docs/notebooks/quick-start.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Quickstart\n",
    "\n",
    "This notebook shows how to train and validate a model with ScAPE and how to do predictions for cell/drug combinations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From c:\\Users\\pablo\\micromamba\\envs\\scape-env\\Lib\\site-packages\\keras\\src\\losses.py:2976: The name tf.losses.sparse_softmax_cross_entropy is deprecated. Please use tf.compat.v1.losses.sparse_softmax_cross_entropy instead.\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'0.1.0'"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Disable all warnings\n",
    "import scape\n",
    "import pandas as pd\n",
    "\n",
    "scape.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((614, 18211), (614, 18211))"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_de = scape.io.load_slogpvals(\"_data/de_train.parquet\")\n",
    "df_lfc = scape.io.load_lfc(\"_data/lfc_train.parquet\")\n",
    "\n",
    "# Make sure rows/columns are in the same order\n",
    "df_lfc = df_lfc.loc[df_de.index, df_de.columns]\n",
    "df_de.shape, df_lfc.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From c:\\Users\\pablo\\micromamba\\envs\\scape-env\\Lib\\site-packages\\keras\\src\\backend.py:1398: The name tf.executing_eagerly_outside_functions is deprecated. Please use tf.compat.v1.executing_eagerly_outside_functions instead.\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# We select only a subset of the genes for the model (top most variant genes)\n",
    "n_genes = 64\n",
    "top_genes = scape.util.select_top_variable([df_de], k=n_genes)\n",
    "\n",
    "# Data sources point to the dataframes with the data from which the features are extracted\n",
    "data_sources = {\n",
    "    'slogpval': df_de,\n",
    "    'lfc_pseudo': df_lfc\n",
    "}\n",
    "\n",
    "# Define the way in which the features are extracted from the data sources\n",
    "feature_extraction = {\n",
    "    'slogpval_drug': {\n",
    "        'source': 'slogpval',\n",
    "        'groupby': 'sm_name',\n",
    "        'function': 'median'\n",
    "    },\n",
    "    'lfc_drug': {\n",
    "        'source': 'lfc_pseudo',\n",
    "        'groupby': 'sm_name',\n",
    "        'function': 'median'\n",
    "    },\n",
    "    'slogpval_cell': {\n",
    "        'source': 'slogpval',\n",
    "        'groupby': 'cell_type',\n",
    "        'function': 'median'\n",
    "    },\n",
    "    'lfc_cell': {\n",
    "        'source': 'lfc_pseudo',\n",
    "        'groupby': 'cell_type',\n",
    "        'function': 'median'\n",
    "    }\n",
    "}\n",
    "\n",
    "# Names of the input layers -> feature extraction method\n",
    "input_mapping = {\n",
    "    'in_slogpval_drug': 'slogpval_drug',\n",
    "    'in_lfc_drug': 'lfc_drug',\n",
    "    'in_slogpval_cell_encoder': 'slogpval_cell',\n",
    "    'in_lfc_cell_encoder': 'lfc_cell',\n",
    "    'in_slogpval_cell_decoder': 'slogpval_cell',\n",
    "    'in_lfc_cell_decoder': 'lfc_cell',\n",
    "}\n",
    "\n",
    "# Configuration for the NN architecture\n",
    "config = {\n",
    "    \"inputs\": {\n",
    "        \"in_slogpval_drug\"          : [n_genes, 256, 128],\n",
    "        \"in_lfc_drug\"               : [n_genes, 256, 128],\n",
    "    },\n",
    "    \"conditional_encoder_input_structure\": {\n",
    "        \"in_slogpval_cell_encoder\"  : [n_genes, 32],\n",
    "        \"in_lfc_cell_encoder\"       : [n_genes, 32, 16],\n",
    "    },\n",
    "    \"conditional_decoder_input_structure\": {\n",
    "        \"in_slogpval_cell_decoder\"  : [n_genes, 32],\n",
    "        \"in_lfc_cell_decoder\"       : [n_genes, 32, 16],\n",
    "    },\n",
    "    \"conditional_decoder_input_hidden_sizes\": [32],\n",
    "    \"encoder_hidden_layer_sizes\": [128, 128],\n",
    "    \"decoder_hidden_layer_sizes\": [128, 512],\n",
    "    \"outputs\": {\n",
    "        \"logpval\": (df_de.shape[1], \"linear\"), \n",
    "    },\n",
    "    # Layer config\n",
    "    \"noise\": 0.01,\n",
    "    \"dropout\": 0.05,\n",
    "    \"l1\": 0,\n",
    "    \"l2\": 0\n",
    "}\n",
    "\n",
    "model_setup = {\n",
    "    \"data_sources\": data_sources,\n",
    "    \"feature_extraction\": feature_extraction,\n",
    "    \"input_mapping\": input_mapping,\n",
    "    \"output_genes\": df_de.columns,\n",
    "    \"config\": config\n",
    "}\n",
    "\n",
    "# Create model\n",
    "scm = scape.SCAPE(model_setup)\n",
    "scape.util.plot(scm.model, show_shapes=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"autoencoder\"\n",
      "__________________________________________________________________________________________________\n",
      " Layer (type)                Output Shape                 Param #   Connected to                  \n",
      "==================================================================================================\n",
      " in_slogpval_drug (InputLay  [(None, 64)]                 0         []                            \n",
      " er)                                                                                              \n",
      "                                                                                                  \n",
      " in_lfc_drug (InputLayer)    [(None, 64)]                 0         []                            \n",
      "                                                                                                  \n",
      " in_slogpval_cell_encoder (  [(None, 64)]                 0         []                            \n",
      " InputLayer)                                                                                      \n",
      "                                                                                                  \n",
      " in_lfc_cell_encoder (Input  [(None, 64)]                 0         []                            \n",
      " Layer)                                                                                           \n",
      "                                                                                                  \n",
      " encoder (Functional)        (None, 128)                  198288    ['in_slogpval_drug[0][0]',    \n",
      "                                                                     'in_lfc_drug[0][0]',         \n",
      "                                                                     'in_slogpval_cell_encoder[0][\n",
      "                                                                    0]',                          \n",
      "                                                                     'in_lfc_cell_encoder[0][0]'] \n",
      "                                                                                                  \n",
      " in_slogpval_cell_decoder (  [(None, 64)]                 0         []                            \n",
      " InputLayer)                                                                                      \n",
      "                                                                                                  \n",
      " in_lfc_cell_decoder (Input  [(None, 64)]                 0         []                            \n",
      " Layer)                                                                                           \n",
      "                                                                                                  \n",
      " decoder (Functional)        (None, 18211)                9439187   ['encoder[0][0]',             \n",
      "                                                                     'in_slogpval_cell_decoder[0][\n",
      "                                                                    0]',                          \n",
      "                                                                     'in_lfc_cell_decoder[0][0]'] \n",
      "                                                                                                  \n",
      "==================================================================================================\n",
      "Total params: 9637475 (36.76 MB)\n",
      "Trainable params: 9632483 (36.75 MB)\n",
      "Non-trainable params: 4992 (19.50 KB)\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "scm.model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Baseline zero MRRMSE: 1.4273\n",
      "Baseline slogpval_drug MRRMSE: 1.1278\n",
      "Model size (MB): 36.76\n",
      "Model inputs: ['in_slogpval_drug', 'in_lfc_drug', 'in_slogpval_cell_encoder', 'in_lfc_cell_encoder', 'in_slogpval_cell_decoder', 'in_lfc_cell_decoder']\n",
      "WARNING:tensorflow:From c:\\Users\\pablo\\micromamba\\envs\\scape-env\\Lib\\site-packages\\keras\\src\\utils\\tf_utils.py:492: The name tf.ragged.RaggedTensorValue is deprecated. Please use tf.compat.v1.ragged.RaggedTensorValue instead.\n",
      "\n",
      "Epoch    1: val_loss None -> 1.4151, train loss: 1.3068 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0086, 0.85%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.7969, -25.48%\n",
      "Epoch    2: val_loss 1.4151 -> 1.3815, train loss: 1.2771 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0332, 3.21%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8164, -22.50%\n",
      "Epoch    3: val_loss 1.3815 -> 1.3536, train loss: 1.2502 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0545, 5.16%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8332, -20.02%\n",
      "Epoch    5: val_loss 1.3536 -> 1.3477, train loss: 1.2244 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0590, 5.57%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8368, -19.50%\n",
      "Epoch    8: val_loss 1.3477 -> 1.3360, train loss: 1.1903 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0683, 6.40%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8441, -18.46%\n",
      "Epoch    9: val_loss 1.3360 -> 1.3299, train loss: 1.1699 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0732, 6.82%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8480, -17.93%\n",
      "Epoch   13: val_loss 1.3299 -> 1.3092, train loss: 1.1388 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0902, 8.27%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8614, -16.09%\n",
      "Epoch   36: val_loss 1.3092 -> 1.2762, train loss: 1.0179 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1184, 10.59%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8837, -13.16%\n",
      "Epoch   60: val_loss 1.2762 -> 1.2740, train loss: 0.9563 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1203, 10.74%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8852, -12.97%\n",
      "Epoch   97: val_loss 1.2740 -> 1.2705, train loss: 0.9279 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1234, 10.98%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8877, -12.66%\n",
