replogle_pertspectra_metrics.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "b8270f66-7a27-4263-aa97-25abeb2cf34d",
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "529f64f3-c83e-481d-872c-df89b9dd063b",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "\n",
    "from utils import (\n",
    "    auprc,\n",
    "    factor_enrichment_gsea,\n",
    "    get_gprofiler,\n",
    "    perturbation_signal_recovery,\n",
    "    read_aws_csv,\n",
    "    read_aws_h5ad,\n",
    "    read_aws_pickle,\n",
    "    retrieve_stringdb_neighbors,\n",
    "    split_data_by_cell,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "c57d97ad-da1a-4853-b825-a45aae5ca3dc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# read in trained model outputs generated from ./PertSpectra_load_checkpoints/pertspectra_replogle.ipynb\n",
    "adata = read_aws_h5ad(\n",
    "    \"s3://pert-spectra/PertSpectra_checkpoints/pertspectra_replogle/replogle.h5ad\"\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5b7bf773-5e0c-43b8-8df2-4ce3376be959",
   "metadata": {},
   "source": [
    "# Reconstruction"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3b4b092b-2120-48c6-a492-84656e8b7169",
   "metadata": {},
   "source": [
    "## Spearman Coefficient\n",
    "- Spearman correlation between predicted and observed expression (on DE genes if available)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "e2885bd3-5433-47d6-88d5-47be83df5b5a",
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.stats import spearmanr\n",
    "\n",
    "test_corr_singles = []\n",
    "test_corr_combos = []\n",
    "train_idx, val_idx, test_idx = split_data_by_cell(\n",
    "    adata.X, adata.obs[\"gene\"], test_size=0.2, val_size=0.2\n",
    ")\n",
    "adata_test = adata[test_idx]\n",
    "\n",
    "for pert in adata_test.obs[\"gene\"].unique():\n",
    "    hold_idx = [i for i, x in enumerate(adata_test.obs[\"gene\"]) if x == pert]\n",
    "    recon = adata.uns[\"recon\"][hold_idx]\n",
    "    # correlation\n",
    "    mean_reconstruction = recon.mean(axis=0)\n",
    "    mean_observed = np.squeeze(np.array(adata_test[hold_idx].X.mean(axis=0)))\n",
    "    if (\"+\" in pert) and (\"ctrl\" not in pert):\n",
    "        test_corr_combos.append(\n",
    "            [pert, spearmanr(mean_reconstruction, mean_observed)[0]]\n",
    "        )\n",
    "    else:\n",
    "        test_corr_singles.append(\n",
    "            [pert, spearmanr(mean_reconstruction, mean_observed)[0]]\n",
    "        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "c75473ab-68bf-4be3-b2bf-a5bc3fe081d8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot train correlation distribution vs test correlation distribution\n",
    "test_corr = np.array(test_corr_singles + test_corr_combos)[:, 1].astype(float)\n",
    "sns.histplot(test_corr, label=\"test_corr\")\n",
    "plt.title(\"Correlation between predicted and observed expression per perturbation\")\n",
    "plt.xlabel(\"Spearman Correlation\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "a03e387a-9b61-4b4e-90c2-e1995d46c9e6",
   "metadata": {},
   "outputs": [],
   "source": [
    "np.savetxt(\n",
    "    \"figures/reconstruction_spearmans/model_replogle_spearman_correlations.csv\",\n",
    "    test_corr,\n",
    "    delimiter=\",\",\n",
    "    fmt=\"%.2f\",\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "64fc0b35-f87e-4d8a-87f1-e6080a3ee83f",
   "metadata": {},
   "source": [
    "# Perturbation Embedding Heatmap"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c36ebe2b-2835-44ed-9f71-a52b522125b1",
   "metadata": {},
   "source": [
    "## Hierarchical Clustering + Enrichment\n",
    "- Perform enrichment tests (gprofiler) on hierarchical clustering of perturbation embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "99d3263b-8988-47a6-a57b-0dce4186eb6b",
   "metadata": {},
   "outputs": [],
   "source": [
    "import seaborn as sns\n",
    "from scipy.spatial.distance import pdist, squareform\n",
    "\n",
    "# Compute the pairwise distances\n",
    "pert_embeddings_df = pd.DataFrame(\n",
    "    adata.uns[\"SPECTRA_pert_scores\"], index=adata.uns[\"Spectra_pert_labels\"]\n",
    ")\n",
    "df = pert_embeddings_df.drop(index=[\"basal\", \"ctrl\"])\n",
    "\n",
    "distance = \"euclidean\"\n",
    "distances = pdist(df.values, metric=distance)\n",
    "\n",
    "# Convert the distances into a square distance matrix\n",
    "distance_matrix = pd.DataFrame(squareform(distances), index=df.index, columns=df.index)\n",
    "clustermap = sns.clustermap(distance_matrix, cmap=\"viridis_r\")\n",
    "clustermap.fig.suptitle(f\"Pairwise {distance} distance of perturbation latent vectors\")\n",
    "# clustermap.ax_row_dendrogram.set_visible(False)\n",
    "clustermap.ax_col_dendrogram.set_visible(False)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "7a684ec0-9241-42a4-b65b-b3bdf51b7c7f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiYAAAG/CAYAAABsTxxQAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8fJSN1AAAACXBIWXMAAA9hAAAPYQGoP6dpAADHEklEQVR4nOzdd3xT5f4H8E/2TpomaZvuXToohZZRNsoQEUTxggORpSIOcN2fqNd9Va7jol5xIaJeRcSFAxRQkL1b9iirLaWD7pG0SZPn9wf3PLbQQls60vJ9v159QZKTnCfJyTnf8zzf53tEjDEGQgghhBAPIO7oBhBCCCGECCgwIYQQQojHoMCEEEIIIR6DAhNCCCGEeAwKTAghhBDiMSgwIYQQQojHoMCEEEIIIR5D2tENaAq3242zZ89Cp9NBJBJ1dHMIIYQQ0gSMMVRUVMDf3x9icdP6QjpFYHL27FkEBQV1dDMIIYQQ0gLZ2dkIDAxs0rKdIjDR6XQAzr8xvV7fwa0hhBBCSFOUl5cjKCiIH8ebolMEJsLwjV6vp8CEEEII6WSak4ZBya+EEEII8RgUmBBCCCHEY1BgQgghhBCPQYEJIYQQQjwGBSaEEEII8RgUmBBCCCHEY1BgQgghhBCPQYEJIYQQQjwGBSaEEEII8RgUmBBCCCHEY1BgQgghhBCPQYEJIYQQQjwGBSaEEEII8Rid4urChHgqxhjsTldHN4MQj6KSSZp1NVlC6qLAhJAWYozhlve3YndmSUc3hRCPkhJixPJZqRSckBahoRxCWsjudFFQQkgDdmWWUE8iaTHqMSGkFex6ejjUcklHN4OQDmVzuJDy0tqObgbp5CgwIaQVqOUSqOX0cyKEkCvV7KGcDRs2YOzYsfD394dIJMIPP/xwyeW/++47jBgxAhaLBXq9Hqmpqfjtt99a2l5CCCGEdGHNDkyqqqrQo0cP/Oc//2nS8hs2bMCIESOwcuVK7N69G8OGDcPYsWORlpbW7MYSQgghpGtrdt/z6NGjMXr06CYvv2DBgnq3X375ZaxYsQI//fQTevbs2eBzampqUFNTw2+Xl5c3t5mEEEII6YTafVaO2+1GRUUFvL29G13mlVdegcFg4H9BQUHt2EJCCCGEdJR2D0zeeOMNVFVVYeLEiY0uM2/ePJSVlfG/7OzsdmwhIYQQQjpKu04jWLp0KZ577jmsWLECPj4+jS6nUCigUCjasWWEEEII8QTtFpgsW7YMM2bMwPLlyzF8+PD2Wi0hhBBCOpF2GcpZunQppk6dii+//BJjxoxpj1USQgghpBNqdo9JZWUljh8/zm+fOnUK6enp8Pb2RnBwMObNm4ecnBx89tlnAM4HJVOmTMFbb72Ffv36IS8vDwCgUqlgMBha6W0QQgghpCtodo/Jrl270LNnTz7V95FHHkHPnj3xzDPPAAByc3ORlZXFl//ggw9QW1uL+++/H1arlf/NmTOnld4CIYQQQrqKZveYDB06FIyxRh9fsmRJvdvr169v7ioIIYQQcpWiqwsTQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj9HswGTDhg0YO3Ys/P39IRKJ8MMPP1z2OX/++SeSk5OhVCoRHh6O999/vyVtJYQQQkgX1+zApKqqCj169MB//vOfJi1/6tQpXH/99Rg0aBDS0tLw5JNP4qGHHsK3337b7MYSQgghpGuTNvcJo0ePxujRo5u8/Pvvv4/g4GAsWLAAABAbG4tdu3bh9ddfx4QJE5q7ekIIIYR0YW2eY7J161aMHDmy3n2jRo3Crl274HQ6G3xOTU0NysvL6/0RQgghpOtr88AkLy8Pvr6+9e7z9fVFbW0tCgsLG3zOK6+8AoPBwP+CgoLaupmEEEII8QDtMitHJBLVu80Ya/B+wbx581BWVsb/srOz27yNhBBCCOl4zc4xaS4/Pz/k5eXVu6+goABSqRQmk6nB5ygUCigUirZuGiGEEEI8TJv3mKSmpmLNmjX17lu9ejVSUlIgk8naevWEEEII6USaHZhUVlYiPT0d6enpAM5PB05PT0dWVhaA88MwU6ZM4cvPmjULmZmZeOSRR3D48GEsXrwYH3/8MR577LHWeQeEEEII6TKaPZSza9cuDBs2jN