflfSamrbHV9odRpq2fLpSXnuMv1yzrZ0i4jrofIxFxGMRMY00fX0xcJOknTvYx5GShlfW1Y63rdHZf0NWXf9e0rTuZNIlkLa8vq7jHBFfj4g3ks6y9yNdjunIMtKMSrX/hkbEfV3EW7u8DDihpo0hEbGCNL6rx3Mn0mUg66N6+2aomcAn8mD9NOk6DaRBu5/SDTP3S6rrTNi2JGl/ScfkxPEn0lnWy7n4SuBLkvbNN3a8TtJupISyn6T3Shoo6T2kN5LZnWxmdq5/iqRB+fEmVW7a2crYd5d0Un5j/j/SNaj22J8B9pC0I0A+s7wB+LKk4flGk38Bvpvr3wCcKWmcpBHAZ7vZ/I6ka11rgE2STiBd8+xx+Sz1KuASpZuCBkg6Ih+zG4ATJR0raRBwFqkv7gN+DWzMN8wMza87WNKbctPnkt6sTwf+HbhGW35vdYqko3Iffol03bOjM31IH3A+ImlSHis7Szox93WnY0zS+yWNzvu4Ibf1crXhvM37gAslDVG6yWo6rxy7rfUM6bp1V4aT+nMdsBPwb/U2nsf4pHxcXiDt+8udVL8cOEf5RiulG/ca/drP5aTxPSG3MVrSSbnsJmBq5Xh+Ed/Y2qf12sHJn6KPBG6U9BDput6YXDwQ2Jd0rWIa8K38BmmNGQxcRDprar8x5pxcdgnpjfxnpJuOvk26FroOmEp6U19HugFnakSs7WgDecrvOOBk0tnJKl65iWNb7EBKlk+TZkLeArSfJd9Funa/SlJ7XJ8gveE9AdxLujHpqlz2rbyfvwN+S/owsYk09dbZPn2S1D/rSWc+t27tjkj6W0nPd1Hl06RrqQ+S9vVi0nW/RcD7Sdda1wJvB94eEX/OHy6mkm4eWpLLrwR2kfRGUt99INe7mJR0Z1S2eS1wft7eG/N2OhQRc0mzTN8g9cdi0nVy6HqMHQ8syPt+KXByRLzIq00jnU0+DdxCur5/Rxf9VY8LgfPyVOunO6lzDWmaegXpfpD7G2j/r0jjan1uYx3pA82rRMQtpGNwfZ6ing+c0MC2IPXfrcDPJG3MsU7K7S8APkY6pitzTMsbbN96kSI6m3HpgcalNmB2RBycr2csiogxHdS7HHggIq7Oy3cCMyLiwWLBWb+Rz1Avj4jaKfJ+QdIsYHlEnNfsWMz6o147o42I54Al7VMoeUrqkFz8A9LZLErfsdyPdKZi1rA8tTolT4WPI53J3dLsuMysfyqWaCVdR/qayP5KX8aeTro7dbqk/yVNBbZfc7gNWCfpYeBu4DMd3DVpVi+RvmaynjR1vJD0lQjbTumV30vu6LFns+Mz60rRqWMzM7P+zneqmZmZFeREa2ZmVlCR/yVi1KhR0dbWVqJpMzOzPmfevHlrI2J0R2VFEm1bWxtz584t0bSZmVmfI6n2p2w389SxmZlZQU60ZmZmBTnRmpmZFVTkGu32pm3Gj7dYXnrRiU2KxMzMWo3PaM3MzApyojUzMyvIidbMzKwgJ1ozM7OCnGjNzMwKcqI1MzMryInWzMysICdaMzOzgpxozczMCnKiNTMzK8iJ1szMrCAnWjMzs4KcaM3MzApyojUzMyvIidbMzKwgJ1ozM7OCnGjNzMwKcqI1MzMryInWzMysICdaMzOzgpxozczMCnKiNTMzK8iJ1szMrCAnWjMzs4KcaM3MzApyojUzMyuo20QrabykuyU9LGmBpDN7IzAzM7NWMLCOOpuAsyLiN5KGA/Mk3R4RDxeOzczMbLvX7RltRKyMiN/k5xuBhcC40oGZmZm1goau0UpqAw4FHigRjJmZWaupO9FKGgbcDHwqIp7roPwMSXMlzV2zZk1PxmhmZrbdqivRShpESrLfi4jvd1QnImZGxMSImDh69OiejNHMzGy7Vc9dxwK+DSyMiEvKh2RmZtY66jmjfTNwCnCMpIfyY0rhuMzMzFpCt1/viYh7AfVCLGZmZi3HvwxlZmZWkBOtmZlZQU60ZmZmBTnRmpmZFeREa2ZmVpATrZmZWUFOtGZmZgU50ZqZmRXkRGtmZlaQE62ZmVlBTrRmZmYFOdGamZkV5ERrZmZWkBOtmZlZQU60ZmZmBTnRmpmZFeREa2ZmVpATrZmZWUFOtGZmZgU50ZqZmRXkRGtmZlaQE62ZmVlBTrRmZmYFOdGamZkV5ERrZmZWkBOtmZlZQU60ZmZmBXWbaCVdJWm1pPm9EZCZmVkrqeeMdhZwfOE4zMzMWlK3iTYi7gGe7YVYzMzMWo6v0ZqZmRXUY4lW0hmS5kqau2bNmp5q1szMbLvWY4k2ImZGxMSImDh69OieatbMzGy75qljMzOzgur5es91wK+A/SUtlzS9fFhmZmatYWB3FSJiWm8EYmZm1oo8dWxmZlaQE62ZmVlBTrRmZmYFOdGamZkV5ERrZmZWkBOtmZlZQU60ZmZmBTnRmpmZFdTtD1a0orYZP252CGZm1k/4jNbMzKwgJ1ozM7OCnGjNzMwKcqI1MzMryInWzMysICdaMzOzgpxozczMCnKiNTMzK8iJ1szMrCAnWjMzs4KcaM3MzApyojUzMyuoX/6nAo2q/U8Ill50YpMiMTOz7Y3PaM3MzApyojUzMyvIidbMzKygfnGN1v/Ru5mZNYvPaM3MzArqF2e0jeqJM+BG71T2nc1mZq2prkQr6XjgUmAAcGVEXFQ0qj7OU9FmZlavbhOtpAHAZcBbgeXAg5JujYiHSwdnnfMZsJnZ9qGeM9rDgMUR8QSApOuBkwAn2gY0eha8rfWdeM3M+oZ6Eu04YFlleTkwqUw41lO6S9S1ibjR+laGPzCZtZ4euxlK0hnAGXnxeUmLeqptYBSwtgfb64+26ENd3NiLG63fonp9HLZgv/tvuWe4H7ddT/fhhM4K6km0K4DxleU98rotRMRMYGbDodVB0tyImFii7f7Cfbjt3Ifbzn3YM9yP2643+7Ce79E+COwraS9JOwInA7eWDcvMzKw1dHtGGxGbJH0cuI309Z6rImJB8cjMzMxaQF3XaCPiJ8BPCsfSlSJT0v2M+3DbuQ+3nfuwZ7gft12v9aEiore2ZWZm1u/4t47NzMwK6tOJVtLxkhZJWixpRrPj6askjZd0t6SHJS2QdGZeP1LS7ZIey//umtdL0tdzv/5O0huauwd9h6QBkn4raXZe3kvSA7mv/jvfEIikwXl5cS5va2bcfYmkEZJukvSIpIWSjvBYbIykf85/y/MlXSdpiMdi1yRdJWm1pPmVdQ2PO0mn5vqPSTq1J2Lrs4m28tOPJwAHAtMkHdjcqPqsTcBZEXEgcDjwsdxXM4A7I2Jf4M68DKlP982PM4Bv9n7IfdaZwMLK8sXAVyPiNcB6YHpePx1Yn9d/Ndez5FLgpxHxWuAQUn96LNZJ0jjgk8DEiDiYdBPqyXgsdmcWcHzNuobGnaSRwPmkH2U6DDi/PTlvk4jokw/gCOC2yvI5wDnNjmt7eAA/JP029SJgTF43BliUn18BTKvU31yvPz9I3xG/EzgGmA2I9IX2gbl885gk3YV/RH4+MNdTs/eh2Q9gF2BJbV94LDbUh+2/xjcyj63ZwNs8FuvquzZgfmW5oXEHTAOuqKzfot7WPvrsGS0d//TjuCbFst3I00aHAg8Au0fEyly0Ctg9P3ffduxrwNnAy3l5N2BDRGzKy9V+2tyHufwPuX5/txewBrg6T8FfKWlnPBbrFhErgK8ATwErSWNrHh6LW6PRcVdkPPblRGsNkjQMuBn4VEQ8Vy2L9PHMt5h3QtJUYHVEzGt2LNu5gcAbgG9GxKHAC7wyXQd4LHYnT1WeRPrQMhbYmVdPiVqDmjnu+nKireunHy2RNIiUZL8XEd/Pq5+RNCaXjwFW5/Xu21d7M/AOSUuB60nTx5cCIyS1f9+82k+b+zCX7wKs682A+6jlwPKIeCAv30RKvB6L9ZsMLImINRHxEvB90vj0WGxco+OuyHjsy4nWP/1YJ0kCvg0sjIhLKkW3Au13zZ1Kunbbvv4D+c67w4E/VKZX+qWIOCci9oiINtJYuysi3gfcDbwrV6vtw/a+fVeu3+/P0iJiFbBM0v551bGk/1LTY7F+TwGHS9op/22396HHYuMaHXe3AcdJ2jXPLByX122bZl+87ubC9hTgUeBx4HPNjqevPoCjSFMivwMeyo8ppOs0dwKPAXcAI3N9ke7ofhz4PenuxqbvR195AEcDs/PzvYFfA4uBG4HBef2QvLw4l+/d7Lj7ygN4PTA3j8cfALt6LDbch18AHgHmA98BBnssdttn15Guab9EmlmZvjXjDjg99+Vi4IM9EZt/GcrMzKygvjx1bGZmtt1zojUzMyvIidbMzKwgJ1ozM7OCnGjNzMwKcqI1MzMryInWzMysICdaMzOzgv4ft1RJbZ6rtTEAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 576x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for channel, array_sparse in score_arrays.items():\n",
    "    if array_sparse.nnz == 0:\n",
    "        # skip plotting distributions with no nonzero values\n",
    "        continue\n",
    "    plot_score_distribution(array_sparse, channel)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Write to disk\n",
    "\n",
    "Save sparse matrices to disk using [compressed npz format](https://docs.scipy.org/doc/scipy/reference/generated/scipy.sparse.save_npz.html), which is space efficifient, but whose reading is limited to `scipy.sparse.load_npz`.\n",
    "\n",