      "Epoch  118: val_loss 1.2705 -> 1.2666, train loss: 0.8632 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1269, 11.26%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8904, -12.31%\n",
      "Epoch  129: val_loss 1.2666 -> 1.2664, train loss: 0.8401 (lr=0.00425)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1270, 11.27%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8905, -12.30%\n",
      "Epoch  131: val_loss 1.2664 -> 1.2651, train loss: 0.8480 (lr=0.00425)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1282, 11.36%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8915, -12.18%\n",
      "Epoch  137: val_loss 1.2651 -> 1.2608, train loss: 0.8515 (lr=0.00425)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1320, 11.66%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8945, -11.80%\n",
      "Epoch  147: val_loss 1.2608 -> 1.2554, train loss: 0.8212 (lr=0.00361)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1369, 12.04%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8984, -11.31%\n",
      "Epoch  149: val_loss 1.2554 -> 1.2513, train loss: 0.8605 (lr=0.00361)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1406, 12.33%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9013, -10.95%\n",
      "Epoch  152: val_loss 1.2513 -> 1.2505, train loss: 0.8256 (lr=0.00361)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1413, 12.38%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9018, -10.88%\n",
      "Epoch  156: val_loss 1.2505 -> 1.2487, train loss: 0.8198 (lr=0.00361)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1430, 12.51%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9031, -10.73%\n",
      "Epoch  160: val_loss 1.2487 -> 1.2444, train loss: 0.8486 (lr=0.00361)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1470, 12.82%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9063, -10.34%\n",
      "Epoch  168: val_loss 1.2444 -> 1.2399, train loss: 0.8618 (lr=0.00361)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1511, 13.13%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9096, -9.94%\n",
      "Epoch  169: val_loss 1.2399 -> 1.2362, train loss: 0.8460 (lr=0.00361)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1546, 13.39%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9123, -9.62%\n",
      "Epoch  170: val_loss 1.2362 -> 1.2361, train loss: 0.8276 (lr=0.00361)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1547, 13.40%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9124, -9.60%\n",
      "Epoch  179: val_loss 1.2361 -> 1.2349, train loss: 0.7935 (lr=0.00307)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1558, 13.48%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9133, -9.50%\n",
      "Epoch  180: val_loss 1.2349 -> 1.2314, train loss: 0.7964 (lr=0.00307)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1591, 13.72%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9158, -9.19%\n",
      "Epoch  183: val_loss 1.2314 -> 1.2246, train loss: 0.7999 (lr=0.00307)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1655, 14.20%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9209, -8.59%\n",
      "Epoch  187: val_loss 1.2246 -> 1.2211, train loss: 0.8200 (lr=0.00307)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1688, 14.44%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9235, -8.28%\n",
      "Epoch  188: val_loss 1.2211 -> 1.2160, train loss: 0.8214 (lr=0.00307)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1737, 14.80%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9274, -7.83%\n",
      "Epoch  197: val_loss 1.2160 -> 1.2111, train loss: 0.8091 (lr=0.00307)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1784, 15.14%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9312, -7.39%\n",
      "Epoch  198: val_loss 1.2111 -> 1.2095, train loss: 0.8144 (lr=0.00261)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1800, 15.26%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9324, -7.25%\n",
      "Epoch  205: val_loss 1.2095 -> 1.2095, train loss: 0.8035 (lr=0.00261)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1801, 15.26%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9325, -7.24%\n",
      "Epoch  206: val_loss 1.2095 -> 1.1976, train loss: 0.7937 (lr=0.00261)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1917, 16.09%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9417, -6.20%\n",
      "Epoch  210: val_loss 1.1976 -> 1.1904, train loss: 0.8134 (lr=0.00222)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1990, 16.59%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9474, -5.56%\n",
      "Epoch  217: val_loss 1.1904 -> 1.1814, train loss: 0.7985 (lr=0.00222)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2081, 17.23%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9546, -4.75%\n",
      "Epoch  220: val_loss 1.1814 -> 1.1746, train loss: 0.8014 (lr=0.00222)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2151, 17.70%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9601, -4.15%\n",
      "Epoch  228: val_loss 1.1746 -> 1.1712, train loss: 0.7788 (lr=0.00222)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2187, 17.94%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9629, -3.85%\n",
      "Epoch  236: val_loss 1.1712 -> 1.1650, train loss: 0.7723 (lr=0.00222)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2251, 18.38%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9680, -3.30%\n",
      "Epoch  237: val_loss 1.1650 -> 1.1646, train loss: 0.7943 (lr=0.00222)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2255, 18.40%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9683, -3.27%\n",
      "Epoch  238: val_loss 1.1646 -> 1.1572, train loss: 0.7763 (lr=0.00222)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2334, 18.92%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9746, -2.61%\n",
      "Epoch  239: val_loss 1.1572 -> 1.1482, train loss: 0.7714 (lr=0.00222)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2430, 19.55%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9822, -1.81%\n",
      "Epoch  240: val_loss 1.1482 -> 1.1392, train loss: 0.8035 (lr=0.00222)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2529, 20.18%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9900, -1.01%\n",
      "Epoch  247: val_loss 1.1392 -> 1.1372, train loss: 0.7736 (lr=0.00189)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2551, 20.32%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9917, -0.84%\n",
      "Epoch  249: val_loss 1.1372 -> 1.1350, train loss: 0.7843 (lr=0.00189)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2575, 20.48%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9937, -0.64%\n",
      "Epoch  251: val_loss 1.1350 -> 1.1340, train loss: 0.7549 (lr=0.00189)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2586, 20.55%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.9945, -0.56%\n",
      "Epoch  257: val_loss 1.1340 -> 1.1257, train loss: 0.7666 (lr=0.00189)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2678, 21.13%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0018, 0.18%\n",
      "Epoch  259: val_loss 1.1257 -> 1.1243, train loss: 0.7734 (lr=0.00189)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2694, 21.22%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0030, 0.30%\n",
      "Epoch  260: val_loss 1.1243 -> 1.1215, train loss: 0.7651 (lr=0.00189)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2727, 21.43%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0056, 0.56%\n",
      "Epoch  261: val_loss 1.1215 -> 1.1140, train loss: 0.7804 (lr=0.00189)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2812, 21.95%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0123, 1.22%\n",
      "Epoch  263: val_loss 1.1140 -> 1.1115, train loss: 0.7661 (lr=0.00189)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2841, 22.13%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0147, 1.45%\n",
      "Epoch  274: val_loss 1.1115 -> 1.1100, train loss: 0.7808 (lr=0.00160)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2859, 22.23%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0160, 1.58%\n",
      "Epoch  276: val_loss 1.1100 -> 1.1055, train loss: 0.7628 (lr=0.00160)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2910, 22.54%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0201, 1.97%\n",
      "Epoch  277: val_loss 1.1055 -> 1.1014, train loss: 0.8005 (lr=0.00160)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.2959, 22.83%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0240, 2.34%\n",
      "Epoch  278: val_loss 1.1014 -> 1.0958, train loss: 0.7529 (lr=0.00160)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.3025, 23.23%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0292, 2.84%\n",
      "Epoch  279: val_loss 1.0958 -> 1.0928, train loss: 0.7609 (lr=0.00160)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.3060, 23.43%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0320, 3.10%\n",
      "Epoch  280: val_loss 1.0928 -> 1.0924, train loss: 0.7886 (lr=0.00160)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.3065, 23.46%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0324, 3.13%\n",
      "Epoch  281: val_loss 1.0924 -> 1.0896, train loss: 0.7497 (lr=0.00160)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.3099, 23.66%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0351, 3.39%\n",
      "Epoch  283: val_loss 1.0896 -> 1.0887, train loss: 0.7864 (lr=0.00136)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.3109, 23.72%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0359, 3.46%\n",
      "Epoch  285: val_loss 1.0887 -> 1.0832, train loss: 0.7448 (lr=0.00136)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.3176, 24.10%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0411, 3.95%\n",
      "Epoch  286: val_loss 1.0832 -> 1.0766, train loss: 0.7736 (lr=0.00136)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.3257, 24.57%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0475, 4.53%\n",
      "Epoch  290: val_loss 1.0766 -> 1.0660, train loss: 0.7499 (lr=0.00136)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.3389, 25.31%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 1.0579, 5.48%\n"
     ]
    }
   ],
   "source": [
    "result = scm.train(\n",
    "    val_cells=[\"NK cells\"], \n",
    "    val_drugs=[\"Prednisolone\"],\n",
    "    input_columns=top_genes,\n",
    "    epochs=300,\n",
    "    baselines=[\"zero\", \"slogpval_drug\"]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "scape.util.plot_result(result, legend=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Inference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20/20 [==============================] - 1s 10ms/step\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></th>\n",