9+5JFHAAB33XUXlixZgtzcXB6kAEBYWBhWrlyJhx9+GO+++y78/f3x9ttv01RhQgghhFyk2YHJ0KFDefJqQ5YsWXLRfUOGDMGePXuauypCCCGEXGXoWjmEEEII8RgUmBBCCCHEY1BgQgghhBCPQYEJIYQQQjxGmxdYI4QQ0jEYY7A7Xe22PpujtsH/tzWVTNJoJXHS+VBgQgghXRBjDLe8vxW7M0s6ZP0pL/3efusKMWL5rFQKTroIGsohhJAuyO50dVhQ0t52ZZa0a88QaVvUY0IIIV3crqeHQy2XdHQzWp3N4ULKS2s7uhmklVFgQgghXZxaLoFaTrt70jnQUA4hhBBCPAYFJoQQQgjxGBSYEEIIIcRjUGBCCCGEEI9BgQkhhBBCPAYFJoQQQgjxGBSYEEIIIcRjUGBCCCGEEI9BgQkhhBBCPAYFJoQQQgjxGBSYEEIIIcRjUGBCCCGEEI9BgQkhhBBCPAYFJoQQQgjxGBSYEEIIIcRjUGBCCCGEEI9BgQkhhBBCPAYFJoQQQgjxGBSYEEIIIcRjUGBCCCGEEI9BgQkhhBBCPAYFJoQQQgjxGBSYEEIIIcRjUGBCCCGEEI9BgQkhhBBCPAYFJoQQQgjxGBSYEEIIIcRjUGBCCCGEEI9BgQkhhBBCPAYFJoQQQgjxGBSYEEIIIcRjUGBCCCGEEI9BgQkhhBBCPAYFJoQQQgjxGBSYEEIIIcRjUGBCCCGEEI9BgQkhhBBCPAYFJoQQQgjxGC0KTBYuXIiwsDAolUokJydj48aNl1z+iy++QI8ePaBWq2G1WjFt2jQUFRW1qMGEEEII6bqaHZgsW7YMc+fOxVNPPYW0tDQMGjQIo0ePRlZWVoPLb9q0CVOmTMGMGTNw8OBBLF++HDt37sTMmTOvuPGEEEII6VqaHZi8+eabmDFjBmbOnInY2FgsWLAAQUFBeO+99xpcftu2bQgNDcVDDz2EsLAwDBw4EPfeey927drV6DpqampQXl5e748QQgghXV+zAhOHw4Hdu3dj5MiR9e4fOXIktmzZ0uBz+vfvjzNnzmDlypVgjCE/Px/ffPMNxowZ0+h6XnnlFRgMBv4XFBTUnGYSQgghpJNqVmBSWFgIl8sFX1/fevf7+voiLy+vwef0798fX3zxBSZNmgS5XA4/Pz94eXnhnXfeaXQ98+bNQ1lZGf/Lzs5uTjMJIYQQ0km1KPlVJBLVu80Yu+g+waFDh/DQQw/hmWeewe7du/Hrr7/i1KlTmDVrVqOvr1AooNfr6/0RQgghpOuTNmdhs9kMiURyUe9IQUHBRb0ogldeeQUDBgzA448/DgBITEyERqPBoEGD8NJLL8Fqtbaw6YQQQgjpaprVYyKXy5GcnIw1a9bUu3/NmjXo379/g8+x2WwQi+uvRiKRADjf00IIIYQQImj2UM4jjzyCRYsWYfHixTh8+DAefvhhZGVl8aGZefPmYcqUKXz5sWPH4rvvvsN7772HkydPYvPmzXjooYfQp08f+Pv7t947IYQQQkin16yhHACYNGkSioqK8MILLyA3NxcJCQlYuXIlQkJCAAC5ubn1appMnToVFRUV+M9//oNHH30UXl5euOaaazB//vzWexeEEEII6RKaHZgAwOzZszF79uwGH1uyZMlF9z344IN48MEHW7IqQgghhFxF6Fo5hBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGNQYEIIIYQQjyHt6AYQQgjxLIwx2J2ujm7GZdkctQ3+35OpZBKIRKKOboZHo8CEEEIIxxjDLe9vxe7Mko5uSrOkvPR7RzehSVJCjFg+K5WCk0ugoRxCCCGc3enqdEFJZ7Irs6RT9EZ1JOoxIYQQ0qBdTw+HWi7p6GZ0CTaHCykvre3oZnQKFJgQQghpkFougVpOhwnSvmgohxBCCCEegwITQgghhHgMCkwIIYQQ4jEoMCGEEEKIx6DAhBBCCCEegwITQgghhHgMCkwIIYQQ4jEoMCGEEEKIx6DAhBBCCCEegwITQgghhHgMCkwIIYQQ4jEoMCGEEEKIx6DAhBBCCCEegwITQgghhHgMCkwIIYQQ4jEoMCGEEEKIx6DAhBBCCCEegwITQgghhHgMCkwIIYQQ4jEoMCGEEEKIx6DAhBBCCCEegwITQgghhHgMCkwIIYQQ4jEoMCGEEEKIx6DAhBBCCCEegwITQgghhHiMFgUmCxcuRFhYGJRKJZKTk7Fx48ZLLl9TU4OnnnoKISEhUCgUiIiIwOLFi1vUYEIIIYR0XdLmPmHZsmWYO3cuFi5ciAEDBuCDDz7A6NGjcejQIQQHBzf4nIkTJyI/Px8ff/wxIiMjUVBQgNra2ituPCGEEEK6lmYHJm+++SZmzJiBmTNnAgAWLFiA3377De+99x5eeeWVi5b/9ddf8eeff+LkyZPw9vYGAISGhl5ZqwkhhBDSJTVrKMfhcGD37t0YOXJkvftHjhyJLVu2NPicH3/8ESkpKfjXv/6FgIAAREdH47HHHoPdbm90PTU1NSgvL6/3RwghhJCur1k9JoWFhXC5XPD19a13v6+vL/Ly8hp8zsmTJ7Fp0yYolUp8//33KCwsxOzZs1FcXNxonskrr7yC559/vjlNI4QQQkgX0KLkV5FIVO82Y+yi+wRutxsikQhffPEF+vTpg+uvvx5vvvkmlixZ0mivybx581BWVsb/srOzW9JMQgghhHQyzeoxMZvNkEgkF/WOFBQUXNSLIrBarQgICIDBYOD3xcbGgjGGM2fOICoq6qLnKBQKKBSK5jSNEEIIIV1As3pM5HI5kpOTsWbNmnr3r1mzBv3792/wOQMGDMDZs2dRWVnJ7zt27BjEYjECAwNb0GRCCCGEdFXNHsp55JFHsGjRIixevBiHDx/Gww8/jKysLMyaNQvA+WGYKVOm8OVvv/12mEwmTJs2DYcOHcKGDRvw+OOPY/r06VCpVK33TgghhBDS6TV7uvCkSZNQVFSEF154Abm5uUhISMDKlSsREhICAMjNzUVWVhZfXqvVYs2aNXjwwQeRkpICk8mEiRMn4qWXXmq9d0EIIYSQLqHZgQkAzJ49G7Nnz27wsSVLllx0X7du3S4a/iGEEEIIuRBdK4cQQgghHoMCE0IIIYR4DApMCCGEEOIxKDAhhBBCiMegwIQQQgghHoMCE0IIIYR4DApMCCGEEOIxKDAhhBBCiMegwIQQQgghHoMCE0IIIYR4DApMCCGEEOIxKDAhhBBCiMegwIQQQgghHoMCE0IIIYR4DApMCCGEEOIxKDAhhBBCiMegwIQQQgghHoMCE0IIIYR4DGlHN4AQQghpT4wx2J2udl2nzVHb4P/bi0omgUgkavf1tgQFJoQQQq4ajDHc8v5W7M4s6bA2pLz0e/uvM8SI5bNSO0VwQkM5hBBCrhp2p6tDg5KOsiuzpN17iVqKekwIIYRclXY9PRxquaSjm9GmbA4XUl5a29HNaBYKTAghhFyV1HIJ1HI6DHoaGsohhBBCiMegwIQQQgghHoMCE0IIIYR4DApMCCGEEOIxKDAhhBBCiMegwIQQQgghHoMCE0IIIYR4DApMCCGEEOIxKDAhhBBCiMegwIQQQgghHoMCE0IIIYR4DApMCCGEEOIxKDAhhBBCiMegwIQQQgghHoMCE0IIIYR4DApMCCGEEOIxKDAhhBBCiMegwIQQQgghHoMCE0IIIYR4DGlHN4B4BsYY7E5XRzejU7E5ahv8P2k6lUwCkUjU0c0ghHgQCkwIGGO45f2t2J1Z0tFN6bRSXvq9o5vQKaWEGLF8VioFJ4QQjoZyCOxOFwUlpEPsyiyhnjpCSD3UY0Lq2fX0cKjlko5uBunibA4XUl5a29HNIIR4IApMSD1quQRqOW0WhBBCOkaLhnIWLlyIsLAwKJVKJCcnY+PGjU163ubNmyGVSpGUlNSS1RJCCCGki2t2YLJs2TLMnTsXTz31FNLS0jBo0CCMHj0aWVlZl3xeWVkZpkyZgmuvvbbFjSWEEEJI19bswOTNN9/EjBkzMHPmTMTGxmLBggUICgrCe++9d8nn3Xvvvbj99tuRmpra4sYSQgghpGtrVmDicDiwe/dujBw5st79I0eOxJYtWxp93ieffIITJ07g2WefbdJ6ampqUF5eXu+PEEIIIV1fswKTwsJCuFwu+Pr61rvf19cXeXl5DT4nIyMDTzzxBL744gtIpU1LqnzllVdgMBj4X1BQUHOaSQghhJBOqkXJrxcWQ2KMNVggyeVy4fbbb8fzzz+P6OjoJr/+vHnzUFZWxv+ys7Nb0kxCCCGEdDLNmhdqNpshkUgu6h0pKCi4qBcFACoqKrBr1y6kpaXhgQceAAC43W4wxiCVSrF69Wpcc801Fz1PoVBAoVA0p2mEEEII6QKa1WMil8uRnJyMNWvW1Lt/zZo16N+//0XL6/V67N+/H+np6fxv1qxZiImJQXp6Ovr27XtlrSeEEEJIl9LsSlqPPPII7rzzTqSkpCA1NRUffvghsrKyMGvWLADnh2FycnLw2WefQSwWIyEhod7zfXx8oFQqL7qfEEIIIaTZgcmkSZNQVFSEF154Abm5uUhISMDKlSsREhICAMjNzb1sTRNCCCGEkIa0qPb47NmzMXv27AYfW7JkySWf+9xzz+G5555ryWoJIYQQ0sXR1YUJIYQQ4jEoMCGEEEKIx6DAhBBCCCEegwITQgghhHgMCkwIIYQQ4jFaNCuHEEIIIe2PMQa709Xk5W2O2gb/3xQqmaTBy820NQpMCCGEkE6AMYZb3t+K3ZklLXp+yku/N2/5ECOWz0pt9+CEhnIIIYSQTsDudLU4KGmJXZklzeqdaS3UY0IIIYR0MrueHg61XNImr21zuJDy0to2ee2moMCEEEII6WTUcgnU8q55CKehHEIIIYR4DApMCCGEEOIxKDAhhBBCiMfomgNUhBBCurzm1vQArqyuB9BxtT2uJhSYEEII6XSutKYH0Py6HkDH1fZoKw0Fd00J3toyQKPAhBBCSKfT3jU9BEJtj64wI6YpwV1jwVtbBmid/5MlhBBySYyxjm5Cm2rLmh6Cjq7t0RauJLhrywCNAhNCCOmC6gYjkxftwHez+3eZ4YcLdeWaHu2lqcFdewRo9E0SQkgXVDcIScsu7TLDD6RteFJwR9OFCSGEEOIxKDAhhBBCiMegwIQQQgghHoMCE0IIIYR4DM/IdCGEeJSWVNRsjiutvtlUVKWTkM6HAhNCSD2tUVGzOVpSfbPJr93FqnQScjWgwIQQUk9HVdRsC7syS1BUVeMx0yAB6sUhncuFvaeX6u1srW3bc36thBCP05SiS4wx3LFoB9KzS9unUc3Ulj0yLUG9OKSzuFzv6YW/rdbatikwIYQ0qilFl2yOWo8NSjxRR19r5XL5Q83J/6Hen86tbnXghi5b0Nze012ZJbA5aqFRyK6oXRSYEEJaTXtcs6Sz8oRrrTQ3f+hyvU3U+9N5CT2dgstdtqCx3zZjDLd9tB37zpQ16XWaggITQkir8aSy1p7irx6Kv85I6/ZEtGevQ2vnD3V0709X0569WXanq15PZ1p26SV7Oxr7bdsctTwoacrrNAVtTYQQ0kYa66Go2xPRUb0OLendEg6cdocLg/61HsBfB0ga1rkyntCb1Vrf36QPtmHx1GSIxWLIWnBlawpMCCGkjTSlh6Kjeh2a27t1uSCLhnWujCf0ZjHGGu2VaU4v34Gz5ejz8joAQA8feXOaDYACE0IIaRcX9lB4Qs5Jc1zuwEnDOq3nSnK1rmS7mrxoB9IaSWSv20PTI9CAL2b2gUgkumwhxsZe71JoCyKEkHbQlfJv6h44O1uAdSmXyvFoan5HawxpddS20tQgYu+ZMiQ8t6bN2tE1fiWEENJJMMZQW1MDp+OvA6CzugZOdy2kCkWnGArpSkGWoDk5HpfK72jvIS3GGJjdzm+762xXbpsd7tq/el5EKlWT25Xor8d/bknC4Lc3tF5jm6hrbVmEEOLBGGP46pm/4+yxw3CKpEDo3QCA9+65AzJWC/+YONz6/PxOEZx0Na2V49HSIa26dURsTjtUMu1ltwO3243M225H9d69/L5qiRwY+zIAIGPAQChdDv6YskcPBC/+GGK1+rLt2Xe2HF+9sB3watbbaBUUmBBCSDupranB2WOHAQAyVosHT71X7/GzRw+htqYGMqWyI5pH/qclOR5XMqTFGMPdq+8G8DcAwNCvh6CXXwI+ve7TRoMTxhgyb70N1fv21btf6XJg1Q+PNfic6r17cSw5Bcru3eH91juNtmd2mRIaJoIDzZ9R0xooMCGEXNKFXcUXulTXcV3N6Ua+Gtz34X8hU5wPQJw11Xjvnskd3CIiaO+hKnutHfuLd0MXu5vfl1aQhuLqYqikKqikF/92mN1+UVAij41FwFsLIBaLIVapIFIqIRKJ4LbbkTFgIF+uev9+nBwxkvesXEjWMfEIR4EJIaRRjDFk3n4H7GlpjS5zqa7julS9eiHki/9ScPI/MoWSekaagTEGt/uvANnlctX5vx0u1/mAWCzufAFwQ+XgAWDo10MBAImWRHw4/EP+vlRSVYPLOw4fxqmRo/htZffuCPpkces2th1QYEIIaRSzV18yKAEu3XVcl33PHjC7HaImjG9fLXgibE01v89ZU+3xSbAOhxNSuP/3f1eD9wtkMtkVvxfGGHbvmYiysj38vppaOYDXAQAbN/aBQno+IDYYkpHca1mT18kYg83thq1OoFPlckHFmjm7pm5w0cyiYtWu6ks+vu/cPvRb2o/fTjAl4O3Uf13+dffvR0ZK72a1xRNQYEIIaZKozZsgVjV8plYXYwys+q8drdtux4nhI/j/L3S1DvHUTYSt6717JrdJEmzdITfhthvnX18kEzdrXa+99hpkovMBiJOJASRfdL8gKCgI06dPv6L34nbb6wUlAKCQOvDxyIcuWrasbDfcbjskkssHwIwxjNtzHDvLq4BaN4T+q+6bDqKPSYcVPSOb1m7GgM9uBHD/+dufjQdm/gy00XZ9oOgARn83Gp+3yat3PApMLocxwGnr6Fa0rbo7