    "Also save the scores to a XZ-compressed CSV file, which is less efficient but more interoperable and widely supported. gz compression is more widely supported but has worse compression ratios and included a timestamp header making for non-deterministic file hashes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# ensure preferred_name never contains commas (to play nice with CSV)\n",
    "assert not gene_df.preferred_name.str.contains(\",\").any()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def array_to_df(array: np.ndarray):\n",
    "    return pd.DataFrame(\n",
    "        data=array,\n",
    "        columns=gene_df.preferred_name,\n",
    "        index=gene_df.preferred_name,\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 51min 25s, sys: 5.46 s, total: 51min 30s\n",
      "Wall time: 51min 31s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "for channel, array_sparse in score_arrays.items():\n",
    "    if array_sparse.nnz == 0:\n",
    "        continue\n",
    "    # save as .sparse.npz\n",
    "    path = f\"data/score-matrices/{channel}.sparse.npz\"\n",
    "    scipy.sparse.save_npz(path, array_sparse, compressed=True)\n",
    "    # save as .csv.xz\n",
    "    path = f\"data/score-matrices/{channel}.csv.xz\"\n",
    "    matrix_df = array_to_df(array_sparse.todense())\n",
    "    matrix_df.to_csv(path, index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "combined_score\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>preferred_name</th>\n",
       "      <th>ARF5</th>\n",
       "      <th>M6PR</th>\n",
       "      <th>FKBP4</th>\n",
       "      <th>CYP26B1</th>\n",
       "      <th>NDUFAF7</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>preferred_name</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>ARF5</th>\n",
       "      <td>0</td>\n",
       "      <td>161</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>M6PR</th>\n",
       "      <td>161</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>FKBP4</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "preferred_name  ARF5  M6PR  FKBP4  CYP26B1  NDUFAF7\n",
       "preferred_name                                     \n",
       "ARF5               0   161      0        0        0\n",
       "M6PR             161     0      0        0        0\n",
       "FKBP4              0     0      0        0        0"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Show what matrix_df looks like\n",
    "print(channel)\n",
    "matrix_df.iloc[:3, :5]"
   ]
  }
 ],
 "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": "3.8.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}