       "      <th></th>\n",
       "      <th>A1BG</th>\n",
       "      <th>A1BG-AS1</th>\n",
       "      <th>A2M</th>\n",
       "      <th>A2M-AS1</th>\n",
       "      <th>A2MP1</th>\n",
       "      <th>A4GALT</th>\n",
       "      <th>AAAS</th>\n",
       "      <th>AACS</th>\n",
       "      <th>AAGAB</th>\n",
       "      <th>AAK1</th>\n",
       "      <th>...</th>\n",
       "      <th>ZUP1</th>\n",
       "      <th>ZW10</th>\n",
       "      <th>ZWILCH</th>\n",
       "      <th>ZWINT</th>\n",
       "      <th>ZXDA</th>\n",
       "      <th>ZXDB</th>\n",
       "      <th>ZXDC</th>\n",
       "      <th>ZYG11B</th>\n",
       "      <th>ZYX</th>\n",
       "      <th>ZZEF1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cell_type</th>\n",
       "      <th>sm_name</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>NK cells</th>\n",
       "      <th>Clotrimazole</th>\n",
       "      <td>0.194414</td>\n",
       "      <td>0.157227</td>\n",
       "      <td>-0.070179</td>\n",
       "      <td>0.009600</td>\n",
       "      <td>0.114648</td>\n",
       "      <td>0.338328</td>\n",
       "      <td>0.076248</td>\n",
       "      <td>0.163527</td>\n",
       "      <td>-0.140055</td>\n",
       "      <td>0.244689</td>\n",
       "      <td>...</td>\n",
       "      <td>0.042096</td>\n",
       "      <td>-0.158234</td>\n",
       "      <td>-0.059915</td>\n",
       "      <td>-0.039544</td>\n",
       "      <td>-0.019006</td>\n",
       "      <td>0.054201</td>\n",
       "      <td>0.150999</td>\n",
       "      <td>0.103660</td>\n",
       "      <td>0.112037</td>\n",
       "      <td>0.037434</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>T cells CD4+</th>\n",
       "      <th>Clotrimazole</th>\n",
       "      <td>0.009779</td>\n",
       "      <td>-0.097105</td>\n",
       "      <td>0.083260</td>\n",
       "      <td>-0.057180</td>\n",
       "      <td>0.010844</td>\n",
       "      <td>0.243668</td>\n",
       "      <td>-0.184838</td>\n",
       "      <td>0.089963</td>\n",
       "      <td>-0.065037</td>\n",
       "      <td>0.227762</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.202045</td>\n",
       "      <td>0.087562</td>\n",
       "      <td>0.036073</td>\n",
       "      <td>-0.092795</td>\n",
       "      <td>0.300995</td>\n",
       "      <td>-0.067515</td>\n",
       "      <td>0.244625</td>\n",
       "      <td>0.075626</td>\n",
       "      <td>0.144302</td>\n",
       "      <td>-0.069032</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>T cells CD8+</th>\n",
       "      <th>Clotrimazole</th>\n",
       "      <td>-0.163914</td>\n",
       "      <td>-0.397257</td>\n",
       "      <td>-0.268933</td>\n",
       "      <td>-0.070648</td>\n",
       "      <td>-0.159012</td>\n",
       "      <td>-0.464534</td>\n",
       "      <td>-0.134033</td>\n",
       "      <td>-0.198160</td>\n",
       "      <td>-0.365885</td>\n",
       "      <td>0.217887</td>\n",
       "      <td>...</td>\n",
       "      <td>0.014628</td>\n",
       "      <td>-0.370442</td>\n",
       "      <td>-0.360676</td>\n",
       "      <td>-0.486203</td>\n",
       "      <td>0.011065</td>\n",
       "      <td>-0.289772</td>\n",
       "      <td>0.052777</td>\n",
       "      <td>-0.105549</td>\n",
       "      <td>-0.061072</td>\n",
       "      <td>-0.138286</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>T regulatory cells</th>\n",
       "      <th>Clotrimazole</th>\n",
       "      <td>0.260759</td>\n",
       "      <td>0.264802</td>\n",
       "      <td>0.388628</td>\n",
       "      <td>0.572786</td>\n",
       "      <td>0.525528</td>\n",
       "      <td>0.398779</td>\n",
       "      <td>-0.064918</td>\n",
       "      <td>0.303745</td>\n",
       "      <td>-0.179417</td>\n",
       "      <td>0.211233</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.056566</td>\n",
       "      <td>0.295044</td>\n",
       "      <td>-0.019873</td>\n",
       "      <td>0.166435</td>\n",
       "      <td>0.413039</td>\n",
       "      <td>0.433954</td>\n",
       "      <td>0.356801</td>\n",
       "      <td>0.264965</td>\n",
       "      <td>-0.105367</td>\n",
       "      <td>-0.027750</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NK cells</th>\n",
       "      <th>Mometasone Furoate</th>\n",
       "      <td>0.761197</td>\n",
       "      <td>0.099600</td>\n",
       "      <td>-0.487814</td>\n",
       "      <td>-0.304756</td>\n",
       "      <td>0.068401</td>\n",
       "      <td>2.010531</td>\n",
       "      <td>0.526351</td>\n",
       "      <td>0.679624</td>\n",
       "      <td>-0.004084</td>\n",
       "      <td>1.631494</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.116993</td>\n",
       "      <td>-0.314878</td>\n",
       "      <td>0.472689</td>\n",
       "      <td>0.101642</td>\n",
       "      <td>0.101359</td>\n",
       "      <td>0.426433</td>\n",
       "      <td>0.040792</td>\n",
       "      <td>0.540312</td>\n",
       "      <td>-1.149533</td>\n",
       "      <td>0.273278</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>T regulatory cells</th>\n",
       "      <th>Atorvastatin</th>\n",
       "      <td>-0.090314</td>\n",
       "      <td>-0.416945</td>\n",
       "      <td>-0.410972</td>\n",
       "      <td>-0.486288</td>\n",
       "      <td>-0.504089</td>\n",
       "      <td>-0.313944</td>\n",
       "      <td>-0.096291</td>\n",
       "      <td>-0.650317</td>\n",
       "      <td>-0.118002</td>\n",
       "      <td>-0.001408</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.279535</td>\n",
       "      <td>-0.523046</td>\n",
       "      <td>-0.400566</td>\n",
       "      <td>-0.560992</td>\n",
       "      <td>-0.406564</td>\n",
       "      <td>-0.144007</td>\n",
       "      <td>-0.370364</td>\n",
       "      <td>-0.139863</td>\n",
       "      <td>0.039296</td>\n",
       "      <td>-0.442674</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NK cells</th>\n",
       "      <th>Riociguat</th>\n",
       "      <td>0.041222</td>\n",
       "      <td>0.170445</td>\n",
       "      <td>-0.020082</td>\n",
       "      <td>-0.026019</td>\n",
       "      <td>0.041145</td>\n",
       "      <td>0.196462</td>\n",
       "      <td>-0.011627</td>\n",
       "      <td>0.164335</td>\n",
       "      <td>-0.200379</td>\n",
       "      <td>0.124760</td>\n",
       "      <td>...</td>\n",
       "      <td>0.128824</td>\n",
       "      <td>-0.045591</td>\n",
       "      <td>-0.197567</td>\n",
       "      <td>-0.049303</td>\n",
       "      <td>-0.030546</td>\n",
       "      <td>0.008790</td>\n",
       "      <td>0.130760</td>\n",
       "      <td>0.022545</td>\n",
       "      <td>0.074422</td>\n",
       "      <td>-0.059573</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>T cells CD4+</th>\n",
       "      <th>Riociguat</th>\n",
       "      <td>0.027641</td>\n",
       "      <td>0.020953</td>\n",
       "      <td>0.023018</td>\n",
       "      <td>-0.161994</td>\n",
       "      <td>-0.205585</td>\n",
       "      <td>0.018626</td>\n",
       "      <td>-0.115283</td>\n",
       "      <td>0.071414</td>\n",
       "      <td>-0.018889</td>\n",
       "      <td>0.036839</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.069316</td>\n",
       "      <td>0.099511</td>\n",
       "      <td>-0.087466</td>\n",
       "      <td>0.044207</td>\n",
       "      <td>0.243455</td>\n",
       "      <td>0.023825</td>\n",
       "      <td>0.109964</td>\n",
       "      <td>-0.047073</td>\n",
       "      <td>0.138471</td>\n",
       "      <td>-0.097747</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>T cells CD8+</th>\n",
       "      <th>Riociguat</th>\n",
       "      <td>-0.099485</td>\n",
       "      <td>-0.306791</td>\n",
       "      <td>-0.288272</td>\n",
       "      <td>-0.156944</td>\n",
       "      <td>-0.162233</td>\n",
       "      <td>-0.419645</td>\n",
       "      <td>-0.079846</td>\n",
       "      <td>-0.237401</td>\n",
       "      <td>-0.370255</td>\n",
       "      <td>0.184281</td>\n",
       "      <td>...</td>\n",
       "      <td>0.007653</td>\n",
       "      <td>-0.323799</td>\n",
       "      <td>-0.373684</td>\n",
       "      <td>-0.442766</td>\n",
       "      <td>0.005768</td>\n",
       "      <td>-0.241377</td>\n",
       "      <td>-0.020977</td>\n",
       "      <td>-0.176173</td>\n",
       "      <td>-0.021656</td>\n",
       "      <td>-0.232762</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>T regulatory cells</th>\n",
       "      <th>Riociguat</th>\n",
       "      <td>-0.086253</td>\n",
       "      <td>-0.415494</td>\n",
       "      <td>-0.391324</td>\n",
       "      <td>-0.470232</td>\n",
       "      <td>-0.472357</td>\n",
       "      <td>-0.354360</td>\n",
       "      <td>-0.073059</td>\n",
       "      <td>-0.500039</td>\n",
       "      <td>-0.140615</td>\n",
       "      <td>0.010131</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.213496</td>\n",
       "      <td>-0.426152</td>\n",
       "      <td>-0.393975</td>\n",
       "      <td>-0.564646</td>\n",
       "      <td>-0.306568</td>\n",
       "      <td>-0.110232</td>\n",
       "      <td>-0.255466</td>\n",
       "      <td>-0.065244</td>\n",
       "      <td>0.024594</td>\n",
       "      <td>-0.255696</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>614 rows × 18211 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                           A1BG  A1BG-AS1       A2M   A2M-AS1  \\\n",
       "cell_type          sm_name                                                      \n",
       "NK cells           Clotrimazole        0.194414  0.157227 -0.070179  0.009600   \n",
       "T cells CD4+       Clotrimazole        0.009779 -0.097105  0.083260 -0.057180   \n",
       "T cells CD8+       Clotrimazole       -0.163914 -0.397257 -0.268933 -0.070648   \n",
       "T regulatory cells Clotrimazole        0.260759  0.264802  0.388628  0.572786   \n",
       "NK cells           Mometasone Furoate  0.761197  0.099600 -0.487814 -0.304756   \n",
       "...                                         ...       ...       ...       ...   \n",
       "T regulatory cells Atorvastatin       -0.090314 -0.416945 -0.410972 -0.486288   \n",
       "NK cells           Riociguat           0.041222  0.170445 -0.020082 -0.026019   \n",
       "T cells CD4+       Riociguat           0.027641  0.020953  0.023018 -0.161994   \n",
       "T cells CD8+       Riociguat          -0.099485 -0.306791 -0.288272 -0.156944   \n",
       "T regulatory cells Riociguat          -0.086253 -0.415494 -0.391324 -0.470232   \n",
       "\n",
       "                                          A2MP1    A4GALT      AAAS      AACS  \\\n",
       "cell_type          sm_name                                                      \n",