LIcNQBe/CJtM3WY7jK5MrFJdNpufMYasSwz91B3nFlytQzy1jpqLghJBWyTB5r60/aLbqv8FJvIQPSyzElv0HchEbkxV7mz08ezsbDidTsjlza8A2pBBA7c3GHS4XDZs3NS3Wa9lc7vPByUAIBWjelQAf2xHWRUKnbVQi0WwuRnUEjHU4kYCOKcN6pzNOK3cfP52zvn7INc0qz3kPApMLoUxYPEoIHv75ZftzJgCwCfn//9aJCCq6dDmtLmgfsD0Xyk4aQPMbr/s0M+FaIjnr0TYtk6CVUGETdBfdL8jsxzM6YaoGTNRHnvssUsGGw6HA6+//nqL2nkpEon6osCEMQaX668TSJfL1iq5Jt03H6x3u7dejR97RV11QXR7o8DkUpy2rh+UAFCLanBaeXtHN6P9ZG+js5l2cLmhnwtnClzN2isR1vp033rBB3O4LupJaSq5XA6ZTAan03nZZR2OhhOi2yr/ZOOmvtDre6Jn0vnptq2VELuz3IZMWzU0UgkgEv3Vi3LFr0zqosCkqR47Dsiv3jO6LsFhA16P7OhWXDWaMvRD/sIYuygJFkCrJsKK5BKI6wQm7kssezmMMSxevBjZ2dmXXbaxnpO2yj8BgPLyNPy5IRFA8xNiL6XfjqP1bvfUqfBpjC8sQP0ApbIQ0IKGjluAApOmkqvpDJsQ0iYaSoQVhnQ8tRqs0+lsUlByKdnZ2aiqqoJGo2nT99echNjmSquwI3HXafRJegcr0h/8Kzh5+3xQhMDewJ0/nD9+eNh36KkoMCGEkA7W3omwre1y+SZ1Mcbw6aef4uzZswDO96a0Rs+JYNDA7RCLVXC77XC57NiydSgA1MtBaYtaJzsMibCJldC4L5j6e2Yn8EoA4JcI/O3z88GJxnT+Macd0JhbtR1dQYsCk4ULF+K1115Dbm4u4uPjsWDBAgwaNKjBZb/77ju89957SE9PR01NDeLj4/Hcc89h1KhRDS5PSINaY3aUw9bw/1uKumhJG2ivRNgrVTdvRCaTNTkwcTgcPCgRXGrmTkOJrUDjwQVjDGnpU1Benl7v/rozdvT6nkju9XWT2ttq8vYB7/S4+P7A3sDfOl8RtLbU7MBk2bJlmDt3LhYuXIgBAwbggw8+wOjRo3Ho0CEEBwdftPyGDRswYsQIvPzyy/Dy8sInn3yCsWPHYvv27ejZs2ervAnSxbXF7KjWyDWh2T2kDXSWirBvv/02//9nn32GmTNnNrsXYvbs2Vi4cCGAhqufNpbYCvyVN3KhTZv7XXTfhcrL07Br902IS/q2We1tE2d2Am/1AEICO7olHkPc3Ce8+eabmDFjBmbOnInY2FgsWLAAQUFBeO+99xpcfsGCBfj73/+O3r17IyoqCi+//DKioqLw008/NbqOmpoalJeX1/sjVzFPnR0lzO4hl8QYg9tmu/iv7qXa7faLHm+sTDfxPDk5OU2anXMhISgBzgc3F37njSW2AufzRur2pDRGq03A4EF7MWhg/X1IRcUB7Em7imYjdiLN6jFxOBzYvXs3nnjiiXr3jxw5Elu2bGnSa7jdblRUVMDb27vRZV555RU8//zzzWkauVp4wuwomt3TZE251g5AhdcudOEMHU8N0ubMmYO33nqrVV4rJycHDocDCoWiwceF3JE9aXegomI/ACAtfQp6Jn12ydetrDyAtPQ70CNx0UWPVVQeAM319TzN6jEpLCyEy+WCr69vvft9fX2Rl5fXpNd44403UFVVhYkTJza6zLx581BWVsb/rjTzm3QhwuyoK/2TtVJw47ABjqqW/XnowaY1taTgmkAovHa1EWbo1M0v+frFp+Cw2684QGGMnS9FX+dPUPe+pq5HJpNdUXsAwGq18v831GsikEjUEIlEPCgBgPLy9Cb1mlRUHGjSEA/xDC1Kfr3o8suMNemsZunSpXjuueewYsUK+Pj4NLqcQqFoNGom5Iq1Zs7KlfScXGU5KlGbN0GkVNa7jk5DLndtnYZ0pevtNDRDJ//EMbwz9W9XPHW4cNEBOLMrGnwsr06hNXmwDuZZiRCLmz3a32y5ubn8/8KQkJAEWzdIaew4k753Zpu3kbSvZgUmZrMZEonkot6RgoKCi3pRLrRs2TLMmDEDy5cvx/Dhw5vfUtJ1XW7GTXNm0zRlpoyn5KxcZRVoRUolsmfe3awelKZWhu1Kwz6X6q240qnDjQUlF3JkVeDce/vgM7vHJT/TunklDoejVSq5Cq8DAGnpd/L7Gxu2qaw8cEXrI56nWYGJXC5HcnIy1qxZg5tuuonfv2bNGtx4442NPm/p0qWYPn06li5dijFjxrS8taTraW7vxeV6KJrbC9EROStXaY4Kq65u8bDO5XSl6+1ceGCf8fYiiEQiLHpwBoDWqQjr+/cUiDX1gwi3w1Wv18SZXXHZ6+fUzS9prXokwuvcddetKC/fy+8vL0+H01nc4tclnUezh3IeeeQR3HnnnUhJSUFqaio+/PBDZGVlYdasWQDO54fk5OTgs8/OR7ZLly7FlClT8NZbb6Ffv368t0WlUsFgMLTiWyGdUmv3XjS3F4Iq+naIutfRYYxddninMS0Z9rmQpw8D/bxgPvJPZvDbQu6JNaobbnnqRd725gQq+f/aBVmwDuYZCRDLJY0+r7k5La11JeHs7OwGr68jFEsjXVuzA5NJkyahqKgIL7zwAnJzc5GQkICVK1ciJCQEwPnxwqysLL78Bx98gNraWtx///24//77+f133XUXlixZcuXvgHQdV9J7cZX2QnRWwnV0GGPIasKsnaZo6QUBVUk9EPTxxxCJRB4ZpNQNSurKzTiCd6b+jd9ubv6JM6sCuc9uhSxQC++74hp8XuGiA5cdznnsscfgdDp570llZSXkcnm94KQlQzxffvklwsKb9RTSRbQo+XX27NmYPXt2g49dGGysX7++JasgVyPqvbjqXMmsndZiT9+LY8kpANo2V6WtpwC3NP/EeaYS+f/c0fBj2RVwVTkh0TQ++2bp0qXIycnht+sWXhO0ZIjn7NmzFJhcpehaOYSQdseLrtUZfolYuwZilQoipbLNei3cdvsle1bse/bAVVwMibd3q7fhmxefRmHGQX77u3+92OzXEMrVM8ZQ66gBADhrahrMP2kteS9thzxYB/X0uAYfrxuUNEYYmqHZlqQpKDAhhLS7rGnTUb1vX737hFyR9pphE7lpI87Mvv+idmQMGMiHd8Rqdau1I/f4UdTtdyg4cYz//8JpsY2RKZSQKhQXXYlYUPeKxOOe+ueVN/p/HFkVqHp/3+UXBPDQQw9BLpdDJpPVu1ifULaekMtp+0nqhLQ1xi6eUnwVFC/rzC4MBupqr8JqIpGo0XYIwzuZd0xu9SGXme9cXIH0u/kv/PX/y/Sk1NY0fiViwdmjh+CsrmlZAxtbb17TLr8gl8uh0WggEonqXaxPqOxKyOVQjwnp3Bqabvx65FVXvKyzity0kfdINDbDpj0SUiPWruHrrqsthnZkiotzQArqJLjW7Um5HGFoBzjf0/L1C0/yZNnv//UCgCFX1tgWEKb7Tp588ZWRP/vsM0yZMqXd20Q6FwpMSOfW2HTjtixedrmCcJfTnIJxl9KUYnIerqGhFKD+DJv2GNoRpi4DgDIhAdUH/iralTFgoMcWcKt7JWJndXW9GTz5J44Boe0fmAB/TRu+UE5ODqqqqjqgRaQzocCEdB2PHT//b1tOG27NcvbAVV/S/lJDOoL2Lp5WNyjpqDZ0VlarlZeYd7vdDS7T0KwdQuqiwIR0He1RwdVTytkDXaqkvVBwrW6xtY4c2hHaBLS8PsrVqO51b5YuXdqBLSGdGQUmhLRUe5WzZwxw1kkGddqAtxLP/7+hoaBWHOKpm/jpttvbLCgQq1QQqVSNFltr76EdABddcLC1k2Bbou40YeD89OArKU3fluoGKYQ0BwUmhLRUexSEu9zQUUNDQa04xJM96z4Ir3KluRaXO7A3tdhaW9YaqevCKc1ZU6ciZNmydrnibmOWv/Q08o4f5bffu2cyL01PSFdB04UJ8WQtGToShnhaQc0F+RbNncpbNxjJnnVfk58XtXkTYvbsRvTuXVAmJl70eMaAgci87Ta4qqrgttnapDfjwvyX6v0HcHriJDgLC9tsnZdTNygRCKXpl//zH+3eHkLaAvWYENJZXG7oqA2vF9TYdNrLqdujcWGQcynCtXTcNttla40A7Te8U3PgAI4PHAQAUCZ2R/Ann7RqEbammvH2Inz8UP1iZXkZR9q1DYS0lc4dmFzptM3Laa1pnZfTBaZ9knbQgdcSqjudtr0wxuolvYavWY2TI0Y2uKwwvCPkqrRHoFC9bz+OJad0SIDy01v/4v+f8fZH+Pihu9tlvYS0h84bmLT2tM3LacspqF1g2mez1A0oLxX8eUrAdrn2eko7O9iFgYTbboeomReUq/taFybC5jz8CP//hbVGgL8SZNuinPylCAFKe663bhG2nxf86xJLEtL5dN7AxJOmbV6p7G1AVWHzZ3h0xgPipQLKC4M/TwjYmtLe9m5nYz2Fl+vha8PthTGGzAsCiYwBA6Hs0aNlr1ddfVEibN2hoIZqjQiEIZ72DlAaWm97qFtUjZCuoPMGJnW117TNxjAGfHYjkLOr5a/Rkh4ZTzhwN1dzAkpPqNPRlPYKgaXG3LbfBWNATSXw+Y3Amctsaw1tT4G9gTt/OP95tnI7G5tRU7137xW/9qXyWy712OUChQunQreWuus1L/ms1V6XkKtF5wpMGAMc/ytnfKmcj/buSXBUXVlQ0lKecOC+Eo0FlG2YxHlFHjsOyFQNB6GvR/514BeJ2mYbvNLg98xO4JWAFgW0dQ/cl5uN0tqFyS6V31J3qKixHJS6gULgor8uoFd3llBLEnsvx56+F1lTpwOht7X6axPSlXWuwOTzm4DC3Rff70lDAO3Re+OpB+7m6sBkzhYRvtfGggPhwA+0zTbY2HqbG+C1oIen7oE7a/oMhH61tNHhkfZMlK0bXNTNQWnI+UBhGr/dnFlCTRW2+jecGjnqr3Xs30+BCSHN1LnqmDT1bLEV6zg0m3CwbdM/ul5Hgxir35N24Zm90ON2YS6Go+r8X3PrUsy5xHVehIN/S173ch47/td1gYBLbHN1tpPHMgD/Xn/dfj0SWHxdi9pWvXdvs2qZNEVLa4LUDS6aEmg0tEzE2jUtWndDTt04vtVei5CrVefqMRFceIYolOxuqFR3Z0wQJU1X94B24VDHZ+OBmWvOf/+NJbHW7VFobi+H7DIBovDadYd46j63pdulTFW/RL3D1oTXEwFn99S/K3vb+cBJoW3Saltay6QpsqbPQPDijxt8rDUKmYlUqkaDqVbt4WnlgI2Qq1HnDEzqDgFc7oDTGRNESdMISceCC3vUcnb+lYPT1CTWC3N26h4UHbbzQUFz1R3iEVzJdnlhAFY3v6Upia3WHkDu/5JSPx0L3PVTk57XlkM01Xv3Nnpdmrr3t1Rr9/AQQtpO5wxM6p4hXu6A014zJsiVE6bBNrVWiNPW8PDenL3AW5eYpnphj1tjuRgXBj6vRwIBvS//PpriShKXG3rPQvAT2BuYvhq41PVccuvMlDm7p8nPqxcstEE59sbK17dVLw0hxDN1zsDk9ci/zjjreiwDWHrb5WdMADTE42las+dLdpmDfVOTbhsKfHJ2Xv55dQOf5lwZuDW2yTM7gUXXAjN/b/7zPh55fuirsUXqBAuXS4BtibrBSFskphJCOofOGZgAjSS4ipo2YwLo2CGeKy2l31ql8j0pOGtKz1dHTo2+XC9MXULg09wrA7fWNnl2z/ngZMqPzXtezs7zOSeihqu11i1qJiTAilqxiFhDwYgiLg41hw612joIIZ6vcwUmD6YDel3rTJXtqANda5fSv5LPoqGkzEtpjUCm7hBAY8MBdXscWjQ1ugnraK7L9cI0xFHVvO+5NYcdz+4Bloxp2rIPpgPvJJ3//6djwe789VJLtysKSgi5+nSuwESuav5UWeEg19jMnfbuNfCkUvoNJWVeypWe0V+Ys1F31kxdV1rf5MtJF6+jLTQW9DB28XsV+PcCblv213bcUCJra/Wc5NXJJblUgPbN9L8WO7sHd385GMAzV7ZuQghpoc4VmLSEXH0++Ggsf6ErFGNrr4JrV3pNnwtzNurOmmlNdafFCutoC3UDoLo+Gw9M+aHhYcWze4A3os73Vk3+vuFl2qI378uJjT+W+9fnZReJsF9qgy72CSgcDEqXq9GntUUCLCGEdP3ABLh0L0X2NqAkC9Caz99uzx6Utqh8erlg50qv63Ml1/TxNI3NAmrqNpC7p+H7c3bWT3htyJmdwKfj/rr92HEADHg9qn47WkvuxdexuVKXSoCloIUQ0lJdKzBpaGd44ayIxzLO7/A/HfvXmfXbiX893tnrnlwu2OmI6/oIhbw86TO91Cyg9gqk6gY2wjV46rZD6FXxUI0lwDLGkDVteiPP8jwURBHiWTp/YFJ3p9JQ1/qFvQNLbz/fzX5hFUxB3eEKT5q10hba8ro+jNUP/oThjea+xqVKzF+Jy/WitfclDRwNTE0+s7PpCawehFVXo3rfJcr1e5i606AJIR2v8wcmdQOHhrrWG6pDcbkptnVLiU/+Hqj9X+VJ4SB+qYClI86+mjLTpSFteRE9R1UDuR4NDG+43Q1fv0aqAj65rn7w0Fiy7OUwdunnCAHaJXN1mvm9ttZ2UDeB1QO57XaI2vGifW2hup1rplAPDSGX1vkDk6Z6KB14O+n8/xtLWrzQmZ3Aq4EX339h+e9612sZ37KDZ0tdaqZLQ/VS6gYAlYWA/ILHG+tBaaveoze7AY7K+vcJFVYvLGbW0mTZy/XWNCVAa+o209Lll14iOdWDZQwYCFWvXgj88AN+Hx14/9LQJ9GaRekI6Yo6V2DSUNd+U3/k8joXKmssabGphGm2DeWjtNVMk8Y0NtOlsZlIddXNrbmcC2ueXJi7UzegaU7S5oVBiaBuUNKc4maNvdblklGBS/c8NTb015jmbmNnWz85tb3Y9+xB1tRp/HY2DY0AOB+UpPd4sKObQUin07kCk6W3AcXpf