       "NK cells           Clotrimazole        0.114648  0.338328  0.076248  0.163527   \n",
       "T cells CD4+       Clotrimazole        0.010844  0.243668 -0.184838  0.089963   \n",
       "T cells CD8+       Clotrimazole       -0.159012 -0.464534 -0.134033 -0.198160   \n",
       "T regulatory cells Clotrimazole        0.525528  0.398779 -0.064918  0.303745   \n",
       "NK cells           Mometasone Furoate  0.068401  2.010531  0.526351  0.679624   \n",
       "...                                         ...       ...       ...       ...   \n",
       "T regulatory cells Atorvastatin       -0.504089 -0.313944 -0.096291 -0.650317   \n",
       "NK cells           Riociguat           0.041145  0.196462 -0.011627  0.164335   \n",
       "T cells CD4+       Riociguat          -0.205585  0.018626 -0.115283  0.071414   \n",
       "T cells CD8+       Riociguat          -0.162233 -0.419645 -0.079846 -0.237401   \n",
       "T regulatory cells Riociguat          -0.472357 -0.354360 -0.073059 -0.500039   \n",
       "\n",
       "                                          AAGAB      AAK1  ...      ZUP1  \\\n",
       "cell_type          sm_name                                 ...             \n",
       "NK cells           Clotrimazole       -0.140055  0.244689  ...  0.042096   \n",
       "T cells CD4+       Clotrimazole       -0.065037  0.227762  ... -0.202045   \n",
       "T cells CD8+       Clotrimazole       -0.365885  0.217887  ...  0.014628   \n",
       "T regulatory cells Clotrimazole       -0.179417  0.211233  ... -0.056566   \n",
       "NK cells           Mometasone Furoate -0.004084  1.631494  ... -0.116993   \n",
       "...                                         ...       ...  ...       ...   \n",
       "T regulatory cells Atorvastatin       -0.118002 -0.001408  ... -0.279535   \n",
       "NK cells           Riociguat          -0.200379  0.124760  ...  0.128824   \n",
       "T cells CD4+       Riociguat          -0.018889  0.036839  ... -0.069316   \n",
       "T cells CD8+       Riociguat          -0.370255  0.184281  ...  0.007653   \n",
       "T regulatory cells Riociguat          -0.140615  0.010131  ... -0.213496   \n",
       "\n",
       "                                           ZW10    ZWILCH     ZWINT      ZXDA  \\\n",
       "cell_type          sm_name                                                      \n",
       "NK cells           Clotrimazole       -0.158234 -0.059915 -0.039544 -0.019006   \n",
       "T cells CD4+       Clotrimazole        0.087562  0.036073 -0.092795  0.300995   \n",
       "T cells CD8+       Clotrimazole       -0.370442 -0.360676 -0.486203  0.011065   \n",
       "T regulatory cells Clotrimazole        0.295044 -0.019873  0.166435  0.413039   \n",
       "NK cells           Mometasone Furoate -0.314878  0.472689  0.101642  0.101359   \n",
       "...                                         ...       ...       ...       ...   \n",
       "T regulatory cells Atorvastatin       -0.523046 -0.400566 -0.560992 -0.406564   \n",
       "NK cells           Riociguat          -0.045591 -0.197567 -0.049303 -0.030546   \n",
       "T cells CD4+       Riociguat           0.099511 -0.087466  0.044207  0.243455   \n",
       "T cells CD8+       Riociguat          -0.323799 -0.373684 -0.442766  0.005768   \n",
       "T regulatory cells Riociguat          -0.426152 -0.393975 -0.564646 -0.306568   \n",
       "\n",
       "                                           ZXDB      ZXDC    ZYG11B       ZYX  \\\n",
       "cell_type          sm_name                                                      \n",
       "NK cells           Clotrimazole        0.054201  0.150999  0.103660  0.112037   \n",
       "T cells CD4+       Clotrimazole       -0.067515  0.244625  0.075626  0.144302   \n",
       "T cells CD8+       Clotrimazole       -0.289772  0.052777 -0.105549 -0.061072   \n",
       "T regulatory cells Clotrimazole        0.433954  0.356801  0.264965 -0.105367   \n",
       "NK cells           Mometasone Furoate  0.426433  0.040792  0.540312 -1.149533   \n",
       "...                                         ...       ...       ...       ...   \n",
       "T regulatory cells Atorvastatin       -0.144007 -0.370364 -0.139863  0.039296   \n",
       "NK cells           Riociguat           0.008790  0.130760  0.022545  0.074422   \n",
       "T cells CD4+       Riociguat           0.023825  0.109964 -0.047073  0.138471   \n",
       "T cells CD8+       Riociguat          -0.241377 -0.020977 -0.176173 -0.021656   \n",
       "T regulatory cells Riociguat          -0.110232 -0.255466 -0.065244  0.024594   \n",
       "\n",
       "                                          ZZEF1  \n",
       "cell_type          sm_name                       \n",
       "NK cells           Clotrimazole        0.037434  \n",
       "T cells CD4+       Clotrimazole       -0.069032  \n",
       "T cells CD8+       Clotrimazole       -0.138286  \n",
       "T regulatory cells Clotrimazole       -0.027750  \n",
       "NK cells           Mometasone Furoate  0.273278  \n",
       "...                                         ...  \n",
       "T regulatory cells Atorvastatin       -0.442674  \n",
       "NK cells           Riociguat          -0.059573  \n",
       "T cells CD4+       Riociguat          -0.097747  \n",
       "T cells CD8+       Riociguat          -0.232762  \n",
       "T regulatory cells Riociguat          -0.255696  \n",
       "\n",
       "[614 rows x 18211 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Predict using a given dataframe index\n",
    "scm.predict(df_de)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 100ms/step\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></th>\n",
       "      <th></th>\n",
       "      <th>A1BG</th>\n",
       "      <th>A1BG-AS1</th>\n",
       "      <th>A2M</th>\n",
       "      <th>A2M-AS1</th>\n",
       "      <th>A2MP1</th>\n",
       "      <th>A4GALT</th>\n",
       "      <th>AAAS</th>\n",
       "      <th>AACS</th>\n",
       "      <th>AAGAB</th>\n",
       "      <th>AAK1</th>\n",
       "      <th>...</th>\n",
       "      <th>ZUP1</th>\n",
       "      <th>ZW10</th>\n",
       "      <th>ZWILCH</th>\n",
       "      <th>ZWINT</th>\n",
       "      <th>ZXDA</th>\n",
       "      <th>ZXDB</th>\n",
       "      <th>ZXDC</th>\n",
       "      <th>ZYG11B</th>\n",
       "      <th>ZYX</th>\n",
       "      <th>ZZEF1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cell_type</th>\n",
       "      <th>sm_name</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">NK cells</th>\n",
       "      <th>Bosutinib</th>\n",
       "      <td>0.330316</td>\n",
       "      <td>0.138537</td>\n",
       "      <td>-0.154608</td>\n",
       "      <td>-0.042543</td>\n",
       "      <td>0.235437</td>\n",
       "      <td>0.781162</td>\n",
       "      <td>0.114396</td>\n",
       "      <td>0.222130</td>\n",
       "      <td>-0.071732</td>\n",
       "      <td>0.424109</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.011349</td>\n",
       "      <td>-0.301028</td>\n",
       "      <td>0.034625</td>\n",
       "      <td>-0.008219</td>\n",
       "      <td>0.021145</td>\n",
       "      <td>0.125907</td>\n",
       "      <td>0.192161</td>\n",
       "      <td>0.209567</td>\n",
       "      <td>0.025681</td>\n",
       "      <td>0.078541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Belinostat</th>\n",
       "      <td>1.666431</td>\n",
       "      <td>3.450031</td>\n",
       "      <td>-8.108703</td>\n",
       "      <td>-16.189913</td>\n",
       "      <td>-3.965121</td>\n",
       "      <td>6.956760</td>\n",
       "      <td>0.939568</td>\n",
       "      <td>0.900427</td>\n",
       "      <td>4.964502</td>\n",
       "      <td>4.974038</td>\n",
       "      <td>...</td>\n",
       "      <td>5.483314</td>\n",
       "      <td>1.852388</td>\n",
       "      <td>-2.629870</td>\n",
       "      <td>0.311158</td>\n",
       "      <td>-1.745673</td>\n",
       "      <td>0.308468</td>\n",
       "      <td>-0.960009</td>\n",
       "      <td>4.471121</td>\n",
       "      <td>-0.727347</td>\n",
       "      <td>-1.623647</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Myeloid cells</th>\n",
       "      <th>Bosutinib</th>\n",
       "      <td>0.599696</td>\n",
       "      <td>0.033600</td>\n",
       "      <td>-0.592653</td>\n",
       "      <td>0.107676</td>\n",
       "      <td>4.732639</td>\n",
       "      <td>2.272205</td>\n",
       "      <td>0.204302</td>\n",
       "      <td>-0.048383</td>\n",
       "      <td>0.280770</td>\n",
       "      <td>0.398841</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.014911</td>\n",
       "      <td>-0.452582</td>\n",
       "      <td>-0.257989</td>\n",
       "      <td>-0.887243</td>\n",
       "      <td>1.013566</td>\n",
       "      <td>-0.028484</td>\n",
       "      <td>0.012789</td>\n",
       "      <td>0.156828</td>\n",
       "      <td>-0.152137</td>\n",
       "      <td>0.071150</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Belinostat</th>\n",
       "      <td>3.609646</td>\n",
       "      <td>2.358377</td>\n",
       "      <td>-8.683269</td>\n",
       "      <td>0.118509</td>\n",
       "      <td>15.380356</td>\n",
       "      <td>6.253202</td>\n",
       "      <td>1.260448</td>\n",
       "      <td>-1.412510</td>\n",
       "      <td>2.200832</td>\n",
       "      <td>4.408970</td>\n",
       "      <td>...</td>\n",
       "      <td>2.190039</td>\n",
       "      <td>-0.478366</td>\n",
       "      <td>-5.948822</td>\n",
       "      <td>-1.081164</td>\n",
       "      <td>1.773736</td>\n",
       "      <td>-0.035427</td>\n",
       "      <td>-3.334975</td>\n",
       "      <td>0.998982</td>\n",
       "      <td>2.543381</td>\n",
       "      <td>-0.463846</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4 rows × 18211 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                              A1BG  A1BG-AS1       A2M    A2M-AS1      A2MP1  \\\n",
       "cell_type     sm_name                                                          \n",
       "NK cells      Bosutinib   0.330316  0.138537 -0.154608  -0.042543   0.235437   \n",
       "              Belinostat  1.666431  3.450031 -8.108703 -16.189913  -3.965121   \n",
       "Myeloid cells Bosutinib   0.599696  0.033600 -0.592653   0.107676   4.732639   \n",
       "              Belinostat  3.609646  2.358377 -8.683269   0.118509  15.380356   \n",
       "\n",
       "                            A4GALT      AAAS      AACS     AAGAB      AAK1  \\\n",
       "cell_type     sm_name                                                        \n",
       "NK cells      Bosutinib   0.781162  0.114396  0.222130 -0.071732  0.424109   \n",
       "              Belinostat  6.956760  0.939568  0.900427  4.964502  4.974038   \n",
       "Myeloid cells Bosutinib   2.272205  0.204302 -0.048383  0.280770  0.398841   \n",
       "              Belinostat  6.253202  1.260448 -1.412510  2.200832  4.408970   \n",
       "\n",
       "                          ...      ZUP1      ZW10    ZWILCH     ZWINT  \\\n",
       "cell_type     sm_name     ...                                           \n",
       "NK cells      Bosutinib   ... -0.011349 -0.301028  0.034625 -0.008219   \n",
       "              Belinostat  ...  5.483314  1.852388 -2.629870  0.311158   \n",
       "Myeloid cells Bosutinib   ... -1.014911 -0.452582 -0.257989 -0.887243   \n",
       "              Belinostat  ...  2.190039 -0.478366 -5.948822 -1.081164   \n",
       "\n",
       "                              ZXDA      ZXDB      ZXDC    ZYG11B       ZYX  \\\n",
       "cell_type     sm_name                                                        \n",
       "NK cells      Bosutinib   0.021145  0.125907  0.192161  0.209567  0.025681   \n",
       "              Belinostat -1.745673  0.308468 -0.960009  4.471121 -0.727347   \n",
       "Myeloid cells Bosutinib   1.013566 -0.028484  0.012789  0.156828 -0.152137   \n",
       "              Belinostat  1.773736 -0.035427 -3.334975  0.998982  2.543381   \n",
       "\n",
       "                             ZZEF1  \n",
       "cell_type     sm_name               \n",
       "NK cells      Bosutinib   0.078541  \n",
       "              Belinostat -1.623647  \n",
       "Myeloid cells Bosutinib   0.071150  \n",
       "              Belinostat -0.463846  \n",
       "\n",
       "[4 rows x 18211 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Predict from a given multi-index\n",
    "scm.predict(\n",
    "    pd.MultiIndex.from_tuples([\n",
    "        (\"NK cells\", \"Bosutinib\"),\n",
    "        (\"NK cells\", \"Belinostat\"),\n",
    "        (\"Myeloid cells\", \"Bosutinib\"),\n",
    "        (\"Myeloid cells\", \"Belinostat\"),\n",
    "    ], names=[\"cell_type\", \"sm_name\"])\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 89ms/step\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></th>\n",
       "      <th></th>\n",
       "      <th>A1BG</th>\n",
       "      <th>A1BG-AS1</th>\n",
       "      <th>A2M</th>\n",
       "      <th>A2M-AS1</th>\n",
       "      <th>A2MP1</th>\n",
       "      <th>A4GALT</th>\n",
       "      <th>AAAS</th>\n",
       "      <th>AACS</th>\n",
       "      <th>AAGAB</th>\n",
       "      <th>AAK1</th>\n",
       "      <th>...</th>\n",
       "      <th>ZUP1</th>\n",
       "      <th>ZW10</th>\n",
       "      <th>ZWILCH</th>\n",
       "      <th>ZWINT</th>\n",
       "      <th>ZXDA</th>\n",
       "      <th>ZXDB</th>\n",
       "      <th>ZXDC</th>\n",
       "      <th>ZYG11B</th>\n",
       "      <th>ZYX</th>\n",
       "      <th>ZZEF1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cell_type</th>\n",
       "      <th>sm_name</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">NK cells</th>\n",
       "      <th>Bosutinib</th>\n",
       "      <td>0.330316</td>\n",
       "      <td>0.138537</td>\n",
       "      <td>-0.154608</td>\n",
       "      <td>-0.042543</td>\n",
       "      <td>0.235437</td>\n",
       "      <td>0.781162</td>\n",
       "      <td>0.114396</td>\n",
       "      <td>0.222130</td>\n",
       "      <td>-0.071732</td>\n",
       "      <td>0.424109</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.011349</td>\n",
       "      <td>-0.301028</td>\n",
       "      <td>0.034625</td>\n",
       "      <td>-0.008219</td>\n",
       "      <td>0.021145</td>\n",
       "      <td>0.125907</td>\n",
       "      <td>0.192161</td>\n",
       "      <td>0.209567</td>\n",
       "      <td>0.025681</td>\n",
       "      <td>0.078541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Belinostat</th>\n",
       "      <td>1.666431</td>\n",
       "      <td>3.450031</td>\n",
       "      <td>-8.108703</td>\n",
       "      <td>-16.189913</td>\n",
       "      <td>-3.965121</td>\n",
       "      <td>6.956760</td>\n",
       "      <td>0.939568</td>\n",
       "      <td>0.900427</td>\n",
       "      <td>4.964502</td>\n",
       "      <td>4.974038</td>\n",
       "      <td>...</td>\n",
       "      <td>5.483314</td>\n",
       "      <td>1.852388</td>\n",
       "      <td>-2.629870</td>\n",
       "      <td>0.311158</td>\n",
       "      <td>-1.745673</td>\n",
       "      <td>0.308468</td>\n",
       "      <td>-0.960009</td>\n",
       "      <td>4.471121</td>\n",
       "      <td>-0.727347</td>\n",
       "      <td>-1.623647</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 18211 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                          A1BG  A1BG-AS1       A2M    A2M-AS1     A2MP1  \\\n",
       "cell_type sm_name                                                         \n",
       "NK cells  Bosutinib   0.330316  0.138537 -0.154608  -0.042543  0.235437   \n",
       "          Belinostat  1.666431  3.450031 -8.108703 -16.189913 -3.965121   \n",
       "\n",
       "                        A4GALT      AAAS      AACS     AAGAB      AAK1  ...  \\\n",
       "cell_type sm_name                                                       ...   \n",
       "NK cells  Bosutinib   0.781162  0.114396  0.222130 -0.071732  0.424109  ...   \n",
       "          Belinostat  6.956760  0.939568  0.900427  4.964502  4.974038  ...   \n",
       "\n",
       "                          ZUP1      ZW10    ZWILCH     ZWINT      ZXDA  \\\n",
       "cell_type sm_name                                                        \n",
       "NK cells  Bosutinib  -0.011349 -0.301028  0.034625 -0.008219  0.021145   \n",
       "          Belinostat  5.483314  1.852388 -2.629870  0.311158 -1.745673   \n",
       "\n",
       "                          ZXDB      ZXDC    ZYG11B       ZYX     ZZEF1  \n",
       "cell_type sm_name                                                       \n",
       "NK cells  Bosutinib   0.125907  0.192161  0.209567  0.025681  0.078541  \n",
       "          Belinostat  0.308468 -0.960009  4.471121 -0.727347 -1.623647  \n",
       "\n",
       "[2 rows x 18211 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Predict from tuples\n",
    "scm.predict([(\"NK cells\", \"Bosutinib\"), (\"NK cells\", \"Belinostat\")])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 71ms/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Axes: xlabel='y', ylabel='y_pred'>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjsAAAGwCAYAAABPSaTdAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8g+/7EAAAACXBIWXMAAA9hAAAPYQGoP6dpAABVEElEQVR4nO3deXhb1Z038O9dpKvFkrxvieM4JBDSEEoTCiSl0NKkpMBAC5S2tIV2ylMKoWToDAyd6UvovCXQmYH2fZjS6fJm4O0Cb8syvGmBuAthh8RNGkpCSIiTGO+LZMmydKV773n/UKzasR3bsmRJN9/P86iNr5aco6S935zzO+dIQggBIiIiIpuS890AIiIiolxi2CEiIiJbY9ghIiIiW2PYISIiIltj2CEiIiJbY9ghIiIiW2PYISIiIltT892AuWRZFjo6OuDz+SBJUr6bQ0RERNMghEAkEkF9fT1keebjNCdV2Ono6EBDQ0O+m0FEREQZaGtrw/z582f8vpMq7Ph8PgCpL8vv9+e5NURERDQd4XAYDQ0N6fv4TJ1UYWdk6srv9zPsEBERFZlMS1BYoExERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2dlIdF0FERES5EYwmEIolUep2oMzrzHdzxmDYISIioozFkya27unAzsNBDCcMeJwqVi0sw6Ur6uFyKPluHgBOYxEREdEsbN3Tgea93ZAlCfWlbsiShOa93di6pyPfTUtj2CEiIqKMBKMJ7DwcRIVXQ5VPg6YqqPJpqPBqaDkcRDCayHcTATDsEBERUYZCsSSGEwb87rFVMX63imjCQCiWzFPLxmLYISIiooyUuh3wOFWEY8aY6+GYAa9TRanbkaeWjcWwQ0RERBkp8zqxamEZ+qM6eiM6dMNEb0RHf1THyoVlBbMqi6uxiIiIKGOXrqgHALQcDqIjFIPXqWLtspr09ULAsENEREQZczkUXLWyARctreE+O0RERGRfZV5nwYWcEazZISIiIltj2CEiIiJbY9ghIiIiW2PYISIiIltj2CEiIiJbY9ghIiIiW2PYISIiIltj2CEiIiJbY9ghIiIiW2PYISIiIltj2CEiIiJbY9ghIiIiW2PYISIiIltj2CEiIiJbY9ghIiIiWyuYsPPQQw9hxYoV8Pv98Pv9OO+88/DMM8+knxdCYNOmTaivr4fb7caFF16It956K48tJiIiomJQMGFn/vz5uPfee7Fz507s3LkTH/3oR3H55ZenA813v/td3H///XjwwQexY8cO1NbWYu3atYhEInluORERERUySQgh8t2IyZSXl+Nf//Vf8eUvfxn19fXYuHEj7rjjDgCAruuoqanBfffdh69+9asTvl/Xdei6nv45HA6joaEBg4OD8Pv9c9IHIiIimp1wOIxAIJDx/btgRnZGM00Tjz76KKLRKM477zy0traiq6sL69atS79G0zRccMEFeOWVVyb9nM2bNyMQCKQfDQ0Nc9F8IiIiKiAFFXbefPNNlJSUQNM03HjjjXjyySexbNkydHV1AQBqamrGvL6mpib93ETuvPNODA4Oph9tbW05bT8REREVHjXfDRjttNNOw+7duxEKhfD444/juuuuw/bt29PPS5I05vVCiHHXRtM0DZqm5ay9REREVPgKamTH6XRi8eLFWLVqFTZv3owzzzwT3//+91FbWwsA40Zxenp6xo32EBEREY1WUGHneEII6LqOpqYm1NbWorm5Of1cIpHA9u3bsXr16jy2kIiIiApdwUxjffOb38T69evR0NCASCSCRx99FM8//zyeffZZSJKEjRs34p577sGSJUuwZMkS3HPPPfB4PPjc5z6X76YTERFRASuYsNPd3Y0vfOEL6OzsRCAQwIoVK/Dss89i7dq1AIDbb78dsVgMN910E4LBIM455xxs27YNPp8vzy0nIiKiQlbQ++xk22zX6RMREdHcs+U+O0RERETZwrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2VhBhZ/PmzTj77LPh8/lQXV2NK664Avv37x/zmuuvvx6SJI15nHvuuXlqMRERERWLggg727dvx80334zXXnsNzc3NMAwD69atQzQaHfO6iy++GJ2dnenHb3/72zy1mIiIiIqFmu8GAMCzzz475uctW7aguroaLS0t+PCHP5y+rmkaamtrp/25uq5D1/X0z+FwePaNJSIioqJSECM7xxscHAQAlJeXj7n+/PPPo7q6GqeeeipuuOEG9PT0nPBzNm/ejEAgkH40NDTkrM1ERERUmCQhhMh3I0YTQuDyyy9HMBjEiy++mL7+2GOPoaSkBI2NjWhtbcW3vvUtGIaBlpYWaJo24WdNNLLT0NCAwcFB+P3+nPeFiIiIZi8cDiMQCGR8/y6IaazRNmzYgD179uCll14ac/2aa65J/3r58uVYtWoVGhsb8Zvf/Aaf+tSnJvwsTdMmDUJERER0ciiosHPLLbfg6aefxgsvvID58+ef8LV1dXVobGzEgQMH5qh1REREVIwKIuwIIXDLLbfgySefxPPPP4+mpqYp39Pf34+2tjbU1dXNQQuJiIioWBVEgfLNN9+Mn/3sZ/jFL34Bn8+Hrq4udHV1IRaLAQCGhobw93//93j11Vdx+PBhPP/887jssstQWVmJT37yk3luPRERERWygihQliRpwutbtmzB9ddfj1gshiuuuAK7du1CKBRCXV0dPvKRj+Bf/uVfZrTCarYFTkRERDT3bFGgPFXecrvdeO655+aoNURERGQnBTGNRURERJQrDDtERERkaww7REREZGsMO0RERGRrDDtERERkaww7REREZGsMO0RERGRrDDtERERkaww7REREZGsMO0RERGRrDDtERERkaww7REREZGsMO0RERGRrDDtERERkaww7REREZGsMO0RERGRrar4bQERERFMLRhMIxZIodTtQ5nXmuzlFhWGHiIiogMWTJrbu6cDOw0EMJwx4nCpWLSzDpSvq4XIo+W5eUeA0FhERUQHbuqcDzXu7IUsS6kvdkCUJzXu7sXVPR76bVjQYdoiIiApUMJrAzsNBVHg1VPk0aKqCKp+GCq+GlsNBBKOJfDexKDDsEBERFahQLInhhAG/e2zVid+tIpowEIol89Sy4sKwQ0REVKBK3Q54nCrCMWPM9XDMgNepotTtyFPLigvDDhERUYEq8zqxamEZ+qM6eiM6dMNEb0RHf1THyoVlXJU1TVyNRUREVMAuXVEPAGg5HERHKAavU8XaZTXp6zQ1hh0iIqIC5nIouGplAy5aWsN9djLEsENERFQEyrxOhpwMsWaHiIiIbI1hh4iIiGyNYYeIiIhsjWGHiIiIbI1hh4iIiGytIMLO5s2bcfbZZ8Pn86G6uhpXXHEF9u/fP+Y1Qghs2rQJ9fX1cLvduPDCC/HWW2/lqcVERERULAoi7Gzfvh0333wzXnvtNTQ3N8MwDKxbtw7RaDT9mu9+97u4//778eCDD2LHjh2ora3F2rVrEYlE8thyIiIiKnSSEELkuxHH6+3tRXV1NbZv344Pf/jDEEKgvr4eGzduxB133AEA0HUdNTU1uO+++/DVr351Wp8bDocRCAQwODgIv9+fyy4QERFRlsz2/l0QIzvHGxwcBACUl5cDAFpbW9HV1YV169alX6NpGi644AK88sork36OrusIh8NjHkRERHRyKbiwI4TAbbfdhg996ENYvnw5AKCrqwsAUFNTM+a1NTU16ecmsnnzZgQCgfSjoaEhdw0nIiKiglRwYWfDhg3Ys2cPfvnLX457TpKkMT8LIcZdG+3OO+/E4OBg+tHW1pb19hIREVFhK6izsW655RY8/fTTeOGFFzB//vz09draWgCpEZ66urr09Z6ennGjPaNpmgZN03LXYCIiIip4BTGyI4TAhg0b8MQTT+APf/gDmpqaxjzf1NSE2tpaNDc3p68lEgls374dq1evnuvmEhERUREpiJGdm2++Gb/4xS/w3//93/D5fOk6nEAgALfbDUmSsHHjRtxzzz1YsmQJlixZgnvuuQcejwef+9zn8tx6IiIiKmQFEXYeeughAMCFF1445vqWLVtw/fXXAwBuv/12xGIx3HTTTQgGgzjnnHOwbds2+Hy+OW4tERERFZOC3GcnV7jPDhERUfGx5T47RERERNnCsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLam5rsBRERE2RSMJhCKJVHqdqDM68x3c6gAMOwQEZEtxJMmtu7pwM7DQQwnDHicKlYtLMOlK+rhcij5bh7lEaexiIjIFrbu6UDz3m7IkoT6UjdkSULz3m5s3dOR76ZRnjHsEBFR0QtGE9h5OIgKr4YqnwZNVVDl01Dh1dByOIhgNJHvJlIeMewQEVHRC8WSGE4Y8LvHVmf43SqiCQOhWDJPLaNCwLBDRERFr9TtgMepIhwzxlwPxwx4nSpK3Y48tYwKAcMOEREVvTKvE6sWlqE/qqM3okM3TPRGdPRHdaxcWMZVWSc5rsYiIiJbuHRFPQCg5XAQHaEYvE4Va5fVpK/TyYthh4iIbMHlUHDVygZctLSG++zQGAUzjfXCCy/gsssuQ319PSRJwlNPPTXm+euvvx6SJI15nHvuuflpLBERFawyrxNNlV4GHUqb9sjOpz71qWl/6BNPPDHjhkSjUZx55pn40pe+hCuvvHLC11x88cXYsmVL+menk3+RiYiI6MSmHXYCgUD610IIPPnkkwgEAli1ahUAoKWlBaFQaEahaLT169dj/fr1J3yNpmmora3N6POJiIjo5DTtsDN6ROWOO+7Apz/9afzwhz+EoqS24DZNEzfddBP8fn/2W3nM888/j+rqapSWluKCCy7Ad77zHVRXV0/6el3Xoet6+udwOJyzthEREVFhkoQQYqZvqqqqwksvvYTTTjttzPX9+/dj9erV6O/vn12jJAlPPvkkrrjiivS1xx57DCUlJWhsbERrayu+9a1vwTAMtLS0QNO0CT9n06ZNuPvuu8ddHxwczGkoIyIiouwJh8MIBAIZ378zKlA2DAP79u0bd33fvn2wLCuTj5zSNddcg0suuQTLly/HZZddhmeeeQbvvPMOfvOb30z6njvvvBODg4PpR1tbW07aRkRERIUro6XnX/rSl/DlL38ZBw8eTK+Ieu2113DvvffiS1/6UlYbOJm6ujo0NjbiwIEDk75G07RJR32IiIjo5JBR2Pm3f/s31NbW4oEHHkBnZyeAVPi4/fbb8Y1vfCOrDZxMf38/2traUFdXNye/HxERERWnjMKOLMu4/fbbcfvtt6eLfmdbAzM0NISDBw+mf25tbcXu3btRXl6O8vJybNq0CVdeeSXq6upw+PBhfPOb30RlZSU++clPzur3JSIiInvLeFNBwzDwu9/9Dr/85S8hSRIAoKOjA0NDQxl93s6dO3HWWWfhrLPOAgDcdtttOOuss/A//sf/gKIoePPNN3H55Zfj1FNPxXXXXYdTTz0Vr776Knw+X6ZdICIiopNARquxjhw5gosvvhhHjx6Frut45513sGjRImzcuBHxeBw//OEPc9HWWZttNTcRERHNvbysxrr11luxatUqBINBuN3u9PVPfvKT+P3vf5/JRxIRERHlREY1Oy+99BJefvnlccc1NDY2or29PSsNIyIiIsqGjEZ2LMuCaZrjrr/33nusoSEiIqKCklHYWbt2Lb73ve+lf5YkCUNDQ7jrrrvwiU98IlttIyIiIpq1jAqU29vb8dGPfhSKouDAgQNYtWoVDhw4gMrKSrzwwgsnPK8qn1igTEREVHxme//OqGZn3rx52L17Nx599FG0tLTAsiz87d/+La699toxBctERERE+TbjkZ1kMonTTjsNW7duxbJly3LVrpzgyA4REVHxmfOl5w6HA7qupzcSJCIiIipkGRUo33LLLbjvvvtgGEa220NERESUVRnV7Lz++uv4/e9/j23btuGMM86A1+sd8/wTTzyRlcYRERERzVZGYae0tBRXXnlltttCRERElHUZhZ0tW7Zkux1EREREOZFR2BnR09OD/fv3Q5IknHrqqQW7vw4RERGdvDIqUA6Hw/jCF76AefPm4YILLsCHP/xhzJs3D5///OcxODiY7TYSERERZSyjsPOVr3wFr7/+OrZu3YpQKITBwUFs3boVO3fuxA033JDtNhIRERFlLKPjIrxeL5577jl86EMfGnP9xRdfxMUXX4xoNJq1BmYTNxUkIiIqPnO+qSAAVFRUIBAIjLseCARQVlaWyUcSERER5URGYeef//mfcdttt6GzszN9raurC//wD/+Ab33rW1lrHBER5UcwmkBrXxTBaCLfTSGatYymsc466ywcPHgQuq5jwYIFAICjR49C0zQsWbJkzGv/9Kc/ZaelWcBpLCKiE4snTWzd04Gdh4MYThjwOFWsWliGS1fUw+VQ8t08Oknl5dTzK664IpO3ERFRgdu6pwPNe7tR4dVQX+pGOGageW83AOCqlQ15bh1RZjIKO3fddde0XvfLX/4S0Wh03HESRERUeILRBHYeDqLCq6HKpwEAqnyp0ZyWw0FctLQGZV5nPptIlJGManam66tf/Sq6u7tz+VsQEVGWhGJJDCcM+N1j/x3sd6uIJgyEYsk8tYxodnIadjIoByIiojwpdTvgcaoIx4wx18MxA16nilK3I08tI5qdnIYdIiIqHmVeJ1YtLEN/VEdvRIdumOiN6OiP6li5sIxTWFS0ZnU2FhER2culK+oBpGp0OkIxeJ0q1i6rSV8nKkYMO0RElOZyKLhqZQMuWlqDUCyJUreDIzpU9Bh2iIhonDKvkyGHbCOjmp3rr78eL7zwwpSva2xshMPBgjYiIiLKn4zCTiQSwbp167BkyRLcc889aG9vn/B1f/nLX9DQwE2oiIiIKH8yCjuPP/442tvbsWHDBvzqV7/CwoULsX79evz6179GMsl9GIhOVjxPiYgKUUZnYx1v165d+N//+3/jJz/5CUpKSvD5z38eN91007hzsvKNZ2PNjWA0wcLGkwzPUyKiXJrt/XvW++x0dnZi27Zt2LZtGxRFwSc+8Qm89dZbWLZsGR544IFpf84LL7yAyy67DPX19ZAkCU899dSY54UQ2LRpE+rr6+F2u3HhhRfirbfemm3zKYviSRO/bmnDfc++jQea9+PbW9/CD58/iM5QLN9NoxwbOU9JliTUl7ohSxKa93Zj656OfDeNiCizsJNMJvH444/j0ksvRWNjI371q1/h7/7u79DZ2YmHH34Y27Ztw//5P/8H3/72t6f9mdFoFGeeeSYefPDBCZ//7ne/i/vvvx8PPvggduzYgdraWqxduxaRSCSTLlAOjNzwBATC8ST2vDeI/3rlMDY+uhu/bmlDPGnmu4mUA8efp6SpCqp8Giq8GloOBzmlRUR5l9HS87q6OliWhc9+9rN444038P73v3/caz7+8Y+jtLR02p+5fv16rF+/fsLnhBD43ve+h3/6p3/Cpz71KQDAww8/jJqaGvziF7/AV7/61Uy6QVk0+obXOxRH20AMbocCpyqjL6rjN3s6AfDUZDsaOU+pvtQ95rrfraIjFEMoluR0JhHlVUYjOw888AA6OjrwH//xHxMGHQAoKytDa2vrbNqW1traiq6uLqxbty59TdM0XHDBBXjllVcmfZ+u6wiHw2MelBsjNzynKqMjFIfbocCrqfA4FSiyBK+m8l/5NsXzlIio0GUUdr7whS/A5XJluy2T6urqAgDU1NSMuV5TU5N+biKbN29GIBBIP7gMPndGbnj9QzqSpgXNkfqrlTAsOBQZlSVOnppsUzxPiYgKXVEdBCpJ0pifhRDjro125513YnBwMP1oa2vLdRNPWiM3vCHdgGkJDOsmhhMGhnUT9QE3Eobgv/KzqNCWeF+6oh5rl9VACIGOUAxCCJ6nREQFoyiOi6itrQWQGuGpq6tLX+/p6Rk32jOapmnQNC3n7aOUkRvbr3a24Uj/MLyaigUVHpR5HeiP6li7rIb/yp+lQl3izfOUiKiQFcXITlNTE2pra9Hc3Jy+lkgksH37dqxevTqPLaPRRm543/vMWbhu9UKcOT8An0uFIkn8V36WFPoS7zKvE02VXgYdIiooBTOyMzQ0hIMHD6Z/bm1txe7du1FeXo4FCxZg48aNuOeee7BkyZL0MRUejwef+9zn8thqmkhdwI2vXbiYmwtm2fFLvAGgypcazWk5HMRFSzlyRkQ0kYIJOzt37sRHPvKR9M+33XYbAOC6667Df/3Xf+H2229HLBbDTTfdhGAwiHPOOQfbtm2Dz+fLV5NpCjw1Obu4xJuIKDNZOS6iWPC4CCpmwWgC9z37NmRJSo/sAEBvRIcQArdfvJRhh4hsKe/HRRDR3OASbyKizBTMNBadHFjHMzsjRd4th4PoCMXgdaos/iYimgLDDs2JQl0yXWy4xJuIaOY4jUVzotCXTBcbLvEmIpo+hh3KOZ6KTURE+cSwQzk3smTa707NmsaTJkLDCThVmedlERFRzrFmh3Ju5JDQ4HACA9EEOkJxJE0LpiVQWeKEy8HMTUREucO7DOXcyJLpt9rD2NcZgWUJKLIE3bAQTZh4+WBfvptIREQ2xrBDc2LNKZXwOlVoqgxTCMiShNNr/VhW52fdDhER5RSnsWhOxA0LNQENp9f7YInUEmq3Q4FumDzqgIiIcophh+bESN1OwhBjjjoIxwx4nSpK3Y48to6IiOyM01g0J3jUARER5QtHdmjOHH/UgSpLeH9DKdYsrsxzy4oXj98gIpoaTz2nOdcZiuG/d7djf3cEpiV4dEQGePwGEZ1MeOo5FZ2X3+3DrrYQ3A6VR0dMIBhNoLUvesIVajx+g4ho+jiNRXPq+KMjAKDKlxqJaDkcxEVLa07a6ZjpjtbY+TvktBwR5QLDDs2pkaMj6kvdY6773epJvwR9ZLSmwquhvtSNcMxA895uAMBVKxvSr7Pjd8hpOSLKJU5j0ZwaWYIejhljrp/sS9BncliqHb9DTssRUS4x7NCc4hL0iR1/WOoIv1sdd1iq3b7DmQQ9IqJMMOzQnLt0RT3WLquBEAIdoRiEEFi7rCa9NP1kNNPRmmL6DqcquJ5J0CMiygRrdmjOuRwKrlrZgIuW1kyrGPVkKFodGa0ZqdHxu1PBpz+qY+2y8QXHM/0O82G6dTijg95IoTVQ3NNyRFRYGHYob8q8zhPeoKdzs7RTEDp+00WvU51ytGaq7zCfpltwPdOgR0Q0Uww7VLBOdLO8dEW97VbvFMNozXTNdHl8JkGPiGi6GHYo5zIZfZnqZhnVDbzybv+UowbFqJBHa6Zrpsvj7RT0iKjwMOxQzsxm75QT3SwP90XxRuuALTfVs4tM63DsEPSIqPBwNRblzGz2TjnR6iRFlmFagqt3CpjdlscTUXFj2KGcmO3eKSe6WZ7TVI5Sj9NWm+rZUTEtjycie+M0FuVENo40OFHR6sio0chncvVO4WEdDhEVCoYdyols7J1yopslV+8UD9bhEFG+MexQTmRz75SJbpYcNcgPO+1rREQnD4Ydypm5GH3hqMHc4KnkRFTMGHYoZzj6Yh/T3Q2ZiKgQFc1qrE2bNkGSpDGP2trafDeLpqHM60RTpTcnQae1dwgvHuhFa+9Q1j+bUngqOREVu6Ia2Xnf+96H3/3ud+mfFYXD58Ukm/UeoeEE7n1mH95oDSJumHCpCj7YVIZ/XH86Sj0cPcqmbKysIyLKp6IKO6qqzmg0R9d16Lqe/jkcDueiWTSFXNR73PvMPvzh7R74XQ5U+zQMxQ384e2e1HNXnjnmtSyqnR2eSk5Exa5oprEA4MCBA6ivr0dTUxM+85nP4NChQyd8/ebNmxEIBNKPhgbWFuTDbHZSnkhr7xDeaA3C73KgoiQ1rVJRosHvcmBHazA9pRVPmvh1Sxvue/ZtPNC8H/c9+zZ+3dKGeNKc8HOD0QRa+6KcljkOd0MmomJXNGHnnHPOwSOPPILnnnsOP/7xj9HV1YXVq1ejv79/0vfceeedGBwcTD/a2trmsMUE/LXeo8SpQlUkWALj6j1mGjLeC8UQN0yUuMYOTJa4VMQNE++FYgCmH7JOFIpyGYCKKVxxN2QiKmZFM421fv369K/POOMMnHfeeTjllFPw8MMP47bbbpvwPZqmQdO0uWoiTaA7HMfbXWHEkxYsIeBQZNQH3GiscKNzMI7HdhzF4f7hGU1vzS91w6UqGIob0Er++rqhuAGXqmB+qXvKU9NHHxY60UqjZ//ShZ2HBwBIWV9qXYzLuLmyjoiKWdGM7BzP6/XijDPOwIEDB/LdFDqBN1oH0BtJwDAteBwKkoaFAz0R/KU9jL5IAq+3Dsx4equpqgQfbCpDOJ5E/1BqWqV/SEc4nsTZTWVoqipJF9VOdVjoZCuNhnQDLx7oQ9KwsjL1Nlq2p/UmkqtRo1yurCMiypWiGdk5nq7r2LdvH84///x8N4UmEYwmsLczjHmlbrT2DaH/2I03aVroG9LRVOHFKdUl40ZeXjnYj8XVPjSWeya9qf7j+tMBADtag+gajMOhyFhzSlX6+nSLaidaaRRPmggNJ6HKMrwu9VgAmnhUKJPvZLojTpmYbNRozeJKxJNWQYzIsGCciOZa0YSdv//7v8dll12GBQsWoKenB//zf/5PhMNhXHfddfluGk1iJEi4HRIgAUIIJEwLScOCJYCjA8Oo8DlR5nVAlWUYloWOwWEc6o0iFEug2ueadHqn1OPEpr9Zji0vH8LrhwZgWgIJ08R/727HNWcvmPZxFROFonjSRCxhwKkq0JMm4kkTLoeSlaXWuV7GffyU3EA0gf965TB+tbMNNX5XXqfMinH6jojsoWjCznvvvYfPfvaz6OvrQ1VVFc4991y89tpraGxszHfTaBKlbgcUWcJ7oThqfC5E4kkEh5NwaDIsIWAKgXe6huBQZCyrC+Cd7ggO9kShqTIWlHuQMMQJd+nduqcDLUdCiCcthGJJHB2IYXdbCPs6w9j0N8undVzFhKEonkQ4bkCVTexqC6XqjEpdKPc6Z73UOpfLuCcaNQpGkwgNJxFPmDi9LoCEYeVt52PuwkxE+VI0YefRRx/NdxNohsq8TpxW48Or7/ZDkSQM6QYUCYAkoczjQOxY0XJrbxSVXg2HeqMAgEWVJQi4/zq6MdH0zsiNfShuoDeiw6Mp8PqcGBxO4qUD/fi/O9rwxdULp1VUe3wo6g7H4XGmgogiS7AsgX2dEZS6Hbh+zcJZjbxk84DU4x0/ahRLmugYjMHnUmFaAkKIdAjKxpTZTOR6+o6I6ESKtkCZisPl75+HxnIv4kkTumFBllNBx6upKPc6sbi6BAnTwuH+KJKmhSXVJTi1piT9/uMLikeEYkmEhlO1Hx5NgcepQpVlBDwOKDLweutAujh3qqLakZVGt1+8FF9e04T6UjfWLK7EsroAZElCwrQgA9BUGWtOqZz1d5KrZdyjR42A1LRR0rQAATgVOT1VNNl3mkvTLRgnIsqFohnZoeJUV+rG1WfPx1O72iEAOBQZqgJE4yaW1PhQ5dNQ7dPwybPm4Yld7XA7VKjKXzP4pNM7QiCeNDGkJ1Hjd8EwLRiWQMKw4HGqMC1rxvUvZV4nQrEkTEugxu9ElU9DwjTRHopBliX0Dul4anc7vrSmaVY1Jrlaxn38qJFTlWCaAsOmidPrfOk252PnY+7CTET5xLBDOXfpinokTQs/fuEQOgdjkCUJHqeC/qgOAYGLl9fi/FOr0R3Rp5zeGV3k2jeko38ogcFYEg5FhmkKWAKo9jvhczkyuoGOvin3DsXxXjAGt0OBqaSC1ButA6jyaVmpMSnzOrM+dTN6Sm4gmkRliYZowkC51wndMLM2ZTZTuZy+IyKaCsMO5VwsYWJwODXKMqQbGIobiMSS2NsRxrI6Hz52eg0AYM3iSvRGdBzojmAgqkORU9NGIzfwYDSBx3YcxeutA6gLuHHuogp0h+PojybhUCx4nSo0RUI4ZkCRkdENdOSm/Js9nTg6MAynIkOSgETCwqKq1DL5Qq4xOX7UyKXKePndvhMWaWcik+Xj0ykYJyLKBYYdypmRUZgX3unFiwf6EIknYViAhNRDllKFv9966i9YvbgCf24bRCSeROdgHIZlocbnwt7OMJ7c1Q5A4I3WAbQcCR6bCpOwoNyDSp8LQgC6acHvTtUC+bRUQW4wmphxIAlGE1g+L4C2/mG82zuUaqesYFFVqpbIPFZnc/wUWaHtHTN61Oj4KTMA6ByMZ9TW2Swf5y7MRJQvDDuUM4/tOIrmvd0IDicwpBswrNR1cexhCcA4thT6ne4IVjaWI5ow0BNJnVRf4dUgSxJ+/toRQAIWVnjhVGQ4FBmHeqMY1k2YlsD8cg9Cw0ksqS5BfakbbqcyLpBMFUaOv4krsoTKEg01fhcayj1wH7uRByP6mBqTmdz88xmIyrxOuJ3KrPe5ycby8VxM3xERnQjDDmVdPGni/+5ow8OvHEHCNDF4rOh30tcbFnojOhyKhN5IAqVuByQJ6InoWFDuQezYKeXlXic0hwLTEpAlCT2ROCRJQkcohoQhcKgvio7BOHwuFQ1lHpS6HdMOIxPdxBOGhbbgMMo8TsgSJqwxmc7Nv1A205ttUJnL5eOFNlJGRMWNYYeybuueDmzb2w3TsuDTHAhGEzhB1gEA6IaF3iEdSdNKFQRbqdVWvREdCcOCKkswji2jbj92qjmEgEORMKSbKHU74HbICMeS6A7HsLAiddTEr1vaprzBT3YTf988Pw73RRFPGhjSk+NqTKZ78y+EzfSyEVRyvfszUDjBkIjshWGHsmrkplrr1xAcTsA6lnIkpKauJhNLmIjEkghGE+hMmhAATEsgEjcwFE9CkiS80TqApGkh4FIRjhtIWgJJ04LfpUBRJBwdSIUgTZXxbu8Q3u4YnNYNfrKbeJnHiZjfxBfPW4iAxzlulGE6N38ABbGZXjaCylwsHy+EYEhE9sNNBSmrRm6qlT4N9aUuxJImLHHioAMAFoAXD/ajdyiBiG5iSE9tQjgwpEM3LMSTJg71RTEQ1WEJAcsScDlkyJKEpCWgyDJq/C40VXpR6dPQNhDDE7vbERpOwBIiPRUGjN/I7vjN+EaM3MQbK7wTbko41ftK3Y6C2UxvOm2dyshKtf6ojt5I6rT53oiO/qiOlQvLZh3aJjuBvsKbWgGX7RPciejkwbBDWTX6pnpqjQ/1pS4IISBl8FmWAAwBmAKQpNTPEd1C96AOv1vF/DIPVFnCUNxEwjDhc6UChRACHqeC1w71Y19XGK++24+XD/Zhb0cYhmmNu8FnehOfzvuyETKyIVtBJVe7PwPcZZmIcofTWJRVx28et6Dcgz+3haAqApoqY1i3YGXwuYaVSuYWAMipnZg9ThVup4ohPTUFdtiMwjh2BpRXUxHVDdQF3AjFktCTJg72RBCJJxHwOMZtZDfVHjCTFcxO9b5C2kwvG/vc5HL5OHdZJqJckYQQU80w2EY4HEYgEMDg4CD8fn++m2NbI0Wmrx/qx573BnGwZwhmFv+WKRLgc6moC7gQTZjojcSRMATcTgWaKsOpyBiMJVGiqbjs/fU4OjCMzlAckXgSqizji6sbcc3ZCyYseD0+1HQOxvDUrna80x2BaYlJC2ZPtHpo5PtoORxENJG6ca/MY9FtIa90Gl1QfnwwZM0O0clrtvdvjuxQVo3cSC9aWoNh3cSfjoTgUCSYRvbSjhBA0hRIWgJNFSUYTpgADNT4NJS4HBhOmAgNJ6EoEhRZxrK6ABZVlqA3EkdvRMeZDWWThoyRPWDiSRO/bmnDr3a24Uj/MLxOFQsqPHA5lAkLZk+0d0yhbaZXyPvccJdlIsoFhh3Kiok25TvaPwwLAj6XA/Gh7BSXSkjV7zgUCfMCbrgcMiQAdQEXHKqMWNKEIgOlXgcciox40oRDlnCoN4pDvUNIWhYefqUVBxdXnnBkZeueDvxmTyf6hhIo8zihKqn+qLKc8ZERuQoZhTxSM1OFFgyJyB4Ydigrjl8y/F4whiMDwwAEVCl7dfACqYLlUMzACwf6oEiAU5WhORQ4VaCqRMPSWh/e6gyjbWAY0biBtoFhHOgZAgAsrvbC7VBTQSai45qzF4y7mY6sCirRVCiyBI+mQJVTfegYjGFemQsD0WRW9pWZjenuSVOMYaiQR5+IqPgw7NCsTbRhXbnXCU2V0R2Ow8ykInkaBFKrtYykBQzpkCQNw/owTCFQoqk4f0klEoaFd3uH4FJlNFV5saiyBIf6hnB0IHX21f7uCNYcN8ozsiqookSDoy8KPWlB1WQ4VRmh4SQO9w+j3DPxqeqzDRYzef9Ue9Jwgz4iohSGHUrL9EY9esM6w7TwTvcQ2kPDGBxOpM/DyrXhpAU9GIPmSE1l3XjBKbj23Ea83RlGKJbEgnIvAm4H9nYO4lBvNHWaOYBo3MRTu9oR1Q1ct7oJwF9XBSUMC/WlLhzqjcISInXGV9xAJJ5EU6UXv3+7Ox0cJgoWy+r8+GBTOWr8rim/z5kGk+nsiPz7t7u5QR8RERh2CLPfon/0kuGOUAwHeiJIGBb0uUo6x5gCMEwLw4nUCh6XQ0FjhRfVPhcSxzYm7AjF4XYoEEIgljBxJBhFPGHh4VeOQIKET5/dMGa5eLnXiaRpYW97GINxAwGXimX1ftT4XWOCw+hRlmq/hr+0h/HigT48ses9LK31T/l9znTn4Kl2RD7SHy2InZuJiAoBNxWk9I1WliTUl7ohSxKa93Zj656Oab2/zOvEmQ2l2HlkAK8e6kN3OI6+oQSSc5t1AABJExhOWPj5a0fx0PMH4HYqWFbnx9GBYRzujyKeNGEKgZ6IjoRpwanIKPM6YFoWto3q88jmeTIkuFQFHk3F6bU+XHZmPc6cX4Zavzu9s29r79CYYHGkfxi9ER1uhww9aSF57GT3yb7PTHYOnmqzQkgSN+gjIjqGYeckN9st+oPRBFr7oojqydS0lSlgWFMfD5Er4thjMG7gwd8fwOd/+hp2t4UwpCfxdlcYwWgCw7oBpyqj2q/B41RhmKkVY7X+v/Z5ZFXQ7RcvxbXnNmJprQ/nnlKBEtdf63RGgsN7oVg6WIwePQp4HDCFgNelnvD7zGTn4Kl2RG4s9xTEzs1ERIWA01gnudHTIfGkiXjShMuhTHlAZGcohl++cRR/fi8EPWnivWAcumFldCxEto20QTcE/nx0EBAS3t9Qhs5QDAd7hiAgAEhwKDKiuoFY0sSiqtSZWsf3uczrxPL6AEo9zkl39p1f6k4HC4ciIWla8LlSNT8ORYbLoUB2YtLvM9Odg0+0J43LoRTMzs1ERPnGsHOSK3U7oKkKdreFEIkbSJqpG7TPpaKhzDPuRhtPmnhyVzv+c/tBtAdjsAQgS4BpIaNjIHIhFWUAWZZgWQKH+6MY0pOwBGBBwBICQqRGpXwuBxZVeXFqjQ/BaHLCcDHVkQ9NVSXp532aClmSMDichBDAoqoSuB0KeiP6pMEl0yMlptqThhv0ERGlMOyc5Mq8TigycKh3CH6XAyUuFUNxA72ROBZWeNI3z5GVWi++04tfvHEE7aE4BCSoCpAwRN6mrU7EtFKhZihuoERTEXA7oEgSgsMJnDEvgFjSQq1fQ6VPQzCanDBcjPR7zSmVACYPDiP//crB/tQqr4SBxnIvGivc6emlEwWX2QSTyfak4QZ9REQpDDsnuWA0AcMSWFRZgohuYDhhQnMoWFSiwbQEOgdjePlgH3YeDiI0nMC+Y3UvwhJQZECBBElKHeFQSEY2HwRSK7QGoglE4km4HAo8TgVup4zzl1RhX2d4wnAx2Qq1r39sCeJJa9LgICBQ49dSK74g0DkYh9/lmDK45DKYcIM+IjrZMeyc5EKxJBKGhfcvKIUlkK7ZkaVUjclTu9qxuy2ECq8Gn1tFMJpEOG4CAAwTyF8p8vSZAtANE/EkEI4bWFjpASDh/FOrcNmZ9ROGi4mWgv9mTyd6Izoufl9tumh45D2jX19T50I4ZqBzMIbl8wL4zAS7NE+GwYSIKPsYdk5yY4tjUyMSANAb0aHKMt7pjqDCq6HM68C2vQMY0o0pPrHwiGP/IUsAICEUTUCV5XTAmey4iJEVaoZpoTei43B/FH9uC+HRHUdR5nFicXUJzl1UgTWnVE66p83R/uE57SsREY3HpecnucmWMHcOxlCiKeiL6HCqEt5qH8TR/iiswh/ImZBTlRFwO+ByyBjSTTSUuScdQTl+Kfg73UM41DsEPZnamBACGBxO4r2BGJr3duO/d7fPaOn4yHL9qZb1ExFRdnBkh8YUx+5pC+HowDCShoV9nWFE4gb2doYxnDCgG8WTdBwyxm5qKAHJY0nN7VCwZnHlpO8dPdpV4kod/ulQZQxEE9AcCipLNCRMCxHdwLxSN97pHoIiS1MuHedZVURE+cGRHYLLoeBDiyuxrzOM11sH8F4oju6hBHrCOmRJoD+aRCxZPEHHKadGckb225ElYF5pasdjj1PFkpoSLKnxTfr+0aNdHaEYhnUDUd2Abpjwu1WoSupQ0KRpwaHKMCwLp9b4Jt3g7/i6nkx3qiYiosww7BAA4KafteDP7w2mVzABqX1zInqh7J4zfaaF9LlcEgCPQ0HSFBBCoNzrxBVnzZ+yCPjSFfW48LQqdARj6Ino6B/SAQBJQ8CyRHrDwKRhwetUccVZ87B2WQ2EEOgIxSCEGLMCa7Y7VRMRUeaKbhrrBz/4Af71X/8VnZ2deN/73ofvfe97OP/88/PdrKIVT5r4wR8P4s/vDRbBuqrpsQBIAnA7JCyt86Mu4Ebs2HLx1YsrprV3jcuhwKHICHgcWFJdgoHhJOJJA8FhHYaVCjg1fheGEgbWLq5BXcB9wqXjUx3cOdlO1URENHtFFXYee+wxbNy4ET/4wQ+wZs0a/Od//ifWr1+PvXv3YsGCBfluXt6MbHyXyd4sW/d0YNve7jEjOsXMrUo4Y34ApgAcsoyrVs3PaO+akZGYuoAbiyq9eLsrgr4hHaaVOi29xu9CQ7kb5ywaG54mWzqe6ZEQREQ0e0UVdu6//3787d/+Lb7yla8AAL73ve/hueeew0MPPYTNmzePe72u69B1Pf1zOByes7bOhdkWvHYOxvCrnW2IxO1xArYMQHMoCMcNWCK1g/KvdrZhzeJKNFV6Z/RZoVgSkXgSoVgC3WEdQgCqIuGU6tTxD185fxGW1wdmtH8Oz6oiIsqPoqnZSSQSaGlpwbp168ZcX7duHV555ZUJ37N582YEAoH0o6GhYS6aOmcyKXgdvez5qV3tONI/DI+zqDLvhBQJ0FQJsixBlVNnezkVGUf6h/HUrvYZf55LlbGvM4xdRwfRE9YRGk5gKG6gPRhDPGnOKOiMuHRF/QnreoiIKDeK5i7X19cH0zRRU1Mz5npNTQ26uromfM+dd96J2267Lf1zOBy2TeA5vuAV+OtGdi2Hg7ho6djRguNHgRRZwpH+YWgOGe3B4t34TlOAar8LA9EkkqYFyxIYTphwOxSoigSvpuJAdwTBaGJG4aR5bzeCw6kRL6eaWtcVjhtQZQkQmZ3tzrOqiIjyo2jCzghJGnujEUKMuzZC0zRomjYXzZpzMyl4DUYTeGzHUbzeOoC6gBv1pW60B2PoCMVgmhaGk8W34goAltaW4KqVDXh+fw8C7iQO9UUBAANDOnTDgkuVsaDCA8MSMyoADkYTeL21Hz6XioAkIaobMIWAKgEuhwy/W51VQTGPhCAimltFE3YqKyuhKMq4UZyenp5xoz0ng+kUvI6M5rx8sA8tR4JwKDJURUKZ14f6UjdcDhldseJc8lzpdeCfP7EMW9/sxIJyL06rkREcTmBYN2FJAlHdwJLqUlT5NCiSNKYAeKqC7lAsCdMS8DhVqLKEcq8ThiUgAQgNJ6GpMguKiYiKSNHU7DidTqxcuRLNzc1jrjc3N2P16tV5alX+THbMw+iN7EZqepKmgFOR4VRkHOqN4p3uCFyO1D4vySLaFXlEmUfFh0+tQolLTR/TUOJyYEm1D26ngjK3E2UeB0pcKgZjyfT3EU+a+HVLG+579m080Lwf9z37Nn7d0pY6AmKUUrcDpR4nSj0OxJLmsT11UiM8hmXh7KZyjswQERWRohnZAYDbbrsNX/jCF7Bq1Sqcd955+NGPfoSjR4/ixhtvzHfT8mL0MQ8doRi8ThVrl9VgzSmV2H00iJcP9qHCq6HEpaK1LwpZAhRZQWcojgVlHnSF9aJbcu53qagLuLF6cSUaK7zwOFUEhxMYiCbQHYnDFAL90QQcqgRZwpgC4IlOMh9ZHXXVyr/Wco0Eyb5jGwkODifRF9FhWsD5Sypxzdkn7zYHRETFqKjCzjXXXIP+/n58+9vfRmdnJ5YvX47f/va3aGxszHfTcuZEUy7HF7y6HDJePtiH7//+AHoicRzsGcIpVSVY5vGjPuDGgZ4IFFmCaQm8eqgf7QOxPPVq5hQJ8LsdOLXahys+UJ9eXr9qYRn+6+XDCMWS8Gkqyr1OKFISXk3BysaydIiZqKDb5wIi8SReOdg/rqB7dJAMuhNQZBnnNJXj02c38BwrIqIiIwkhiuzf9pkLh8MIBAIYHByE3++f0997phv/ZbKHzq9b2tIjF05VxosHeqEbFk6tTp0Dtb87jNBwEpYQMEwLxVCXrMnAsnkBnHdKJS48tQpLanxjvr/OUAwbH92NvqgORZbgUGTUB9wo8zqgSBJuv3gpyrxOtPZF8UDzftSXuqHIEt7pjqAjFEc8aSJpWvjsBxfgS2uaxn23s9mwkYiIsmO29++iGtkpRplu/DfdKZcRwWgCLx/sg0ORUeJS4XYoWFTlxb7OCN5sD8HtVOB2KJC9EhKGgZ5I4SedEqeM1adU4s5PnI6mqpIJXxM3LNQENJxe74MlUqNdbocC3TDHrEobXdDdOxTHod4o3A4FTjVVtvZG6wCqfNq475Yrp4iIil/RFCgXq0w3/pvJoZHxpInHdhxFy5Eg3u4M4+WDfdjbEcaiSi8aK9wYTpgwTAHXsQDkdjogZ7ZVzJzRFAlrllRCliVgkq0FgL+uSksYAmUeJ9zHAuTxxzCM1OF0DsZwqDcKpyJDkoBE0sKiyhLUBdw8kJOIyKYYdnIo05OuR/bQ8bvHDrz53SqiCQOh2NjjHbbu6cDrrQOQJaSPSTjUO4RDvcMo82go8zhwzqIKrDlW1KsUeNJRJKDSp+HtziG80z2EF9/pHbdiasR0VqWNuHRFPc5pKkfStFIbEApgUVUJTq0pmfS7JSKi4sdprBzK9KTrmRwaGYwm8PqhAcSSJpKmQGg4Dqciw+1UcLA3gnmlblT7XTDM1LSVy6FACFFwq7AkAAKp9O13OaCpMiwBVPs1vPxuHzyaMuH0HTD5qrTjj2FwORRcc/YC7O+OIGkK1Je60yNBwYjOAzmJiGyKYSeHMj3peiaHRoZiSRzoiSA0nESpxwFVkRCOGQgNJ+ByKFhW50NUN/Dqu30ocTmwsNKD3kg8tx2foaoSBxaUe9ETiWM4YaJEU+FxqqgrdeHUGh+C0eSER2CMmMkxDGVeJ9YsrkTz3m4MxQ3IEnggJxGRzTHs5NBsTrqe7mgFhEA4ljqzqURzoERzoNJroW9Ix1DcwDvdQzAsAVmW0D+kozcSQzRRWMM6EiTUBVz42LIa/OlIEJUlGvxuR7qAe6qRsBHTLSae9ndLRES2wLCTY5neWKc9WiFJ8LtVDA6npsycqoyEaUFAQixpIhI3UOXTUO51IqqbONofRWrCqDBUelWc3VSBb6w7FaUeJ470D0OSpDEr1aYaCZspHshJRHRyYdjJsdneWKcarSh1O7C4ugTvDcQQ0Q1E4gYcigy/S0VfJFX/4nGm/pg9ThRccfKiKh8+srQqvbQ805GwTHBZORHRyYFhZ47k6sZa5nVixbxSdA3qWFTpgVdzIGlYeC80DJdDGRNuTEsgaRbO/jr1AQ2fPrthzCgXp5iIiCjbGHaK2MiGhXvaQxjSk+gIxeB3q1hS7cMnzqjDU0Y7OgbjEBDwaiqGE+axQ0FT709MvJp7Tvg0BfddtQLnL6kec51TTERElG0MO0Vs9C7L5y6qQG9YR3ckjtNqS+DVVKiKjFjCRDSehENVoKmpUR4BCZoiI2HmJ+14nDI+srQKy+tLJ30Np5iIiChbGHaK1PHHQ2iqgvnlHmgOBc1v9cCjKVhQ4UE4nkR7MIaobiCqHytNFgIJ04IiIS/77ciShNPrAgwzREQ0J7iDchEaOR7ijdYB7GkLYfs7PdjbEYZhWrCEhaPBKJyqjPCwAcMQaCjzwOdSYVoCiixBVSR4nAocytwWKysSoEqAU5HxkdOqp34DERFRFnBkpwg9ues9PLW7HeFYEoCEiG6gP6LjveAwhnQDg8NJHOiOIG5YKPM44FIVmBYgSRK8TiUdehRJBpCaypIB5LJ0WQbgcSqwBDCvzA3tBIegEhERZRPDThYEo4l0MS2AnBbWBqMJPPWnDgzpJvxuB6K6AQiBiG4iHB9CuccJv8cBwxQYHE7CKUtQPTJMy4IipwKPQ5VQWaIhnjQRHYgBAGQJECJ3O/BIEhBLmijzOLCkuoTHMhAR0Zxh2JmFkdVQOw8HEYkn0RdJAJJAZYkGn8uBVQvLcOmK+jEb5M3WkYFhdIZj8LlUlGgq+od0DMaSqdobIbCgwgOXQ8bB3igkKRW8nKoCQILbocAwLZgW0BOJI55MjeUoSB0gmsvyHZcqw62l2uFUZdbrEBHRnGHNziyMrIaSJQnDCRPtgzG0h+KIJgzIkoTmvd3Yuqcju7+pOBZJRKrQt8rnQq3fBU2VoampP87eoQRMS8C0LAzpJmJJE/PKXPA41fT1pCEgAXAqEiBlfwpLOvYY+TUkCZqioKnSC9PCpCe+ExERZRvDToaC0QR2Hg6iwquhxKWiJ6KjzONAqduBvkgCPpeKCq+GlsPBrN7YGyu8qAu4ENENRHUDhmXBFCJ1YrgsoSeiQ5UlzC9zo9rngscho9zrxClVJagLuODVVFSWOFFX6sJZC0rRWOHJWtuAsQFHANAUCfNKXZhX6oZDleHRFOiGiVAsmdXfl4iIaDKcxspQKJY6i6q+1H1ssz4LPlfq64zEDcST5rQPsJyJMq8TV5w1Hz9/7QjiydSojQTA71IRP/ZrIYBh3YSqyFg2L4D6gAtfPG8hIEn44fMHUeZxpg/afPXdPggRzUrbNEVCqceJSDwBwwJUWUKZV0NdwAVZlhDVDbw3EMOK+RprdoiIaM4w7GSo1J06cyocM1DiUuFQZCQMC0Kklla7HErWD7Ac8cmz5sGhSHjlYH+6GLqh3IXH/9SBgWgCfUM6FFnGvDIXltb6MRDVEfA4Uep2oNTjHHPQ5oIKD3YdDcLKwjyWqkgo8zigKhIMU2BhpQd9kQTihgmnKsMUAkO6gVNrSlizQ0REc4ZhJ0NlXueYQyurfRoO9AwBABZXexGJ5+4Ay+OPVIAQeHJXO0zLQsDjSNfuxJMW3u4Ko6nCm14ddvxBm8ICNFVGMjH7tJMwLHRH4qjxuxFwOdBY4YHf5UTHYAyRuAHTEmgs9+Ly98+b9e9FREQ0XQw7szD60EqPU8G8gBuQBLxOFUKInB9g6XYq+P3b3Xj5YB92HB5AJG7AMAUcqgynIkGRJcSSJj6xvC4duCY6aPODTRX44/7ejFdjSUgtLXcoMlRZxiVn1AIS8KcjIdQF3JhX5kLfUAJR3cAlK+pQV+rOzhdAREQ0DQw7szDRoZVAbvfZGW1kNZhTkaEbFiAEBASEEDAsQDcslHsVfLCpfMxeQMe3OZ408en/fAVtwfi0f++RynYLgFOV4XcpmB/wIGFZeLsrAgAYThh4t2cIlT4n/C4HLjytiqeXExHRnGPYyYLjD62ci3qU0avBHIqEpCHgUBV4FAmmJVDjdyF27FjzP+7vxpH+GIYTBjxONb3/z+h23vSRJfj3595GX3R6q6RGJr0kAJUlTgTcDhwODgMATq/zo9KnIeB2onMwhjPmBXDN2QtYp0NERHnBpedFamQ1mN+dyquaQ4YlBExLwBJA0rQgAJhC4OWD/ZAlCfWl7kn3//nkWfNw44WLUaJNbwPEkSXmTkWCQ5EghIBuWHDIEipKNGiqgiqfhrqAG0f6h7PYcyIioplh2ClSo1eDuRwKKrxOlGgqDNNCwkiNu1SWaFAkCTU+F6p8fw0gE+3/43Io+Mr5i/DZcxZgquNBJQAORUKF1wG3U8FANIm4YaFEU1BybAn8CL9bRTRhcF8dIiLKG4adIjWysqo/qiMSN1BR4sRwwkTCFKnVVabAYCwJn0tFlV8b897JAkg8aaLW54LbOflfC1UCXGrqyIdqnwuNFV6Ue504c34pfJoD8qhl7QBytvyeiIhouhh2itilK+qxdlkNhEgFGwDwaSpKPQ6osgRVluB0yAjHjDHvmyyAbN3TgZcO9uGUqhK41YnHdwwB6KmDuBCOJxHVDbgdChRJgltT4HIoiMQN6IaJ3oiO/qiOlQvLWK9DRER5wwLlIjayGmzlgjL827b9OLXah4pjp5mPhI53eyPoHEydbO53p6a9Jtr/p3Mwhl/tbEPfUAKyBJR5NVhD+rFgM5YlUnv4VJbIiCctlLhUOFUZ156zAICEPW2h9LL2XC+/JyIimgrDjh1IqT11Ko/V5YxMI0lSqm5n+bwAjvYPjwkgaxZXorUvml4i/9SudhzpH0aZxwmXU8ZwwoQsS8CosKMcG+yxBGBagGkJfP7cBbjwtGrU+F3p8HTx+2rnbPk9ERHRVBh2Cszo/XCmGxRGFytX+cbWy/hdDnzm7AUAUiu4XKqMl9/tw/d/dyC9FP30Oh/2dYbhdapQFQmDw0kMJ0wo8tipLElKHYWhyBKSpoWA24FzF1VgaZ1/XNubKr1Z+kaIiIhmp2jCzsKFC3HkyJEx1+644w7ce++9eWpRdsWTJrbu6cDOw8Fx++GMLvidyETHQEw0XVXmdeLXLW1o3tuNCq+G+lI3wjEDzXu7MaQbWFDhQWtfFIPDCciyBMexsKNIqXOvJEgo0VQMJQwYlkB/VMfDrx7Gvq4wAAl/bgvNuO1ERES5VjRhBwC+/e1v44Ybbkj/XFJSksfWZNfIbsjHhxAAuGplw5Tvn+gYiOPrZUZvRFjlS63QqvIp0JMmOkIhNFV4Ma/UjYFoAhAAJAluVULCFDAtQJIsDMaSSJgCboeMpbV+uB0qfv76UUAAK+aXZtR2IiKiXCqqsOPz+VBbW5vvZmTdZCEESIWXi5ZOfZjoZEdXdA7G01NiIxsR1h93NlWVX4PfraJ3KIG6gAs9ER3xpIHhpImAx4monsSQntq7R5YESjQFKxeUYVmdH0lLIKabkCTA51KP7eUzs7YTERHlUlGFnfvuuw//8i//goaGBlx99dX4h3/4Bzidk99IdV2Hruvpn8Ph8Fw0c8YmCyF+t4qOUAyhWHLagaHM64TbqUw4JbZmceWktT1LqktwxrzSVO2OpqAvogMSEPCmgtLAUAKSJOBxqvjwqVUIuFPtiegJSFJqIGhkFVimbSciIsqFogk7t956Kz7wgQ+grKwMb7zxBu688060trbiJz/5yaTv2bx5M+6+++45bGVmTlRgnMmGfCeaEjtRbc9VKxsQjCZwoDuCe595G4PxJAQAWZKwYn4pPJqMN98LYyhupMOOy6FAiFTxMjcTJCKiQiQJIcZvpDJHNm3aNGUY2bFjB1atWjXu+uOPP46rrroKfX19qKiomPC9E43sNDQ0YHBwEH6/f3aNz7LRhcMThZDpCkYTuO/ZtyFLUnpKDAB6IzqEEPj6RUvw8rt9aDkcRDSRCiQrjysmbu2L4oHm/Sj3OmGJVIhxOxTohonXDvWjRHNgQbkn3c497aF0zc5s2k5ERDSRcDiMQCCQ8f07ryM7GzZswGc+85kTvmbhwoUTXj/33HMBAAcPHpw07GiaBk3TJnyu0EynwHg6ppoSixvWuNqe46eZRkaaEoYYE5iOn+4aaSc3EyQiokKW17BTWVmJysrKjN67a9cuAEBdXV02m5Q3ExUYZ1LrMt0psTKvc9LPn2gpe29YR3ckjrXLavDF1Qsn3A+ImwkSEVEhKoqanVdffRWvvfYaPvKRjyAQCGDHjh34u7/7O/zN3/wNFixYkO/mZdWJQsh03z+dPXemMjIq8/qhfrx2aDC1QaFbxV/aw/h1SxsuXVE/7rNm23YiIqJcKIqwo2kaHnvsMdx9993QdR2NjY244YYbcPvtt+e7aQVpulNiJ9qteWSkaVg30TmoY3FVCSp9GvfQISKiopPXAuW5NtsCp2IzWZiZ7m7NUxU7337xUo7kEBFRzs32/i3noE00R4LRBFr7oghGExM+X+Z1oqnSOy6QjCxNlyUJ9aVuyJKE5r3d2LqnY8zrRoqd/e6xA4B+t4powkAolsxuh4iIiHKgKKaxaKzZnKM1k92as73/DxERUT5wZKcITXdkZiIzGa0ZKXbuj+rojejQDRO9ER39UR0rF5ZxCouIiIoCw06ROX5kJnUWlYYKr4aWw8FJp7RGjB6tGW2y0ZpLV9Rj7bIaCCHQEYpBCME9dIiIqKhwGqvIzPYcrZkuTc/W/j9ERET5wrCTZSdazp0N2aijmc7S9OP7wT10iIioWDHsZMlsioZnIhubBp5otGau+kFERDRXWLOTJbMpGp6pbNXRTLQ0fS77QURENBc4spMFM1nOnQ25qqOZ634QERHNBY7sZEG+Nt+bbNPATHETQSIisiOGnSyY6XLuQmWXfhAREY3GsJMFdtl8zy79ICIiGo01O1ky3ZPGC51d+kFERDSCp55nWa732ZkrdukHEREVv9nevzmyk2V22XzPLv0gIiJizQ4RERHZGsMOERER2RrDDhEREdkaww4RERHZGsMOERER2RrDDhEREdkaww4RERHZGsMOERER2RrDDhEREdkaww4RERHZ2kl1XMTIMWDhcDjPLSEiIqLpGrlvZ3qc50kVdiKRCACgoaEhzy0hIiKimYpEIggEAjN+30l16rllWejo6IDP54MkSfluTlo4HEZDQwPa2tpydhp7vtm9j3bvH8A+2gX7WPzs3j9gfB+FEIhEIqivr4csz7wC56Qa2ZFlGfPnz893Mybl9/tt+xd3hN37aPf+AeyjXbCPxc/u/QPG9jGTEZ0RLFAmIiIiW2PYISIiIltj2CkAmqbhrrvugqZp+W5Kzti9j3bvH8A+2gX7WPzs3j8g+308qQqUiYiI6OTDkR0iIiKyNYYdIiIisjWGHSIiIrI1hh0iIiKyNYadPPvOd76D1atXw+PxoLS0dMLXHD16FJdddhm8Xi8qKyvx9a9/HYlEYm4bOgs/+MEP0NTUBJfLhZUrV+LFF1/Md5My9sILL+Cyyy5DfX09JEnCU089NeZ5IQQ2bdqE+vp6uN1uXHjhhXjrrbfy09gMbN68GWeffTZ8Ph+qq6txxRVXYP/+/WNeU+x9fOihh7BixYr0ZmXnnXcennnmmfTzxd6/iWzevBmSJGHjxo3pa8Xez02bNkGSpDGP2tra9PPF3j8AaG9vx+c//3lUVFTA4/Hg/e9/P1paWtLPF3sfFy5cOO7PUJIk3HzzzQCy2z+GnTxLJBK4+uqr8bWvfW3C503TxCWXXIJoNIqXXnoJjz76KB5//HF84xvfmOOWZuaxxx7Dxo0b8U//9E/YtWsXzj//fKxfvx5Hjx7Nd9MyEo1GceaZZ+LBBx+c8Pnvfve7uP/++/Hggw9ix44dqK2txdq1a9PnshW67du34+abb8Zrr72G5uZmGIaBdevWIRqNpl9T7H2cP38+7r33XuzcuRM7d+7ERz/6UVx++eXp/xMt9v4db8eOHfjRj36EFStWjLluh36+733vQ2dnZ/rx5ptvpp8r9v4Fg0GsWbMGDocDzzzzDPbu3Yt///d/H/OP4mLv444dO8b8+TU3NwMArr76agBZ7p+ggrBlyxYRCATGXf/tb38rZFkW7e3t6Wu//OUvhaZpYnBwcA5bmJkPfvCD4sYbbxxzbenSpeIf//Ef89Si7AEgnnzyyfTPlmWJ2tpace+996avxeNxEQgExA9/+MM8tHD2enp6BACxfft2IYQ9+yiEEGVlZeInP/mJ7foXiUTEkiVLRHNzs7jgggvErbfeKoSwx5/jXXfdJc4888wJn7ND/+644w7xoQ99aNLn7dDH4916663ilFNOEZZlZb1/HNkpcK+++iqWL1+O+vr69LWPf/zj0HV9zHBmIUokEmhpacG6devGXF+3bh1eeeWVPLUqd1pbW9HV1TWmv5qm4YILLija/g4ODgIAysvLAdivj6Zp4tFHH0U0GsV5551nu/7dfPPNuOSSS/Cxj31szHW79PPAgQOor69HU1MTPvOZz+DQoUMA7NG/p59+GqtWrcLVV1+N6upqnHXWWfjxj3+cft4OfRwtkUjgZz/7Gb785S9DkqSs949hp8B1dXWhpqZmzLWysjI4nU50dXXlqVXT09fXB9M0x7W/pqam4NueiZE+2aW/Qgjcdttt+NCHPoTly5cDsE8f33zzTZSUlEDTNNx444148sknsWzZMtv0DwAeffRR/OlPf8LmzZvHPWeHfp5zzjl45JFH8Nxzz+HHP/4xurq6sHr1avT399uif4cOHcJDDz2EJUuW4LnnnsONN96Ir3/963jkkUcA2OPPcLSnnnoKoVAI119/PYDs9++kOvV8rmzatAl33333CV+zY8cOrFq1alqfJ0nSuGtCiAmvF6Lj21lMbc+EXfq7YcMG7NmzBy+99NK454q9j6eddhp2796NUCiExx9/HNdddx22b9+efr7Y+9fW1oZbb70V27Ztg8vlmvR1xdzP9evXp399xhln4LzzzsMpp5yChx9+GOeeey6A4u6fZVlYtWoV7rnnHgDAWWedhbfeegsPPfQQvvjFL6ZfV8x9HO2nP/0p1q9fP2YWA8he/ziykwMbNmzAvn37TvgY+ZfyVGpra8el2GAwiGQyOS7xFprKykooijKu/T09PQXf9kyMrASxQ39vueUWPP300/jjH/+I+fPnp6/bpY9OpxOLFy/GqlWrsHnzZpx55pn4/ve/b5v+tbS0oKenBytXroSqqlBVFdu3b8f/+l//C6qqpvtS7P0czev14owzzsCBAwds8edYV1eHZcuWjbl2+umnpxd32KGPI44cOYLf/e53+MpXvpK+lu3+MezkQGVlJZYuXXrCx4n+tTXaeeedh7/85S/o7OxMX9u2bRs0TcPKlStz1YWscDqdWLlyZbrCfkRzczNWr16dp1blTlNTE2pra8f0N5FIYPv27UXTXyEENmzYgCeeeAJ/+MMf0NTUNOZ5O/RxIkII6Lpum/5ddNFFePPNN7F79+70Y9WqVbj22muxe/duLFq0yBb9HE3Xdezbtw91dXW2+HNcs2bNuG0f3nnnHTQ2NgKw1/8Wt2zZgurqalxyySXpa1nv32yrp2l2jhw5Inbt2iXuvvtuUVJSInbt2iV27dolIpGIEEIIwzDE8uXLxUUXXST+9Kc/id/97ndi/vz5YsOGDXlu+fQ8+uijwuFwiJ/+9Kdi7969YuPGjcLr9YrDhw/nu2kZiUQi6T8jAOL+++8Xu3btEkeOHBFCCHHvvfeKQCAgnnjiCfHmm2+Kz372s6Kurk6Ew+E8t3x6vva1r4lAICCef/550dnZmX4MDw+nX1PsfbzzzjvFCy+8IFpbW8WePXvEN7/5TSHLsti2bZsQovj7N5nRq7GEKP5+fuMb3xDPP/+8OHTokHjttdfEpZdeKnw+X/r/W4q9f2+88YZQVVV85zvfEQcOHBA///nPhcfjET/72c/Sryn2PgohhGmaYsGCBeKOO+4Y91w2+8ewk2fXXXedADDu8cc//jH9miNHjohLLrlEuN1uUV5eLjZs2CDi8Xj+Gj1D//Ef/yEaGxuF0+kUH/jAB9LLmIvRH//4xwn/vK677johRGo56F133SVqa2uFpmniwx/+sHjzzTfz2+gZmKhvAMSWLVvSryn2Pn75y19O/32sqqoSF110UTroCFH8/ZvM8WGn2Pt5zTXXiLq6OuFwOER9fb341Kc+Jd56663088XePyGE+H//7/+J5cuXC03TxNKlS8WPfvSjMc/boY/PPfecACD2798/7rls9k8SQoiZjwcRERERFQfW7BAREZGtMewQERGRrTHsEBERka0x7BAREZGtMewQERGRrTHsEBERka0x7BAREZGtMewQERGRrTHsEBERka0x7BAREZGtMewQERGRrTHsEFHReuSRR1BRUQFd18dcv/LKK/HFL34xT60iokLDsENERevqq6+GaZp4+umn09f6+vqwdetWfOlLX8pjy4iokDDsEFHRcrvd+NznPoctW7akr/385z/H/PnzceGFF+avYURUUBh2iKio3XDDDdi2bRva29sBAFu2bMH1118PSZLy3DIiKhSSEELkuxFERLOxcuVKXHXVVfj4xz+Os88+G4cPH0ZDQ0O+m0VEBULNdwOIiGbrK1/5Ch544AG0t7fjYx/7GIMOEY3BkR0iKnrhcBh1dXUwDAOPPPIIrrnmmnw3iYgKCGt2iKjo+f1+XHnllSgpKcEVV1yR7+YQUYFh2CEiW+js7MS1114LTdPy3RQiKjCcxiKiojYwMIBt27bh2muvxd69e3Haaaflu0lEVGBYoExERe0DH/gAgsEg7rvvPgYdIpoQR3aIiIjI1lizQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES2xrBDREREtsawQ0RERLbGsENERES29v8B38Co/dD/qSQAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot correlation between predicted and observed values\n",
    "df_cmp = pd.DataFrame({\n",
    "    'y': df_de.loc[(\"NK cells\", \"Prednisolone\")].values.ravel(), \n",
    "    'y_pred': scm.predict([(\"NK cells\", \"Prednisolone\")]).values.ravel()\n",
    "})\n",
    "\n",
    "# Make the plot square with same X and Y limits\n",
    "df_cmp.plot.scatter(x='y', y='y_pred', alpha=0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save model\n",
    "scm.save(\"_models/config.pkl\", \"_models/weights.keras\", \"_models/result.pkl\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 2s 2s/step\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></th>\n",
       "      <th></th>\n",
       "      <th>A1BG</th>\n",
       "      <th>A1BG-AS1</th>\n",
       "      <th>A2M</th>\n",
       "      <th>A2M-AS1</th>\n",
       "      <th>A2MP1</th>\n",
       "      <th>A4GALT</th>\n",
       "      <th>AAAS</th>\n",
       "      <th>AACS</th>\n",
       "      <th>AAGAB</th>\n",
       "      <th>AAK1</th>\n",
       "      <th>...</th>\n",
       "      <th>ZUP1</th>\n",
       "      <th>ZW10</th>\n",
       "      <th>ZWILCH</th>\n",
       "      <th>ZWINT</th>\n",
       "      <th>ZXDA</th>\n",
       "      <th>ZXDB</th>\n",
       "      <th>ZXDC</th>\n",
       "      <th>ZYG11B</th>\n",
       "      <th>ZYX</th>\n",
       "      <th>ZZEF1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cell_type</th>\n",
       "      <th>sm_name</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">NK cells</th>\n",
       "      <th>Bosutinib</th>\n",
       "      <td>0.330316</td>\n",
       "      <td>0.138537</td>\n",
       "      <td>-0.154608</td>\n",
       "      <td>-0.042543</td>\n",
       "      <td>0.235437</td>\n",
       "      <td>0.781162</td>\n",
       "      <td>0.114396</td>\n",
       "      <td>0.222130</td>\n",
       "      <td>-0.071732</td>\n",
       "      <td>0.424109</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.011349</td>\n",
       "      <td>-0.301028</td>\n",
       "      <td>0.034625</td>\n",
       "      <td>-0.008219</td>\n",
       "      <td>0.021145</td>\n",
       "      <td>0.125907</td>\n",
       "      <td>0.192161</td>\n",
       "      <td>0.209567</td>\n",
       "      <td>0.025681</td>\n",
       "      <td>0.078541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Belinostat</th>\n",
       "      <td>1.666431</td>\n",
       "      <td>3.450031</td>\n",
       "      <td>-8.108703</td>\n",
       "      <td>-16.189913</td>\n",
       "      <td>-3.965121</td>\n",
       "      <td>6.956760</td>\n",
       "      <td>0.939568</td>\n",
       "      <td>0.900427</td>\n",
       "      <td>4.964502</td>\n",
       "      <td>4.974038</td>\n",
       "      <td>...</td>\n",
       "      <td>5.483314</td>\n",
       "      <td>1.852388</td>\n",
       "      <td>-2.629870</td>\n",
       "      <td>0.311158</td>\n",
       "      <td>-1.745673</td>\n",
       "      <td>0.308468</td>\n",
       "      <td>-0.960009</td>\n",
       "      <td>4.471121</td>\n",
       "      <td>-0.727347</td>\n",
       "      <td>-1.623647</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 18211 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                          A1BG  A1BG-AS1       A2M    A2M-AS1     A2MP1  \\\n",
       "cell_type sm_name                                                         \n",
       "NK cells  Bosutinib   0.330316  0.138537 -0.154608  -0.042543  0.235437   \n",
       "          Belinostat  1.666431  3.450031 -8.108703 -16.189913 -3.965121   \n",
       "\n",
       "                        A4GALT      AAAS      AACS     AAGAB      AAK1  ...  \\\n",
       "cell_type sm_name                                                       ...   \n",
       "NK cells  Bosutinib   0.781162  0.114396  0.222130 -0.071732  0.424109  ...   \n",
       "          Belinostat  6.956760  0.939568  0.900427  4.964502  4.974038  ...   \n",
       "\n",
       "                          ZUP1      ZW10    ZWILCH     ZWINT      ZXDA  \\\n",
       "cell_type sm_name                                                        \n",
       "NK cells  Bosutinib  -0.011349 -0.301028  0.034625 -0.008219  0.021145   \n",
       "          Belinostat  5.483314  1.852388 -2.629870  0.311158 -1.745673   \n",
       "\n",
       "                          ZXDB      ZXDC    ZYG11B       ZYX     ZZEF1  \n",
       "cell_type sm_name                                                       \n",
       "NK cells  Bosutinib   0.125907  0.192161  0.209567  0.025681  0.078541  \n",
       "          Belinostat  0.308468 -0.960009  4.471121 -0.727347 -1.623647  \n",
       "\n",
       "[2 rows x 18211 columns]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scm = scape.SCAPE.load(\"_models/config.pkl\", \"_models/weights.keras\", \"_models/result.pkl\")\n",
    "scm.predict([(\"NK cells\", \"Bosutinib\"), (\"NK cells\", \"Belinostat\")])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Save during training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Baseline zero MRRMSE: 1.4273\n",
      "Baseline slogpval_drug MRRMSE: 1.1278\n",
      "Model size (MB): 36.76\n",
      "Model inputs: ['in_slogpval_drug', 'in_lfc_drug', 'in_slogpval_cell_encoder', 'in_lfc_cell_encoder', 'in_slogpval_cell_decoder', 'in_lfc_cell_decoder']\n",
      "Model will be saved to _models\\example_model.keras\n",
      "Epoch    1: val_loss None -> 1.4205, train loss: 1.3073 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0048, 0.47%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.7939, -25.96%\n",
      "Epoch    2: val_loss 1.4205 -> 1.3863, train loss: 1.2798 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0295, 2.87%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8135, -22.92%\n",
      "Epoch    3: val_loss 1.3863 -> 1.3620, train loss: 1.2507 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0479, 4.57%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8280, -20.77%\n",
      "Epoch    6: val_loss 1.3620 -> 1.3297, train loss: 1.2050 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0733, 6.83%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8481, -17.91%\n",
      "Epoch    8: val_loss 1.3297 -> 1.3186, train loss: 1.1916 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0824, 7.61%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8553, -16.92%\n",
      "Epoch   11: val_loss 1.3186 -> 1.3063, train loss: 1.1504 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.0926, 8.48%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8633, -15.83%\n",
      "Epoch   17: val_loss 1.3063 -> 1.2954, train loss: 1.1098 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1018, 9.24%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8706, -14.87%\n",
      "Epoch   26: val_loss 1.2954 -> 1.2917, train loss: 1.0342 (lr=0.00500)\n",
      "\t > Improvement factor/percentage zero                 (err=1.4273): 1.1050, 9.50%\n",
      "\t > Improvement factor/percentage slogpval_drug        (err=1.1278): 0.8731, -14.53%\n"
     ]
    }
   ],
   "source": [
    "scm = scape.SCAPE(model_setup)\n",
    "\n",
    "result = scm.train(\n",
    "    val_cells=[\"NK cells\"], \n",
    "    val_drugs=[\"Prednisolone\"],\n",
    "    input_columns=top_genes,\n",
    "    epochs=30,\n",
    "    baselines=[\"zero\", \"slogpval_drug\"],\n",
    "    output_folder=\"_models\",\n",
    "    config_file_name=\"example_config.pkl\",\n",
    "    model_file_name=f\"example_model.keras\",\n",
    "    result_file_name=\"example_result.pkl\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 1s 633ms/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "1.2916558015647297"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check that the error is the same\n",
    "scm = scape.model.SCAPE.load( \"_models/example_config.pkl\", \"_models/example_model.keras\", \"_models/example_result.pkl\")\n",
    "\n",
    "y_true = df_de.loc[(df_de.index.get_level_values(\"cell_type\") == \"NK cells\") & (df_de.index.get_level_values(\"sm_name\") == \"Prednisolone\")]\n",
    "y_pred = scm.predict([(\"NK cells\", \"Prednisolone\")])\n",
    "\n",
    "scape.losses.np_mrrmse(y_true, y_pred)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.11.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}