91uSd5CU/mnAMx16dkMnpjUCZw/qFYVtm69lObUPGksafNSZ9KXGk6pG+i0Ve2Zhnqe2mrb6iLqXm24btVWKid/nlssh10fjsdLAcacqGG1Hd0kQjqFS1zpywNdGCQ09UwYaP6Y/9ldDQcl1l7ni2QJPh17Pk/Ck3x+48UH94DewLyc/01L/Z/Z29pm/RdOhRVcanhDGE5paCip7vNejwQ+HnHln/mFvSMN9jzRFWkvpS2vNkwIuXp1rh4TQd18kaYGHM0d82/MhV30Z/c0fBDuSGcamA6cs/N8L0PdA7+0zv/n7APUpoZ7IoQhtKUTGx9y8O0O5O//63ZDQxlNGd5o6Pu8cBhFuOBcS3s0mto7QrkShBDS7jpnYFKnhHaTA44rzStp7ms3dqG+plyAr7WGKhos2lXnLLdujslbieeHYCZ9+de6hVL+/7358sNCdYOSlmqsjHtDrqRHw2lvWu9IawWzhBBCmqxzBiZ1z6DbMuBoqaZeqK+xqamtVeRNqgQ+H19/fQG9G1/+zM7z5dI7QkPDKU15Tmuu/0KeuG0RQkgX17lyTDoLp/3KEk+FIm+Oqis7+H46ruE6Lp7os/HNf6/N6dG4XK0X6h0hbajuFGqaTk3IpXXOHhNPV3fHc6nqqpe6bo3Qm3IlvSd1z/itPYDcRop1WXs2nOh7Yd5IW2pK4bsLNadHo26dkIaCEOodIW2EMQZHxTf89nf/erEDW0OI56PApC18cUGxrMZyRppy3ZqGrujbkhyUxoISoPEp0e0VlAgudQXcK1U3aZeCkFZBZ/5NVQu48/itghPHOrAthHg+CkzaQl6dg2Ddi8JdKpi4VI/GhbkodSvPdqWDQxtcAZe0HbrGDCGkLVBg0h6Ei8JdqhrspXo0LiQUOwtI6VqBCelU2vsaM4SQqwMFJm2ppaXUL8xLYQxYckMDBeaaMYOFkCagMJcQ0tEoMGlLshZeL+fCi8o5qmiYg7Q5BuBuP5+ObgYh5CpH04UJIQAAu0iE/UpFRzeDEHKVo8CkLTkvqPLa1HwQyhshhBBylWpRYLJw4UKEhYVBqVQiOTkZGzduvOTyf/75J5KTk6FUKhEeHo7333+/RY3tdN6qU/JduPhcTeXlC6d54oUBCSGEkHbQ7MBk2bJlmDt3Lp566imkpaVh0KBBGD16NLKyshpc/tSpU7j++usxaNAgpKWl4cknn8RDDz2Eb7/99oob3+kIs2le9gc+uhaoPNfwcmf3nL9IHfWcEEIIuco0OzB58803MWPGDMycOROxsbFYsGABgoKC8N577zW4/Pvvv4/g4GAsWLAAsbGxmDlzJqZPn47XX3/9ihvfqZ3dDbx9iRk7OTuBkqzzwUtzK6ISQgghnVSzZuU4HA7s3r0bTzzxRL37R44ciS1btjT4nK1bt2LkyJH17hs1ahQ+/vhjOJ1OyGSyi55TU1ODmpoafrusrAwAUF5zlfUgvNa9o1tAriI2AC67i992ORgqXa7Gn0AuycVcsDuqwJgTDqez3dfvFDG4a/46qamEGLVo+YVBq8HgxvnXq0E13KK23R+LRE5UVbXtkHYNXHCLKlvt9cqr3XC1oKe7QsTq/faa4kp/n9Vw1ds+6j3mcMEFEZxgcNc5FreEsI7mVIpuVmBSWFgIl8sFX1/fevf7+voiLy+vwefk5eU1uHxtbS0KCwthtVoves4rr7yC559//qL7g/7dehsQIaQh5fVu9emgVnQZx8Z1cANW8v/1asVXfaMVX6vjDWq1V/K/omeXNfsZV/z7PNrwZUCevNLXbUBFRQUMBkOTlm1RHRPRBaXVGWMX3Xe55Ru6XzBv3jw88sgj/Lbb7UZxcTFMJtMl10MIIYQQz8EYQ0VFBfz9mx62NSswMZvNkEgkF/WOFBQUXNQrIvDz82twealUCpPJ1OBzFAoFFIr69RS8vLya01RCCCGEeICm9pQImpX8KpfLkZycjDVr1tS7f82aNejfv3+Dz0lNTb1o+dWrVyMlJaXB/BJCCCGEXL2aPSvnkUcewaJFi7B48WIcPnwYDz/8MLKysjBr1iwA54dhpkyZwpefNWsWMjMz8cgjj+Dw4cNYvHgxPv74Yzz22GOt9y4IIYQQ0iU0O8dk0qRJKCoqwgsvvIDc3FwkJCRg5cqVCAkJAQDk5ubWq2kSFhaGlStX4uGHH8a7774Lf39/vP3225gwYULrvQtCCCGEdAki1pw5PIQQQgghbYiulUMIIYQQj0GBCSGEEEI8BgUmhBBCCPEYFJgQQgghxGN0+sDEZrNh+/btvJrs/v37G1zuxIkT9W6vXbu2zdvWGvLz8/n/z549e9nlX3jhBbzwwgv4xz/+gccffxwvvPBCg891uy9/DQpnB1zfoyHV1dX44osvMH/+fHzxxRew2+0d3aQmKykp6egmXHVyc3PBGEN+fj7++c9/4r777sM///nPRi+bcTk//fQTXn/9dfz888+t3NIr09R9X0PLrlmzBitXrkRxcTH+9a9/YdeuXQ0+ryn7idaSl5eHZcuWoaqqCgCwfv36i5bZvHlzi1//s88+u+ivrlOnTl3yM7zQhctXVrbskilHjhxp9LH333+fl+IAwD+bffv2NWs/6Ha7UV5eDpfLha1bt6K6urpFbW0vHj8rJy0tDVu3boVYLMbevXsRERGBLVu2YPz48ejZsyfuuusuKJVKnDlzBi+++CI++OADrFmzBmlpaTCbzfjuu+9w5513Yvr06fjwww+xadMm+Pv7Y+nSpVi0aBG2b9+OH3/8EVOnTkVUVBQ2b96MAQMGAAB+/PFHDBkyBGfPnkVgYCB+/vlnFBYW4r777sPSpUsxYcIEfP311/Dz80NFRQXGjh0LpVIJ4PzFC3/88Ufceuut6NHj/FWEq6ur8c0332Dy5Mk4d+4cPvroI+Tk5CAmJgYjRoxAbW0tunf/68J9M2bMgE6ng81mw8SJEzFv3jwAwKuvvorly5fj7Nmz2LNnD1JSUnD27Fn88MMPGDJkCEJDQ3H06FEMGzYMtbW1+PTTT/HQQw9h4cKFyMzMxKZNm7Bo0SKYTCbY7XY88MADGD16NIYPH461a9di8+bN+OWXX/Dzzz/jlltuAXB+x1ZcXIwpU6Zg4MCB2L9/PyorK3H69GkEBQVh27ZtCA8Px9GjR3HzzTcjJiYGzzzzDGbNmoUnnngChYWFSElJQVVVFaZOnYqQkBDo9fp6n/fZs2fx0Ucf4YcffsDtt9+OmJgYjBs3DtOnT4e3tzemTp2KrKwsvPvuu/j+++/x3//+F1VVVbjtttuwatUqnDp1CnPnzoVerwdwfke9fv16WK1WfPPNN5g2bRqUSiW8vb1RU1ODJ554AlqtFj/88ANsNhv++9//4u6774ZIJMKqVaug1WqRmJiI77//Hjk5OThy5Ajuu+8+fPDBB/D29kZAQABuvPFGfPbZZygsLMTdd9+NPXv2oLq6GnK5HH/++Sf27duHe++9F4mJiUhKSsJHH32EBx98EDt37sRHH32EDz74AN9++y3fDvr06YMdO3agpKQEt9xyCzQaDdavX4+wsDAsX74chw8fxvHjx3HrrbfiyJEjMBgMKCsrw88//wytVos33ngDFosFfn5+mDx5MuLj4+Hv7w+lUonbbrsNFosFP/zwA6RSKYxGI8LCwvD777+jd+/eWL9+Pc6ePYv8/Hyo1WqcOnUKy5Yt4+1zu92QSqXYsWMHBg4cCLFYjHXr1uGbb74BAP5dvvnmm/j4449x5513YvXq1YiJiUFERAQyMzOxd+9eFBUV4auvvsKaNWswbdo0OBwO3HXXXXA6nfDx8YFarUZAQABqa2uxdetWVFVVoby8HEajEbfeeiu2bt0KtVqNbt26Yfr06VixYgUWLVqErVu34oUXXsDy5cvRu3dvFBcXIzY2FjNmzMB//vMf7Ny5E3/88QdeeeUV/Pzzz0hNTcXLL7+Mr776CqGhoRg6dCjmzp0Lo9GI8PBwvp5ff/0V/v7+yM3NxaRJk3DLLbfg0KFDSEhIQF5eHiQSCYYOHcr3GePG/XV9nI8++gj9+vXDtm3bMHz4cNTU1KC2thYPPPAA8vLyoNFoUFtbC7VajSeffBKHDh2CzWbD888/jz/++AMZGRkYO3YsnnrqKQwePBgJCQlYtWoVHA4HNm/eDF9fX6xevRpTpkyBw+HA5MmT8dlnn2HatGlISEjAsWPHYDabcf311+Omm27Cjz/+CKPRiB07diAsLAzZ2dl49dVXsXbtWnz77bfIz8+Hr68v3G43Nm3ahPnz5yMiIgIlJSV4+umncfLkSdxzzz3Q6/UICwtDQUEB+vfvD5vNhp49e6Jv375Yvnw5Nm7cCJvNhri4OBQUFGD8+PGwWq3Yv38/7r//fmzduhXXXnst/21FRUXhk08+wYQJE/Dbb7/h5MmTkEqlmDZtGpYuXQqVSoXKykocP34coaGh+PTTTxEYGIg1a9ZALBajrKwM/fv3h4+PD5577jm88sor2LRpEwYOHIjKykpotVr4+/sjNjYWU6ZMgZ+fH9xuN/r3749bb70Vw4YNw4YNG3D69GlERETg66+/xmeffYa7774b//jHP/Diiy/igQcewKuvvooZM2bg4MGD6Nu3L/bs2YOAgADcfPPNePjhh2E2m3HttdfirbfewsKFC/Hnn3/Cz88PkyZNwr///W+MGjUKo0aNwtq1a7FixQpkZmaiW7duCAsLww033ACXy4WSkhIsWrQINTU1OHDgANatW4cjR47gl19+gV6vR2ZmJpRKJSQSCV5++WV88skn2Lt3L3788Uf8/vvv+PHHH2EymbBnzx589913+M9//oP169dDrVbjxIkTCAsLQ2lpKQYMGAAvLy+EhYVh48aNMBgM+Oabb3DXXXfhiy++QHx8PIxGIx5++GE899xziIiIgFqtxmuvvYZ+/frB398fw4cPxyeffILbbrsNffr0gdVqxWuvvYZVq1bBarXCbDbjwQcfxPHjx+v9Li7HowOT1157DTt27IBGo8Hnn3+OoKAgVFVVISoqCnl5eRCLxcjOzoZGo4FEIkFxcTF0Oh0MBgMcDgeKiopgNpuRl5cHrVaLyspK6PV61NTUICEhAYWFhcjJyYFKpUJERAQOHz4MiUQCAKitrYXb7YZYLIZUKoXdbodIJIJMJkN0dDRsNhvKy8uh1WpRWlqK8vJyKJVK1NbW8nVUVlZCpVLB7XbDy8sLZWVlUKvVkEqlKCoqQmBgIHJycuDn54eCggIEBQUhJycHMpkMfn5+OH36NEwmEwoLC6HRaGC323HnnXfiiy++gMFgQHV1NcrKyuByuRAcHIzS0lJER0fj4MGDqK6uhtFoxI033oicnBxs2rQJJpMJLpcL586dg9vthlKphF6vR21tLSQSCex2O2pqavhlBJRKJWQyGb/Sc2hoKDIyMuDn54fKykqUlZVBoVCguroaMpkMDocDfn5+yM3NhVKphN1uR0hICM6dO4fU1FTk5+cjNzcXpaWlkMlkCAwMRG5uLpxOJxwOB1QqFWpra1FbWwu5XA7GGLRaLaqqquB0OqHT6RAcHIxDhw4BAIKDgyESiZCbm4uqqirIZDK43W5069YNvr6+OHbsGJxOJ86dOwcvLy/Y7XaIxWLYbDYoFAp4eXmhZ8+ecLvd2Lt3L6qrq1FdXQ2LxYKysjJUV1fDarXygzJw/sxDLpfD4XBAqVRCo9Ggb9++2LRpEyorKyESieD63xU/xWIxRCIRGGMICQlBVVUVRCIRqqqq+OcTHR0NnU6H3bt3w2q14ty5c/z5TqcTFosFOTk5EIvFYIyBMYaAgADk5uZCKpWitrYWjDHIZDKIRCJUV1dDLBbzbVcsFvPXU6lUMBqNvOfAYDCgtLQUSqUSYrEYDocDWq0WCoUCFRUV/P2p1Wrk5+fDbrfD5XLBaDSiqKgILpcLCoUCPj4+fJvy8fFBdnY2jEYjSkpK4O3tzT9XlUqF6upqzJgxA99//z2io6Nx7tw5FBUVQSQSoaamBk6nE06nE1KplH/mwcHBsNvtsNlsKC0txaRJk7BixQpotVoUFxdDrVZDrVYjJiYGW7ZswYABA7Bz505otVrIZDLccMMNyM3NxapVqzBy5Ehs2LABNTU1kEqlUCgUfHt0u93Q6XSorKyE2+2GSCSCVquF3W6HxWJBZWUlGGOw2+1QqVR8exCLxejXrx8YY9i6dSv69u2LXbt2wWazgTGG8PBwnDx5kn9PEokE4eHhyMjIgMPh4NuIVCqFy+WCy+WCVCqFVqvFsGHDsHPnTthsNlRUVEAkEqG2trbetmU2m1FaWsp7OM1mMwoKCqBWqyGRSFBVVQWDwQCdTgeZTIa8vDzY7Xb4+PjwHhTheibV1dXQ6/VQKpUoLy+Hw+GAr68vxGIxSktL4XK5YLfbodFoUFVVBZVKxX+7wrqE9judTiQkJEAmkyEzMxPh4eE4ffo0UlNTcfDgQdTU1KB79+4oKSnB4cOH+e9HeF8ikYhv92q1GpWVlZBIJPz9BgUF8e1S2OZFIhGkUini4uIQGBiIPXv2oLKyErW1tTwoFE5MhH1BTU0NGGPQ6XQQiUTo2bMndu7cCYlEgpqaGshkMrhcLnh5eSE/P58H6FKpFAkJCUhLS4NarUZFRQVCQkKQl5fH34NOp0NFRQXkcjlkMhmUSiV/XCqVYvLkyfjvf/+LHj164OjRo6ioqIBYLIZSqeSv6e3tjbCwMOzcuRNTp07F559/DolEApFIBKPRiJycHPj6+vJjjtPphFKpRGJiIiwWC9avX4+KigpoNBrU1NRg6NCh+PPPP5GSkoLdu3cDALy9vZGbm8uPMxaLhV9qxmaz8e9VJpPB6XQiIiICZ86cgdvthkKhQFhYGI4cOQK9Xo+KigrodDq43W5MmTIF27dvh0KhaLAHrFHMg91///3swQcfZCEhIUyn07GYmBgmkUiYQqFgEomE6XQ6JhaLmVarZRKJhJlMJqZUKplSqWRyuZyFhYUxsVjMdDodk8lkTKVSMQBMpVIxmUzGoqOjmUgkYmq1mlmtVqZUKplYLGZ+fn5MJpMxmUzGRCIRk8vlbOLEicxsNjOxWMy8vLz4YyKRiEkkEqbRaJhEImEGg4GJxWI2aNAgJpVKmUajYRaLhT8nPj6eyWQyJpVKmV6vZwCYWq1mGo2Gty8wMJAB4PdJpVKWmJjIADCdTsdUKhUTiUQsMTGRBQUFMZlMxiZPnsxEIhHz9vbmj6tUKhYdHc0kEgkDwGJjY5mXlxcDwJRKJTOZTMzHx4epVCpmMBiYVCplBoOBP/f222/nn7FcLmfdu3dnAJhUKmVisZgpFAqWmprKAgICmEKhYGKxmCUkJDCxWMxkMhkDwBQKBV+fcL/JZGIBAQFMLBYztVrN9Ho9E4lELCgoiKlUKqZQKJjRaGQ6nY6JRCL+Wmq1mkmlUqZQKFi3bt2Yl5cXCwgI4N+RVCrl3wUAJpFImL+/P5NIJKxbt25MKpXy1wsMDGRisZglJyczhULBfHx8LnodqVTK34NMJmNjx45lMpmMRUVFMalUyqKiophCoWAGg4ElJSXx55tMJubv789EIhHz9/dnAwcO5NuHRqPh24XQRqvVytRqNZNIJHy7MxgMTCKRMJVKxUJCQphMJmM+Pj5MLpezXr16MavVysxmM9+WZ8+ezaKjo1l0dDQTi8XsuuuuY2q1mk2dOpUBYCkpKcxisfDt1cfHhwFgISEhTKFQ8Pbq9XqmVqsZAGaxWPi2KhaLmbe3NxOJREypVDKDwcC8vLyYRCKp95tMTk5mIpGIhYSE8O9ZJpMxpVLJFAoFUyqVzM/Pj4nFYv7bE7Z1uVzOjEYj8/LyYnFxcUwulzO5XM58fX2ZwWBgOp2OAeDbkfB7TElJ4Z+9SCRi11xzDZNKpfy3IxaLmVgsZhqNhpnNZiaRSJjRaGQRERH899yzZ08mEomYQqFgcrmc3Xjjjfy2Wq1mN910E1+vUqlkYWFhfB0SiYQNGTKEf7ZSqZTJZDKmVqv5+9br9Uyj0fDfl1wu5/sQjUbDRo0axcRiMQsODmbe3t5s/PjxzM/Pj+831Go102q1TKlU8jYEBQUxo9HIlEols1gsLCgoiK9HeO1u3brx/YLw+QifubC9CfuhsLAwJpFImFgs5o8LvxfhvQwbNozJZDI2cOBA/tkIy+j1ev7dRkZG8t+IVqtlMpmMfw/CdyLsP4X7hM9PLBYzlUrFVCoV8/LyYmq1miUnJ7P4+Hg2fPhwZjAYmMFgYCEhIfy59957L/8dy+Vyplar+TYpfKdSqZSp1WomFov5PrHu71DYz3l5ebGgoCDm6+vL1Go1k8vlTCwWM6PRyAYOHMj0ej2Ty+VMp9Mxo9HI5HI569atG9NqtSwqKor16NGD6XQ6JpfLmUKh4N+hsO8UvjOdTseCg4P55yB8Bj4+PkwmkzGtVstMJhOLj49n99xzD9Pr9byNdbc14TMUvnN/f3+m1Wr59i/sx3Q6HfPx8WEhISFMKpWy4OBgZjab+fYqFov55ye8nrBtCL9z4VhqtVqZt7d3ve9LWE9cXBzTaDRs3Lhx7JprrmGjRo1i48aNa9ax36NzTAoKCrB27VqUl5+/FHtWVhYYY/xsWqVSQSQSISkpCVKplEe7LpcLEokE2dnZCA4Ohsvl4mf/FosFarUaMpkMjDF4e3sjMTER11xzDfr164eePXvi5ptvhtVqRWRkJI/ES0tLodPpoFKpeE+JwWCAxWJBQEAA/Pz84O3tjQkTJkCv12PHjh0Qi8UYPnw47HY7KisrIZVKYbVacd1110EikcDpdEKhUCA2NhYajYZfuDA6OhoSiQQ9evSATqeDWCxGZmYmJBIJZDIZvLy8oFarcebMGeTl5cFoNGL9+vXQarUoKSmB0WiEXC6HVquFRqOBUqnEyZMnsWrVKjz11FMQi8Xw9fWF2WzmUbRcLkdcXBwcDgcYY3A6ndi4cSMsFgvEYjECAwPx+uuvY+TIkbBYLPwqz6WlpUhISMDAgQNhsViQnp4OlUoFi8UCk8kEq9UKtVoNs9kMkUgEt9vNP/vIyEh+Wzg7cDgccDqdCAkJwejRo9GrVy90794dGo0G11xzDRQKBVJSUnDs2DGUlZWhvLwcQUFBUCqViIyMhE6nw6233opu3bphxowZqKmpgcViQVZWFgYNGgSlUgmn08nzELKzs6FQKDBt2jRMmjQJERER6NmzJxQKBTQaDXx8fPjZ0Z49exAUFITTp09Do9Hgxx9/5D1GBw4cgEKhQGBgIG6++WZ89dVXEIvFGDNmDL777ju888478Pf3h8VigUwm41falMlkOHfuHAICAiAWi2GxWKDT6eB0OiGXy+Hr64s+ffrAaDTCz88PwPmejoCAAAwfPhyMMQQGBmL58uUoKCjgPXu7du0CYwxfffUVRCIRUlJS8MYbb8Df3x+MMcTFxaFfv378gpoWi4X/hoQzrokTJ6JHjx78rFqr1cLHxwd+fn4ICAhAaGgonn76achkMnTr1g0hISE4ePAgGGMoLS0FYwwFBQXQarWoqamBy+WCw+FAbGwsRo4ciREjRmDChAmwWCyw2+1wOp2w2WwICwtDZmYmtFotXC4XSktLea/d6NGjYTQa4XQ6eU/L4cOHodFoMHXqVBgMBqSmpiIwMBC33XYb5HI5brjhBoSFhaF79+7w9vbmPU0+Pj7Q6XQoLS3lPRo1NTVwOBw4duwY+vTpA5fLBZvNhs2bN8Pb25v3lMTExEChUMDtdkOv16OgoAAqlQrz5s1DaGgo7/ZWKBS8R1LYTwi9Hk6nk69z7969iI2NRUFBAZxOJ0pKSuB0Onmvq9PphJeXF9//eXl54cyZM6isrERsbCzcbjd69+7Ne5qEs+4TJ06AMYbQ0FAYDAb87W9/g5+fHyIjIzF27Fj4+voiMTERGo0GMTExSE5ORnx8PGJjY+Ht7Y2RI0eCMQaJRAK32428vDzU1tYiPT0darWafy/Cvkwul8NgMCArKwtutxtFRUWQSCSYOnUqevTogVGjRmH+/PkwmUxISEiAVqvF0KFDERERgaCgIMjlcj5MOnr0aEyfPh2hoaEIDAzkOR1VVVWw2+0oKiqCTqfDDTfcAH9/f+h0OhQWFkKtVkOv1+PRRx/lvT9CT6ZIJIKvry8GDBgAhULB90/Dhg2DQqHAgAEDMHfuXH7M0Wq1kMvlMBqNqKqqwt69e+FyueDt7Q2xWAy73Y7a2lpkZ2fzz7myshIajQY333wzxo8fj/j4eFgsFvj6+kKhUCAmJgZisRh6vR4OhwM+Pj6QSCTo3bs3vL29IRKJEBkZCZPJBJvNxof6jEYjdDodJBIJIiMjYTabIZVKIZPJYLVaMXz4cIjFYhQVFcHhcECtViM8PBzR0dGIiIhAcHAwvL29MWzYMCiVSowYMQI33HADBgwYAIlEAqlUiuLiYsTFxcFoNCIiIgJSqRS33XYb9Ho9RCIRVCoV4uPjccMNN0AkEiExMRG9e/eGRCKBr68v5HI5iouLUV1djeLiYuTm5uKNN97gaQhN5dFDOYcOHcLq1auhUChQUFCAo0ePIjg4GJs2bYJOp8OiRYvwySef4M8//8SkSZNw22234fHHH8dLL72EX375BUVFRbjxxhvx0EMP4fDhw0hOTsazzz6Lzz77DFlZWbjvvvt4d2V8fDxOnDiBiIgI/P3vf+fdYydOnIDT6cQLL7yAoKAgfPTRR5DL5QgJCUGvXr0wbNgwPPvsszhz5gxKS0vx7bffYvXq1aiqqsLBgwdx5swZpKenw2azYdSoUfj+++9RW1uL4cOHw9vbG3/++Sf++OMPvPvuu4iLi8MHH3wAq9XKx2hjY2NRXFyMoKAgPmygUCiwYsUKxMbG4v/+7//w448/YsmSJRg0aBAiIiKwevVq7Nu3DzabDcD5/BCNRoPo6GisXLkSr7/+On766SfMmDEDc+bMQUBAAEpLSxEbG4uQkBC89NJLOHToEE6dOoVVq1Zh79698Pf3x/XXX8/HU1944QUkJCTgl19+wTfffIOJEyfi4YcfRlpaGkpLS2G329G/f38EBAQgISEBX3/9NcRiMdauXYuioiI8/PDDqKiowKpVq/Drr79i5MiReOaZZ/Dyyy/jyJEjGD58OMaPH49Vq1bhsccew/79+6HX6/Huu+9i7dq1OHjwIHr37s2HwwoLC3HjjTciOzsbL730EmQyGV566SV4e3ujuLgYc+fOxcKFC/Haa6/hn//8J0JCQrBs2TJcd911kEqlGDNmDD766CPExsYiKioKixcvxqhRo/DEE0/g3nvvxauvvoqUlBQMHToUSUlJSEpKgr+/P3bu3InevXvDarWid+/eOHr0KMrLy+Hl5YWpU6di8+bNuOWWWzBlyhRUVlZi/fr1qK2thdFoxE8//YShQ4di8+bNCA8Px5dffolJkyZh48aNcLvdiIyMRFBQEH744QfEx8fjueeew8qVK1FaWorVq1fD4XBg7dq1UCqVcLvd0Gq1AACHw4GYmBgMGDAARqMRMpkMvr6+6NGjB3bv3g2FQoF169bhuuuuw/jx47Fv3z6o1Wr88ssv6N27N1QqFVatWoXAwEC43W74+vpCpVJh6NChePfdd7FgwQKo1WoEBwfj8OHDePLJJxEREYHjx48jKioKjz76KJYsWQKRSITQ0FCo1WocOHAACQkJ+PPPP3H06FHcdNNNEIlEMJvNcLvdyM7ORkBAAKRSKXbt2oVVq1YhNDQUp0+fxj/+8Q/84x//QGhoKKZNmwaxWIydO3finnvugUqlwowZMyCXy+F0OlFeXo4ffvgBDocDEyZMwI4dOzB8+HB4eXnh22+/xVdffYU//vgDdrsdZ86cwaZNm6BQKBAdHY1x48bhn//8J4KCgnDPPfegtrYWvr6+SEtLQ2FhIb799lsUFBRALBbj/fffx7fffguHw4HevXujpqYGmzdvxrBhw1BaWoqDBw9i1apVGDFiBIKDg+Hj44NHH30Uixcvxp9//onrr78eH374IZKTk/n+bvz48Vi0aBFOnz6NsLAwTJ8+Hb/++ituvPFGGAwGiEQi7Nu3D5mZmejXrx8fXnniiSeQn5+PxMRE9OjRAx999BEmTZqE+Ph4HDp0CL/88gsGDhyItLQ03Hjjjbj77rv5MGhNTQ2GDx+OI0eOoHfv3ti2bRu++OILbNmyBUOHDkVkZCRv39dff41PP/0UMTExeO6557BixQrY7XZce+218PLygtFoRFZWFs6ePcsPojk5Odi4cSN8fHwAAN9//z2sViuef/55/Pzzz/jzzz/x5JNPorKyEvfeey/Gjh2LXbt2weVyYdeuXdDpdCguLoZGo4HL5UL37t1hMBjwr3/9C4sWLcKkSZPw22+/4dixY/D398eJEyfgdrths9kgl8uRmJiInTt3Ijc3F8nJyYiJiUFOTg7GjBmDAwcOIDo6GtXV1XjmmWdQU1ODXr164aGHHsKrr74Ku92Obdu2wdvbG3q9np+QAkBUVBRGjRqFESNG4PPPP8enn36Kmpoa+Pj4IDc3F6GhoYiPj0dgYCB27dqFsWPHYty4cTh16hReffVVqNVqOBwODBgwAMeOHcPLL78Mb29vXH/99RgyZAiOHDmCkpIS+Pn58XyqF198EZMnT8bBgwdhMBjwwAMP4Nlnn8WsWbNQXl6OYcOGoX///hg9ejQkEgmOHz+OoqIizJkzB2PGjIHJZMJbb72FX375BYcOHUJUVBSee+45DB06FH/88Qf+/e9/IygoCP/+97+xefNmXHPNNfj8889hsViQl5eH0aNHY+nSpVi8eDG8vLwgk8lQW1uL2bNnY+3atcjOzoZer0e/fv0gkUgwZ84cZGZm8hMq4cS7KTw6MAGAt99+G1u2bOEH+MDAQFRVVSEhIQGnTp2C3W7n4+0SiQSMMYwYMQIjRoyAzWaDy+XCzz//jKNHj0Kj0WDkyJFYtmwZkpKSMH36dLzwwgs8OBEiXofDgbvvvhvbt2/HunXroNfrMXz4cJw4cQLr1q2D0WhEUlISZsyYgVtvvRUAMHv2bADAHXfcAQAwmUwAwBM/f/31Vzz66KOIiopCcHAwAgMD8emnn6Jv377Yvn07iouLUVJSgpkzZ/LkyXfeeQejR4/Gp59+ioCAAKSnpyMxMRHDhw+HyWTC8ePH8d///hc1NTXo1q0bdDodysvLERISgp9++gkDBw6E0+mEt7c3Vq9ejcjISJw4cQLx8fGora3FkSNHeM7L888/j2eeeQYWiwWZmZlwOByora2Fl5cXYmNjsW3bNkRFRaGgoAB9+vSBTCbD2rVr+Vm+MC5vNBpRUFAAnU7Hz/qsVitGjx6NLVu2QKlU8mXMZjOsViu2b9+OmTNnQiQSISAgADt27MC3337Ld5QKhQJlZWUAzicQe3t7o2/fvvyMacuWLYiIiMDgwYPh4+OD2267DT/88AO2bt2KrKwsBAcH4+zZs4iNjYVCocCiRYvQp08fZGRk4K677sKJEydw7tw5nDlzBrm5uZDJZPzscv/+/TCZTKisrERUVBTWrFkDxhgfZ87KykJ5eTlEIhHPMZLL5YiKisLBgwfh7++PoKAgnDt3DsePH0dcXBxOnToFAOjWrRvOnj2La665BidOnMDRo0chEongcDgQFxcHf39/fP7554iJiUFYWBh27dqFhx9+GNu2bcPp06eRm5uLsWPHoqioCIMGDcLXX3+NiooKZGVlQavV8nwJIddBoVCgpqYGvr6+GDNmDPr374958+bxM3StVouIiAg899xzePHFFzFz5ky88cYbqKio4PkKmZmZ/IBvNBpRWFiIqqoq3qPocDh4Do7D4QAAaDQaxMXFYcuWLYiJiYHL5eJJo2KxGBUVFfyMWxjrZozBYrEgJiYGGzduRFxcHOLj47F27Vr069cP69atQ3BwMJxOJx5//HG8/PLLGDduHE9MttvtCAwMxKJFi7Bw4UKsWrUKJpMJIpEINpsN586dAwDeexEfHw9fX1+UlJTg4MGDkMvl0Ov1MJvN8PX1xfr166FQKGCz2XiPi06ng1wu57kG+/fvR01NDQ8UZ82ahaSkJMyfPx8ajQYHDx5EVVUVtFotpFIp8vLyIJfLIZFIUF5ejjNnzmDfvn3QarX8WmLV1dX8uzx9+jQSExNx8uRJ2Gw2nkemUqmQl5cHlUqFkJAQREVF8STUrKwslJSU8HyympoaVFdXw263Q6/Xw2q1YsKECTh8+DAqKirwxx9/8B7N2tpaAODbdXV1NSIiIpCfnw+tVouysjLeg2y1WlFZWcn3A0K+h5BfJ8wmYYzBaDSitrYWM2bMwPLly/HUU09hwYIFCAoKwoYNG1BdXQ3GGO9FDgkJwXXXXYeXX34Z3bt3R3h4ONatWwe5XA6FQoFBgwYhNDSU56n07t0bc+fOxYQJE2AwGODl5YWjR49i7969fH8n5EwIeYNSqRRKpRIVFRU8j8xqtcLhcKCqqgrh4eFgjOHo0aNwuVy4/vrrsXXrVixYsADffPMNDh06hIyMDDz00EMoLy/HH3/8gfz8fMjlcgDnZ9Po9XrYbDb06dOH5+WcPXsWp06dQkhICA4cOACTyYTS0lLeI20ymRAUFITs7GwUFhbyHtzi4mIMHjwYBw8ehF6vh5eXF0pLS6HRaFBQUICpU6finXfewY033ohly5bBx8eHP15UVAS9Xo/w8HD4+fkhOzsbBw4cQPfu3bFnzx70798f8fHxWLduHQoLCyESiXgOJwDeGyfk7uj1esyYMQN//PEHsrOzkZCQgF27dsFgMKB79+6IiIgAACxcuLDpB/6W5n+0h7vuuotFRESwbt26MYlEwiZNmsRkMhkfW4uKiuJj8lqtlueMBAcHM5VKxZRKJdNqtSw5OZnJ5XJmNpv5+KtGo2GLFy/mry2RSFhkZCTr1q0bMxgMrFu3bkyn0/H8AmGszmKxMKVSyWJiYpharWYxMTHs5ptvZtHR0SwsLIxFRUWxXr16sdDQUBYTE8M0Gg0flxPG2I1GI5PJZHyssG4eBgD+J4w9CssbjUYmEon448I4rVwuZ6GhoXwcU8jnEMaehfHaQYMGMR8fH2YwGJi/vz8LDg5mGo2GeXt7s4EDB7KbbrqJhYWF8ZyPm266iVmtVj4mKXxWZrOZabVaFhAQwEaPHs10Oh2LjIxkarWaBQUFMaVSyaRSKfPz8+M5BhqNho0ZM4aNGTOGj1UGBgYylUrFtFptvRwLo9HI/Pz8mEajYZGRkcxqtbKAgABmMBhYaGgoE4lETKfT8TF74fMwGo0sODiYf2/COLOQ16HVauuNzwrvU6FQsIiICKbT6di9997L1Go1M5vN9XKItFotzw2pO7YqjIFrNBqeS6BWq9mAAQOYSCTi26RGo+H5UABYUlISM5lMTC6X8+9SGK8X/hU+R+H9CmPjwrg56uQnmUwm5ufnx/R6fb08iLi4OBYYGMhMJhPr27cvHzPXarUsNDSUqVQq1r17d+bj48O0Wi3r1asXX5fZbGYymYzpdDqm1Wr5Ov38/JhUKuXbmVar5ePc/v7+DAAzGAwsLCyM59wIY+NC/oJYLGbdu3dnXl5eLCwsjN19993M29ubryMqKornQgifEerklwjj2tdddx3/HN98800ml8sZY4zJ5XIWHBzMunfvzvbs2cPzl+69915mMpmYwWBgcXFxPFfKYDAwlUrF1Go1k8lkLDIykq9PyHUR9gEGg4E9+uijPLdk9uzZ7Nprr2Xh4eH8uQqFgg0ePJiZTCae86XVallwcDDz9/dnERER/Pcq/NaF9yqRSFhAQADz8vLiuWhCbpbBYKiXHyZsj0FBQSw+Pp4vGxkZySIiIngOxK233sr++c9/1svpMpvNTK/XM6PRyKRSKevZsyf/noXPeMyYMXy7Wr16NevVqxd/TMjdqLvtzpgxgyUkJDBfX18mk8nYNddcw3PGhPyO4OBgFh4ezt+bwWBg4eHhPMfIYDDw7U6hUPD14H85GkI+m8lkYikpKUylUjFfX1/Ws2dPZjKZWFJSEouJieGv3atXL6ZSqZjFYmE+Pj7MYrEws9nMdDodz/+755572NChQ3neUlBQENNoNMzf35+3Qa/X81wKYZsXjg3CcSg1NZX/LiQSCbNYLPx45ePjU2+fHRsbyzQaDX+toKAgvl0DYMnJyUyv17PZs2ezadOmMaVSyYKDg3mujpBT4uvry6RSKd8XGQwGFhQUxLfVadOmMb1ez7y8vPg+UMhjMZlM/Njn5+fHADBfX1+mUCj4NivkF8lkMjZs2DAWHR3NfH19mV6vZ3/729+Y0Whk48ePZ1qtlgUFBfH8znHjxvGcm1mzZjXr2O/ROSbCOFh1dTXMZjMqKiqgUqmg1WqhVCrRt29feHl5ISgoiM/OMJlMPJua/S+D/tChQ3A6nQgMDIRCoYBWq4W3tzfuv/9+VFZWIi4uDjqdDiaTiUfIx48fh9PphNvt5uO7LpcLsbGxMJlMiIyMRG1tLU6dOoWffvoJGRkZyM7OxsmTJ5Geno7MzEwcPXqUn0EL0TkA1NTUQKFQ8FkTwviv0AUWExMDqVSKm2++mc/qCQgIgMlkQlxcHKRSKWbNmoXo6GgkJSXB19cXlZWVEIvPf51Op5NnszudTt7tfujQIZSVlcFms6GoqAjV1dWoqqpCSEgIGGOoqqpCQUEBvLy80L17dwQFBUGv16OyspLP+jCbzVi3bh3cbjfOnj2LHTt2oKqqCvn5+aipqeEzT+RyOcrKyuB0OlFbW4uqqips27YNa9euhcvlgtvtRklJCWprayGTyXhOjDAjpqysDDU1NcjJyUF1dTXOnTuHiooKnl9jMBh4tr5SqURycjIqKiqQl5fHZyf4+/vzmR4A+GciRPJSqRRmsxnA+eEui8WCjz/+GA6HA4WFhXzclTGGiIgI+Pn5YeDAgRg0aBDCwsIAgJ8hWq1WSCQSnlW/c+dOAMCQIUPQvXt3TJo0CVKpFGPHjoVIJMKBAwdQVFTEZ8EYDAao1Wr06tULQUFB8Pb2xuTJk/kUZ+H3YLVaeZ6MsE2q1WoAQGFhIaqrq/nZ6pkzZ1BRUcFnmh07dgw2m4339lRVVWHKlCkIDw/n31N1dTWCg4MRFhaG8vJyaDQano8knN0LvSsSiYR3SQtnpsLsA5lMxqf8ut1ujBo1CgkJCUhMTERoaCgUCgVOnDiB1NRUxMbG4uuvv4bRaOQziKqqqqBQKCCRSDBo0CBER0ejR48emDdvHqKjo2GxWFBdXc2HDh0OB/7v//4PDocDUqkUDocDxcXFOHz4MKZNm4bi4mI+fFZdXY2amhoMGjQI1113HYKDg8EY471xKpUKubm5AABfX1+88847iIqK4r2ywhmx2+3GiRMn8N///pfnPkgkEpw6dQq1tbU4ceIEysrK+Aw4YdYZYwyxsbFQKpXQarWIi4vDvffeixkzZuDmm2/m+XDCvsPHxwcKhQL33HMP1Go1z91ITEyEwWCAXC5HSUkJz4USeqGEM3Sh5/Phhx9GbGwsvLy8+NTUmpoa2Gw2dOvWjc8qUSqVMJlMUKvVKCgogEgkgo+PD+644w4YDAaeZ+Pt7Q0vLy9IJBJoNBqIxWJUVlYiIyOD560dPHgQIpGI1/gQem7sdjsYY3zGR3Z2NrRaLR+Gkclk0Gg0MBqNcDgc0Gg06NWrFx/6k8vlfMaa8B727t2L8vJyZGRkIDMzk8/uOnz4MFQqFe/Vk0gkPJfNarXyfc2ZM2f4fePHj4dEIkFeXh4YY3A4HHC5XNDr9VCpVAgICMC4ceOg0Wj4bCphSNBgMAA433suzBYU1imXy5GcnAy32w2z2cxzQfR6PUpLS6FSqeDv74/p06cjJiYGMpkMf/75JzIyMhASEoKwsDD4+vpCIpHAy8sL3t7evG3CvrC2tpb3BLlcLhw6dIjPJrVardBoNDh37hxkMhlsNhtEIhG8vb0BABKJBKWlpTAajZBIJLjjjjt4nqPT6URaWhqOHTuGkpISREZG4uDBg5BKpdiwYQNqa2ths9l4z1NwcDCio6ORlZWF77//vlnHfo8eyklJSUFycjJ++uknlJWV8QOw8GP38vICAPztb3/DuXPnMGvWLOzcuRMBAQFYuHAh5s2bh5dffhkPPfQQpk6dinnz5uHDDz+EXq9HRkYGnxolTD8TEtPKy8vh7e0Nk8kEjUaD4uJiHDx4kE8nlEgk6N+/P37//Xfo9XpMnToVUqkU3t7ekMvl+P3337FlyxZIJBLMnj0bCxcuxLlz53iybXBwMOLj47F161aUlpaiT58+yM/PR3JyMm+fRqPBli1b0L9/fyQnJ+P333+HWCzmc8NvueUW7N69G8899xwAYPHixTh37hzsdjvuuece3HzzzSgvL0dgYCDy8/P5vPdbb70VX3/9NXx8fHDy5Ek+/TMgIAB79+4FYwx+fn5wuVxQqVS49tpreZdeVFQUqqqqkJubC7lcjgEDBuDw4cN8qlhqaiq6d++O+fPnw2Kx8ORHYejN19cXI0eOREZGBoqLi9GnTx/U1tbi4MGDyMjIQFxcHMLDw7F582Zs27YNFosF/fv3x9dff80PDFVVVThz5gxPPBOSsoKCghAaGorff/8dYWFhSE5Oxm233Yb58+ejoKAACQkJSE9PR3BwMLZs2cKnjQo/pOrqaigUCqjVagwcOBBlZWU4cuQID5KEhFipVIojR45g2LBh2LJlC4qLi3myW0JCAs/1EHbW8+fPR3R0NI4fP468vDz07dsXoaGhkEqlOHfuHE+clEqlAAAvLy9UVFRAKpUiLCwMJSUlSE1NxQcffIDvv/8e69evx6BBg7B27Vqe33Trrbdi//79PFmypqYGpaWlUKvVSExMxM8//wylUsmD+XXr1uGBBx7AkiVLUF1dDafTyaeL63Q69O7dG08++ST+/ve/IzMzEzExMdi3bx9UKhXuvfderFq1CnFxcVi2bBmeffZZzJs3jyfI1p3CWl1dzcfIzWYzjh8/Dj8/Pxw/fhwajQYlJSV8Z1pYWAgAfMjIarViz549eOmll3DvvfdixYoVePPNN5GdnQ2bzcan8Qr1UR599FEUFhZi69atuP3227FlyxYcOHCAB+zC6/fq1Qs7d+6EWCyGl5cXxGIxP/gqFAqYTCY4nU4sXboUa9euRWpqKvR6Pd566y0+HJOZmQnGGCorK6FQKCCTyRAVFYWTJ08iJiYGwPmANSYmBrm5uSguLsaRI0cgl8v5dG5hSi0AJCcnQyqVYvDgwTh27Bj27t2LM2fO8ANXWVkZHy6tqKiA2WxGVVUV/Pz8YDAY+LTNHj16oKSkBPn5+Xy4qbi4GE6nE35+fsjJyYGPjw9KSkqgUqkQHBzMkzb79++PjIwM9OrVC+vXr4fdbkd5eTnUajUvR1BdXY2AgACkpaVBp9PhwQcfxOTJk/H0009j//79yMjI4AdvYShXmJKfm5uLo0eP8gkACoUC586dQ0REBPbs2cOnB4eEhCAnJ4dPFggLC0N0dDReeeUVREdH48UXX0R6ejq2bduGIUOGwGw2Y/HixRgyZAiqq6vh6+uLnTt38ryQPXv2wOl0oqqqCqNGjcLZs2exf/9+PrQpDINkZGRALpcjLCwMLpcLJpOJl1LYu3cvqqqq0K9fP/j6+iIvLw9+fn6IiIjAl19+CbvdjqCgIOzcuRNPPPEE3n77bfj5+aG0tBRhYWHYv38/wsPDERMTA39/f2zcuBE33HADPvzwQyiVSp5cC5w/eUhJScEdd9yBL7/8Ek6nE1lZWejevTvS0tJ4m0+dOgWr1QqbzYaQkBCkpaVh6NChOHr0KJ566im8+uqrkEgkPIVBGEK02+3o06cPcnNzMXDgQKxcuRIKhQIzZ86EVqvF3r17eYmGH374AWFhYTCbzdixYwfsdjukUinf51ZWVvJp1CKRCLGxsQgICMC2bduQnJwMm82GW2+9Fddddx2Av9IbmsKjA5PMzEyUlZXhnXfewblz51BZWYnU1FQUFxdj9+7dyM/PR15eHnx9feFwODBlyhQsXrwYRqMRNpsNZWVl8PHxQX5+PhQKBVQqFQIDA3Hs2DHodDrMmTOHj/Nu3LgRBw8ehFgsxn333YcxY8bgiy++wJAhQ3Ddddfhjz/+wMKFC7Fjxw6+0R44cAApKSnQ6/WYNm0aJkyYgOTkZFRWVsLb2xtGoxErV64EAPzyyy8QiUS4/vrr8csvv+DkyZP48ssvkZubC5vNhgkTJmDOnDlYvnw5br31Vvz5559YsmQJZDIZzpw5A6vVirfffhv//ve/sXfvXuTk5MDhcCAgIAA9evTgBz6bzYYnnngC8+fPR2hoKI4cOcLPPoOCglBeXo7y8nL4+/vD398fSUlJsNls+OyzzxAZGQmFQoFhw4Zh3bp1OHbsGGpqahAaGorS0lJUVFSgtrYWfn5+/MyhpKQE06ZNw+rVq7Fjxw6+04yPj0dWVhbCwsJw5swZ3HLLLThx4gQfexw3bhwqKipQUVGB1NRUXvNi/vz5mDx5MpYuXcrn/ZeUlKCgoIDXVBk3bhyio6Nxww034Msvv8T777/Ps/sHDx6M7777DjNmzMCKFStw6tQpZGRkIDAwkPcWBQQEYNiwYdi3bx8vxCfMz4+KisKRI0fw/fffQ6fTYfLkyUhLS4Pb7UZMTAySkpJw7Ngx/lk/8sgj2LhxIw4fPoyqqip069YN6enpuOOOO/hY7g033IB77rkHbrcbZWVl0Gq1PGPe4XAgLy8PiYmJ2LBhA59hJpxJAcDYsWPx/PPPw+Fw8F4KhUIBu90Of39/BAcHY/fu3bDZbLjlllvw7bff8sqO//d//4f33nsPvXr1QlFREU6cOAGbzQaj0Yji4mL4+fnB398fFRUVmDNnDj788EN88cUXOHnyJD755BOcO3cOarUap0+f5j1in3/+OR577DE+S+nTTz9FYWEhPvroIzzwwAMICgrC+++/j6KiIsTHx2Pfvn3Yvn07LBYLDh06hPDwcNx88834+uuvee2FO++8E+np6cjNzYVCoUBERAQKCwuRnp4OvV6PiIgIbN26FTKZDEajEQqFgteUEOr/mEwmHmgJZ5EWiwVBQUG49tprcfz4cdTW1uLpp5/GiRMncOedd4IxBh8fHwwcOBDp6ek4ffo0r73DGINYLOb1V66//npeeVQI5kUiEXbs2IEnn3wSDz74IG644QacO3cOR48eRXV1NX799VdERUXhl19+QXR0NLZt24aePXvi119/RUFBARhj2LdvH/r06QOxWIz169cjODgY27dvR3BwMIqKirBkyRLk5OTg6aefhslkQm5uLgwGAyoqKvDaa69h6dKlEIlEUKvVGDNmDJKSkvDkk09i165dqKqqgsVigVwu5zV0NBoNcnNzeXAUHh6O7Oxs/N///R9ef/11GAwGmEwmfgD+4YcfMHfuXIhEIuTl5WHnzp1ITU3Fjh07cPvtt2PPnj04deoUevXqhf79+2Pnzp2ora3lJw5CfZScnBzI5XKUl5fDZDIhJiYGaWlp/PsSeniFHCWhd6m2tpb3Fubn5yMwMBApKSl8X6ZQKODv749z585BLpejb9++mDt3LqRSKTZv3ow777wTQUFBKCwshEwmQ1FREa/ZJNS38vLy4rkxwqzNEydOQCKR8Nkz5eXlvN6MxWKB1WrF4cOHERkZCaPRiL1798JkMqGqqgplZWVQKpU8Zw8ABgwYgIyMDH7i4efnx3sbP/roIzz00EPYt28fkpKSEBcXhzNnziAwMBB//vknHnjgAfz973+HWq3GlClT8Pnnn+ONN97Af//7X2zfvp2ftJpMJhQVFfFZf6GhoVi1ahUGDBiAoKAg/PHHH7j33nvxxBNPICAgAIWFhdDpdHx7cDqd/Hgq5OMIvccKhYKfXO/btw8PPPAAFi5ciBEjRuCee+6B2WzGkSNHEBISgmeffRanT5+GSqVCTEwMXnvtNSQlJTX52O/RgQkA3HvvvfDx8cH777/PuwOFbkJvb29eIVOIyHNzc/nORCQSYfny5XjooYdQW1uLgQMHorq6GllZWbz7T5hyarFY+FmmMJ0zIyODF/0RumLz8/P5j00YihGJRJg8eTLeeecdpKSkAAAfttm7dy+Av5JjhUqd69atg1QqxenTp/nyYrEYffr0werVqzF79mxs2LAB2dnZvIvZy8sLNpuNT60Ui8WQyWS8+10ovDR48GB89dVXCAoKQmZmJiZPnox169Zh5cqVSE1NRWVlJT9zU6lU/EAlTF0TCpAJU49FIhFKS0sxduxYrFixApGRkZg2bRoef/xx6PV6fsYtTEPbuHEjIiIikJGRgZKSEp6dLxQ5i4iIwKhRo/Ddd99h06ZNSE1Nhb+/P7Kzs/n3KxRvEolE0Gg0OHPmDFwuF3x8fFBbW4uAgACcOnUKMpkM06dPx7Fjx1BVVcUPYKmpqRCJRDh+/DjS0tJgMBjQt29fbNu2jRdcKy8vh6+vL4xGI3x8fPD7779Dp9MhJCQEe/fuhY+PDz9zFBLkhODu3LlzkEqlfMhQeD3hbFj4XoSuVX9/f7jdboSEhGD79u18WERIdM3KyuIHDoVCgYCAAJ4garPZMHDgQF60S6VSoV+/ftixYwdSU1Oxd+9eLFmyBImJiUhNTcXx48fhdrsxevRobN26Fddddx1OnTqFnJwcaLVaHD16FEqlEjqdjhdtE3a+drudJzwfOnQI2dnZkMvl/GxJIpFgypQp+OSTT/Dbb79hxIgR6N69O/bu3QuFQoHg4GCcOnWKT3EVpsUK1XiFkwMhg19IpDt8+DAUCgUvYqZUKnkRtDFjxuD7779HQkICjhw5An9/f5SUlPAhq1WrVmHIkCGIi4vDoUOH4Ovry4uRAeDtCAwM5EO1wjCVkKAZExODY8eO8ecIieTCMGqvXr1w5MgRREREYNu2bTygqnsws1gsOHXqFJ8yK/Q6arVaHDt2DED9Eu+i/xXZ+tvf/gaTyYTMzEysX78eXl5eOHv2LEaOHInffvut3hBkcXEx4uPj4XA4+GU2hKGNqKgomM1m5ObmIiIiAmvWrOFTlVNSUlBbW4vTp0+juroaPXv2xIcffoiePXti5MiRCAwMRGRkJJ555hk+XT8jIwPdunXD6dOnea+iVqtFYWEh9Ho9ZDIZCgsL+VC1yWTiw63CkKJWq0V+fj6sVivkcjlOnz6Na665BvHx8fj444+h0WgQHByMvLw8WCwWPvwo/Gbsdjtuu+02LF26lE87l0gkKCws5CUjhB60yspKqNVqOJ1OLFiwAHPmzOEBjslkQnFxMUJCQqBQKJCbm8uLobndbhQXF9cbVv38888xZ84c7Nq1C1arFSqVChMnTuRBv0wm4zM3zWYzDh8+DK1Wy6sGp6en86KDwjR8IVGUMYbx48fj1KlT2Lx5Mx9y8vHxQVlZGeLi4nDkyBEkJSUhIyMDRUVFvAdM2DaFYEFIERB6gLy9vWGz2WC1WnlVWWEKv5C8LeyTcnJycPPNN2P//v2ora1FWFgYjh49yourCcc9oWdQKCwoDK1JJBIolUrk5+fzk0ahnUajEd7e3khOTubF9z7++OOmH/iblZHSzt544w0WExPDEhMTebEsjUbDjEYjT/xSKBTs2muvZb1792ZarZZpNBqeBGSxWJhOp+OFX/R6PfP29mYymYzdc889LDAwkEmlUpacnMySkpJYnz59mLe3N4uLi2Pz589nUqmUmUwmplAomE6n40mQ8fHxbNKkSbzIlJCoJiRTKZVKnnyrUChYYGAgCw0N5X+3334769atG4uNjWVRUVFs+vTpTKPRsBkzZrDTp0+z06dPs0mTJvG2y2QyFhwczCQSCYuKimJGo5Ent1ksFqZQKFhMTAxLSEhgOp2O9e/fnydtCkmJJpOJTZkyhY0ePZpFRkayuLg43n6LxcLuu+8+Nn36dJ5cdvLkSaZUKln37t3ZsGHDmEajYddddx1PtJTL5Uyj0bDnnnuO6fV6ZrVamY+PD/P29uZJakqlkqnVap5UaDAYmI+PDwsNDWWDBw9m4eHh7MUXX2QWi4X17t2bmc1m5uvry/z8/JhOp+MFtfC/5MABAwYwg8HAZsyYwRNae/bsyZKTk9mcOXOYj48PM5vNzGQyMZPJxAtgqVQqptFomEgkYgaDgSecyWQy5u3tzQICAnhRveTkZNazZ0+mVCp5YqyQiKrVaplOp2Pdu3fnyY0SiYSFh4fz71OhUDCpVMomTZrE1Go18/f3Z2q1mvXr149FR0ezxMREplAo2KxZs5jBYGCBgYE84REAi4iI4AWihAJPBoOBGY1G5u/vzxMVVSoVf56QQOzj48O8vLx4MTGhAJqQ+N2jRw82YcIEJhaLWUxMDJs0aRLz9/dnUVFRzGQyMb1ez8LCwnjxuwEDBvD3EBsbywsZBgQEMKvVyj788ENe4MpgMDC5XM4/X5VKxVJTU3kCp5CEKXx+er2eabVads0117A777yTb9NhYWHMz8+P9e7dm4WFhTGFQsHi4+OZSqVif/zxB0+01Gq1bN68eTyJHAAvHigkF5rNZhYVFcVUKhWzWq08oVf4jB9++GE2ePBgvq8AwLp3784Lil1YBExINBbec3BwME8KHDhwIDty5Ai7++67WXR0NE8Q7d27NzMYDGzixInMx8eHb5dCknRISAgLCQlhQUFBLCQkhIWGhjKj0cjCw8OZj48P8/X15e3XaDRsxIgR7Prrr2dWq5XJZDK2bds25u3tzbRaLUtKSuJJmELCrbAPsVgsTK/Xs169ejG1Ws3Cw8N5wcJ+/foxo9HIfv/9dxYdHc1/31qtlieMzpw5k4WGhrLk5GTenjlz5vD1qNVqds0117Bhw4axkJAQZjAYeBKt8LvT6/UsIiKChYWFsalTp/KkVqVSyZMxhdcTJjQEBQWxadOmMZVKxUJDQ3nRTG9vb2Y2m1n37t2ZWCzmyctGo5GZzWbm5eXF9u7dy6ZOncp69+7Nv0OdTsf0ej3/3iIiIvjn4OXlxXx8fFhsbCyzWq3Mz8+P9e3blyd0Dx48mPXq1YtpNBoWFRXFlEol69evH5NKpczb25sZjUYWGxvLCyoK+08hEVvYlgICAlhsbCxbunQpT1QWPi/hveN/hUAtFgufpKBSqXjSsFwuZ35+fuyuu+5iwcHBPDm/d+/eTKFQMK1WyyIjI3kCsVgsZiNHjmRms5mNGDGCaTQadvvtt/NEV6lUym6++WaeTC1M0hC29zlz5jAAzN/fv14BvNDQUKbT6VhYWBgbMmQICwsLY6NHj2ZyuZzdcMMNjDHG5s6dy+bOndusY79HByaTJk1icXFx7LrrrmM6nY7FxsbyHbHwRQLgs22sVivPxBaJRMxisbDExEQWHx/PAxC9Xs8kEgkLCQnhWcpCRUBhxodWq2UJCQl8Bkfdg7xWq2VGo5ENGTKE9erVi3l7e/OKtMIPy9vbmx9YNRoNP4jWrRZbt1qf8GMTqkgKGfrCcsKOR3hfwqyWgIAAptVq+XsQZpIIByyJRMKr4IaEhDA/Pz/mdDpZSkoKn80iFot5Jr9arWaDBw9mRqOR7yBiYmJYt27dWFBQEBs/fjwLCgpi119/Pd9pC7NGhIPpiBEjmFgsZoGBgcxqtfKDizCTQ5hV5eXlxatfKpVK5uXlxQ/AwswR4XMQPjdhWeHgLhx0hYBReD1hZkXdnaYwu0KoiCisU9gJCN+XxWKpVzVUCEjCw8NZZGQkCwkJYcnJySw1NZUfgO6//34WGhrK+vTpwwICAljv3r35dy8EEr6+vnz7EQ6iYrGYWa1WvqMRiUSsZ8+e/H2qVCqm1+uZXq9no0ePZkuXLmU+Pj58x2q1WplCoWBJSUlMpVIxs9nMgoKCWEBAABs3bhwLCwtjRqORr0eocikEjsJ3ZrFY+ONCdVEhCFGpVPyzE347APgMLSH4F4vFbOzYsfwAL1TQFJYXqn1aLBY2adIkfiAVvithRo/wnQo7zLozQITvymQyMQDM29ubVzQNDg5mL774Ig/OhWWE2WnC8/r168duvvlmJpVK+YmJ8BtUKBTMbDazXr16MYvFwtRqNYuOjmbXX389S0xM5FWJQ0JC+PuWSCS8crRQHVmo/izMxhI+O71ez77++mv+2wfAwsPD+TYpVAG2WCxs2bJlTKfTsZdffplJpVJeBTk0NJRv81KplIWGhvLHheBDJBKxyMhI/n0KszyEmUVC20wmE9Nqtax3795Mo9EwPz8//puue1IjkUhYWFhYvdfq1asXPxEUTkqEGSp6vZ5XvhbaWXedwnciBLTCZ4j/zeaSSCT8REGn07ERI0awRx55hL+GMBNMoVCwPn36MC8vLzZ48GA+29BsNjOlUskGDhzIgyuhnVFRUXx/HB0dzSZNmsRPxvz9/dnbb7/N950ymYzvt4TfkNls5hVdhQBP2McKB27h9xAaGsq8vLyY1Wrl321ycjIPcIVZXgEBAXwbEYvFLCwsjM/ACQkJYampqWzOnDk8eBo2bBgLDQ3lVcX79OnDZ+gplUo2ZMgQFhQUxLp168ZnOwr7IOH3Jfz2hYrKPXr0YFFRUXy/7O/vzwwGA5swYQIzmUz8pE0IVCIiIvjxZ9CgQczLy4vddNNNbPbs2WzYsGH8N3nHHXewmJgYds011zTr2O/xQzmPPfYYvvzyS5w7d44Pm+j1eqjVajz66KN4/PHH+Xz2jz76CO+++y769euH999/H8HBwbzLXSwW49y5c3j99dfx0UcfobCwEH5+fvDx8UFaWhrGjRuHsrIySKVSfPvtt7zLShgLdLvdKCgoQHFxMc+Uz8nJ4YlFI0aMgNVqRUlJCc6cOQOJRMJnklx77bUICAjgc/kBwG63IzMzk9e0yM/Ph0qlgslkgtlsRnR0NC+ytHnzZnzwwQf8GjQTJkzgBZ2ErlStVovIyEisWrUKycnJSE9Ph8FgQF5eHoqKigCcr2IodBnr9XqkpqZiw4YNfPaRXq9HVVUV75oVrndgt9vhdrv5DAuhaq2QwW80GrFz507+//LycvTt2xe33XYbfvrpJ7jdbuzevRsDBgxA3759cfLkSb5eHx8fBAUF4ZprrsGKFSt4LQi73Y4XX3wR27Ztw/r163n9BYfDwSs2ZmdnIyQkBDabDd7e3jh9+jRef/11PProo0hKSkJmZib69OmDffv2Qa/Xw9vbGzqdDj/99BPPgSkoKOD1NIRuSLVajfHjx+Pzzz+Hl5dXvesSaTQa6PV6JCUloVevXli6dCkYY/xaSSdPnoRer4fBYEBMTAy0Wi0eeOAB7Nq1C7/88gtKS0t5vpOPjw80Gg1OnjyJJ598Ek8++SRPPhNyf4qKipCfn89n1yiVSpjNZuj1et4dnZ+fj0ceeQQvvPACZs2ahX/96188mVD4Tp988kleDO7zzz9HfHw8Nm7ciH79+vHkw2uvvRYrVqxAUlISr2YbExPD80t8fX153oxQ/VcYmmKMYcCAAejZsye/qCFjDM899xy+++47HDp0CEOGDOF1VXJzc5GZmcmrSdbW1vKZDSEhIbwKcFRUFM8lEZJmq6urER0dzetjCPlPQs2Rv/3tb9i1axc2bdqE8vJy1NTUIDw8HIcPH+YzKIR/fXx8+NCtkKcl5Gt4e3vDarXCbrcjIyMDMpkMVVVVPM8gKSkJ5eXlKCkpQUJCAgoKChAVFQU/Pz98++23vKqrkFQsDCvU1NQgJCQEMTExMJlMcLvdMBgM2LFjB7+WlFARuaqqis+KEIaOhOEAh8MBi8XC92GTJk3C6NGjMX/+fJw+fRrjxo2Dv78/tm/fjsjISPj4+MDLywvLli3jQ17CDIri4mI888wz2LZtG0JDQ7F582YkJSXh6NGj2LdvH8aMGcOvY3TgwAFotVpUVFTw6z5JpVKUlJTw2lHCrBqTyYSSkhLo9Xrk5+ejuroa3bp1Q01NDYqLi2EwGHhROuH6R8LwqDA8VFtbyxM3dTod3n//fdx2222YPHkyPvjgAz5MGBUVhTNnzkCn0+Hw4cNwuVx8WFCYmadQKPh1rmJiYhAZGYn169fzujYWiwU33XQT0tPTceTIEV6EDjhfuDA8PJznu9hsNmRkZCAvLw86nQ52ux1Dhw5Ffn4+jh8/jrqHVoVCgcjISPz666+YOHEi/P398dtvvyEwMBDHjx9HYmIiXnzxRYwbN47PptJqtXA4HCgpKUH37t2xf/9+WK1W+Pr6wuVy8domwpCLn58f/12eOXOGz5ITcgUjIiL40OPZs2dx33334dSpU9iyZQufHVV3JpQwSy8sLAy9evXC9u3bYbfbeW5bVlYWQkNDsWfPHnh5efFkbuF6Z/369QMA3HDDDUhKSsKQIUOafNz36MDktddew/r16zFkyBB89dVXMJvNcDqd2L59O6qrqzF48GBs27YNVqsVZ86cgVKp5IV5hIvbCWOjQsKgw+Hg46bFxcV86tu1116L9PR0LF26FGPGjIFMJkNycjI2bNiAiRMnYs2aNdDpdDh16hRMJhOfPmU2m3H69GlexlsoCnTu3DnU1tYiJiYGJSUlMBgMOHLkCM9V8fPzQ1xcHMrKyrB27Vqe1FhWVsY3TH9/f8hkMkycOBGvv/46UlNTsX79etxxxx3QaDRIS0tDTk4OcnNzoVar+ZS4lJQUZGdnw+VyYdKkSViyZAm/4JmQWyLko0gkEkRERKCqqgqxsbFYunQprrvuOvTs2RNbtmzhF98SpuGWlJRAJpPxgj3CTmjSpEn49ttv4ePjg4yMDCQnJ+PYsWMIDg7GiRMnUFNTw8fihcJQZWVl0Ol0GDFiBLZs2YL4+Hikp6fDbrcjNDQUEokEaWlpMBqNvGiV+H8XmwsPD0evXr3w4Ycfonv37jh37hwfJwbAd0rCNPPIyEgUFxcjPz8fVVVVqK6uRmpqKg4dOoSAgABs2bKFT+EsKytDUVERH8MWgjLhSrAAIJfLkZSUhD179vCdu1AOXiKR8KnGwvh0aWkpqqurYTKZ4Ovri759+2LVqlXQ6/V8vFjIUzl9+jRmzZqF//znP4iKiuLfb1FREcaPH4/8/Hzs2rWLXzXVZrPBbDbzxNrS0lIemEVFRSEtLY3nSgHnC56x/+WW9O3bF4MHD8bixYsRGxuLAwcOIDAwkJcUFy7AKLRLyIk5fPgwevTogT179vCKjhEREfD398fmzZv5eLO/v3+9XJ28vDz+/dvtdlRXVyMyMpInWgqJzsOGDcPGjRvhcrmQnJyM/fv3IyIiAkePHoXZbObfj0QiQUJCAg4ePIhRo0Zh3bp1GDNmDCorK7F582Z+UUaJRILc3Fx89913GDNmDMxmM0+AZIxh2LBhOHz4MJKSkrB+/Xr+G3Q4HPy3HhwcjIqKCpSUlCA2NhZHjx6FTCZDRUUFLBYLzp07x3OyiouL+cX7evXqhYyMDD51VyjvLmwncrkckZGRGDhwII4fP461a9fC29ubV9yU/q/s+IwZMzB//nw+S0atVqOkpASlpaX8Ymt33XUXjh49ihUrVsBgMCAwMBBRUVE4ffo0jh49CrVaDY1Gw3NhhNwAoR1CmXyTyQTGGLp164Z169bxKcMqlQoVFRWorKzkFxhVqVTo0aMH0tLSIJfLeZEyl8uFv//97/j444/5xea8vLxgsVhw4sQJXvBv0KBB2L9/P/Lz83nOkTCdWCg9UFVVhRtuuAErV67EM888g/nz52P06NHYs2cPunXrhqioKCxatAhGoxFutxt+fn44cOAAZs2ahaVLl/IimhqNhp/IlJeX81w9YdtNT0/nM8mEqr47duzAyZMnkZiYiMOHD/M8L6GKrN1u5/unuLg4foXmgwcPIjw8HG63G1lZWSgsLLzo9xoQEIDq6mp+AT2Xy4WdO3fivvvuw4EDB3gSt7DdCIEWAMyfPx/PPvssv3htTU0NT+IVLngpXL5FKGAo5MX5+fnxwobCiejo0aOxfPlyxMXFwWw2Y9OmTXzaupBrJZfLkZOTw4OQqKgong924sQJzJ49G99++y2mT5/Op00LQeGDDz7Y5GO/R9cxOXToEPr06YN33nkHJ0+exKZNm7BhwwaedLpr1y7ExcXh/fffx5AhQ9C/f3/06dMHkZGRSE5Ohlqt5jU5EhISMHfuXCQkJMDlciE9PR2hoaEYPHgw/Pz8kJmZiZycHNx0000oKipCWVkZ/vjjD6hUKqxcuRLdunXDiRMnIBaLeeBRXFzMp+iKxWJoNBpoNBo+R93Pz49nZzudTnTr1g0SiYT3wBw7dgxr167FXXfdxRM6JRIJgoKCeCJWbW0tfvvtN/j7+/MpqCtWrMDGjRv5rBZhCrPFYuHz/U0mEwwGA/r378+T0CIiIuB0OqHRaDBhwgTExcUhICAAp0+fRmZmJg4dOoT4+Pj/b++8A6qs9z/+Qjgs2SJ7CoIylOnAgXtiiokmJkTuzFFJaqV5naVlpqkp5kxb5iYVGyp63QgIiooc9pK99/n94X2+F0zN+nWvdjuvvyzgnOc853m+z+f7Ge837u7ubN++nfj4ePr37y9GW62trYUkdF1dHeXl5QwaNAgTExMSEhJobGwUmizt27ensbFRaKRI2ivm5uY4ODhgaGiIo6NjCzdZTU1NIUteUFBAWloa7u7u1NfXU15eLs55UVGRmKeXtBskbRg1NTWcnJyEzkF5eTnt27enpKSEvLw8VFVVMTExQUdHh4sXL1JRUUFSUhLt2rUjKytLTAxIQY2kXmpubk5ZWRn29vZMnDhRKLq+8MILqKmp4e3tjaGhobAJt7Kyorq6Wii+SouwnZ0deXl5DBgwQDwss7OzSU5O5tKlSxgZGdGhQwfOnTsnGoqbN1qq/Mv3RlVVFUNDQwYMGICenh6mpqY4OjpSWlqKQqFALpfzyiuvUFFRQWVlpXBclXZbampq1NTUEBsbyyeffEJ5eTnXrl2jpqaG4uJiamtrha7F3bt3xW5VW1tbjNHfuHGDYcOGYWxsTJ8+fUhMTOTatWt4enrSrl074Z1hYGBAhw4dKCsrw9nZmbZt21JbW4u/v79Qk5XG1qUMiuS6q62tLXxLbt++LTKX0veuqqoqbCPOnDkjND4k3Ye6ujpycnIoKCigc+fOHDhwQDQj9u3bV9xnkZGRVFZWiuzQggULqKmpEVlDqXFz3Lhx6Onp0adPH2xtbZk0aRIymYxevXoRGBgoHGW7du1KU1MTPj4+JCQk4OXlRUBAAOPGjSMgIICQkBBGjx4tPLays7M5cOAAd+7cEUFRVlYWlpaW4hiWLl1KfX09Ojo6ZGRkkJiYSEVFhVi/SktLiY+P58yZM2hqajJq1Cju3bvHsWPHMDExEb5ReXl52NjYoKGhQbdu3VBXV2fMmDEYGhpib2/Pnj170NTUpK6uDrlczsqVK+nbty9t27bFzMyMIUOGoK2tLRR6pUk0aapD8nsaO3YsvXv3pqGhgbKyMjEuHR8fT1hYmMiYZmZm4ufnJ/xWPD09GThwIIDwg7G0tOTSpUvo6+uzatUqGhoaOHjwIIWFhZw4cYJvvvkGIyMj3NzchE6N5N1TW1uLubk5rVq1Ijg4WOh3SEGQgYEBBQUFxMTEUFNTQ2ZmJhYWFkIjRdq0mZmZoaurK1yatbS08PPzw8HBgZqaGtq1a4e9vT15eXl06NBBTE6mpqZSWlqKubk5zs7OItCXfNyGDx+Om5sbQUFB6Ojo0L9/f+Li4rCzs6OhoQEXFxcMDAzEePSOHTuoqqri7bffJi8vj8bGRuEZJjUBA2ItadOmjZje09HRwcDAgOLiYsrKyqiurkYmk6Grq8v169eFFlNKSgrq6uqoq6sLmYCXXnoJAwMD6urq8PT0RF9fn5s3b3LlyhUx6fTpp5+SlZVFXFwcp0+fJiIiglu3bnHr1q3f9ex/rjMmI0eOZObMmQQGBmJpaUnXrl3FtIkUtbVq1Yp3332XGzduEB0dTVFRkZgEkDrgJXEwAwMD0ZHs6OiIXC4Xvi6tWrUSi6g0CSGlUktLS4W+gYqKisiCSLtCaRclpR8fRtqpShMAEtIOxdfXl/j4eJH6fRhJIEvaOdXU1KCmpoa+vr4o00gZAgnpprOzsyM5ORlNTU3at2+PTCbj2rVrjBgxgtOnT6OtrU3Hjh25cOGCSMENHTpULPKS54TiX2N7kk6BtNuT0sE6OjpkZWWRmZlJQ0ODCBakz2xgYECvXr1IS0sTHeFNTU1iCqP5uTI0NKS8vByAPn36cPbsWVxdXcnOzqZHjx7I5XJycnLEaKikaSIJ60lI6WbpdaXJmtraWkaOHMnp06epqakRn7tNmzZC4Oju3bts27aNpUuXkpiYCCCmjx7+niWRpYqKCvT19WloaBBW59JC3aNHD06dOkVTU5MwtJOCFXgg5CUFfyoqKhQWFoqFW09Pj+zsbHGckveRVGKTuvKlRVj67l1cXADEFJokI25lZUVVVRUFBQXCNE/1X5b1kmGlFLg0NTWhUCiwtLSkqqqKsrIyTExMxHSKlpaW+J6l+0FfX1+k9hsbG8VUj3QdSfeAi4sLd+/eFe/Vrl074UVjamoq7mVLS0tat25NbGyssF+QyqgymYycnBzatm0rJnW0tbWprq4WI4+amprcv38ff39/Lly4gKqqKp06dRL3rFwuZ+jQoRw9elRsemQyGdXV1RgZGYmFvrGxETMzMyoqKmjXrh3Jyclikkq6ZqWNiRRcSSlvFRUVdHV1KSwsREtLi6KiIrEuaGtro6Kigr6+Pnl5eUIL5d69e9jZ2eHv78/FixfJzMwUQaokwiYdn6QpYWRkRHl5OX5+fpiamnL06FHKysrEPauhoUFubi6enp4kJyfTpk0bkbkyMDAgJyeHF198ERMTE/bv3y+uMek6UPxLiBEQxna6urqUlpbSuXNnYbQqPfAaGxuZMWMGu3fvFhkq6fs3NjbGwsJCyPBL04De3t5cuXKFtm3bUlhYKHb0ZWVl9O/fnzNnzgDg6OhISUkJ+fn59O/fn8TERJERlvRp4MHUUs+ePTl79ix6eno0NTXh4eHB+fPnMTQ0xNPTE7lcTn19PdnZ2UIzRsrsSfYZUnA/duxY3nnnHaqrq+nYsSMrVqzgrbfeEqWprKwsfHx8WLhwIffu3WPZsmWUlZVhZmaGl5cXR44cETIWdnZ2eHl5cfz4cXGOTUxMyMjIwMzMTAihqaqqUlpaipGREUuXLuXQoUOcPn1arOPFxcVCeE1NTQ1bW1tiYmJEgG9iYoKlpaXIIn399dcYGRmRnJxMdXW1EEKTMiuKf4nfSZNCgCjZSe0MUlbMycmJ6upqNDQ0SE5OFuaY0prg5uaGnp4e+/bt+9Wz7XE81xmT0aNHk5OTw7x587C2tuby5cuoq6tjZmZGx44dxcNg0aJFnDx5En19feFD0a5dO/T19TExMSEoKIgFCxawdOlSXnjhBdq1a0dFRQU2Njaihunm5kZ9fT35+flCS8HU1JSmpiZxomtqamhsbKSgoABjY2MyMzNp3bq12MV07dpVqF66ubkxfPhwevTogbe3N8eOHSM+Pp4PP/yQhIQEfH198fDwQFtbm+zsbMzNzXFxccHPzw+FQiE0UkaMGIGJiQm9evUSrpSSGq1Ue5cyKfb29vj6+rJx40YGDBiAmZmZiKgbGxuJiIhAR0eHDh06iJs0NzeXixcv0rp1a6ysrNDS0uKXX36hqqoKPT09LCwshAJlly5d0NDQEOOWNjY29O7dGwMDA2JiYoRjZseOHdHT08PW1hYtLS2xQP3888/Ex8eLHa6EpqYmGhoaaGpqYmtrS4cOHfDw8MDc3BxPT08AkpKSKCoqws3Njerqatzd3enSpQt+fn5i19i6dWsRfEr1VV1dXaysrPDy8mLevHnY2NjQunVrLCwssLe3Z9OmTdjY2NCtWzfWrFlDcXGxSE9OmTKF2NhYXF1dRVlQGus2NDQUYkP19fW0adNG/I6enh6jRo2ivr6ekJAQioqK+Pnnn0VAKZUbJQdalX/54wCi3i8FxllZWSQmJtLY2IipqSkvvPACrq6uYqzYxsaGQYMGoa6uzrhx49DR0cHMzIz6+nrS0tIoKChokcmTxAUzMzPx9PQkPDxcnI+VK1eKa8DNzQ17e3sMDAywtbVl2LBhAHTt2hWA7t2709TUhJOTE4GBgXh7ewul5srKSqEW3KdPHwwNDenTpw9Hjx5lyZIldO7cGW1tbRITE8X4rlS3r6mpEYF+SUkJpaWl3Lx5U/RibNiwgU6dOjFixAjhVxMUFMRPP/3E999/j4ODg7hXjYyMUCgUeHl5oa2tLYQGJXdcyZ27qamJyMhIcW9JmjYymUwEV1L/VF1dHV26dCEpKUn0j0gPBDs7O7y9vZHJZAwZMgR9fX0RbCoUCqF9M3r0aHR0dPD19cXFxYXy8nL27NlDRkYG4eHhpKWlCdGtsrIyCgoKRCDj5OTEwIEDMTY2pqKiQuzuFyxYwLBhwxg8eDBmZma0a9eOL7/8khdffBELCwsx8l9UVISxsTF37twRDr0rVqwQnjlSULF3715RetLR0UGhUODh4SECTKlcJfWKaGhoEBcXR319PW5ubowYMYLx48eLMdG6ujrCw8OZNm0ay5YtE75la9asYdeuXQwdOpT3338fKysr8brV1dXY2NgwZ84cUf7Kz8/H0NCQwMBA4VBsZmYmHK0NDQ3FRkVTUxMvLy+cnZ2FErOtrS3Tpk2jpqYGfX19WrVqxcWLF0lLSyM7O1s4epuamoqRb1NTU/r164etrS0pKSl88MEHomQcHx/P/PnzSUtLo7CwUAT7knHgt99+KwLyqqoqoqOjRaDcr18/dHV1hcK2VE6TdLGWLVtGt27d+OKLL4iLi8PBwYGhQ4eyefNm3NzckMlkREVF4ePjQ/v27dHS0kJdXR1XV1fi4uLE+iRtsMvKykRWq7q6Wqw9kmq6pBcjbeSb++FIWa/S0lIhiSBpddXW1orRemkTkpGRQXZ2Njk5OZw/f57c3Nzf9ex/rjMmAFu2bOHWrVts27ZNpNeqqqowNTVFJpOJzEevXr344YcfUFVVpaamRiz2UupZknqvr6/H2tqa3NxcdHR0OHHiBK1atWLAgAEcPXpUaCRIzrx1dXVinlz6f6qqqri6upKamoqnpydNTU0kJCRQWVkphH9u3rwpdliS/LqkoijtHKRMiJSylmbFpbq31AdiY2NDUVERampqQvVWylhIqrWmpqbk5eVhbW1NeXk5MplM7GJOnz6Ng4MDTk5O5ObmCrXRVq1aIZfLqa2tpWfPni1m2MeMGcP169eF7L6joyNTp07lgw8+wNzcHHiwY/niiy+YNm0a8+fP591338Xb25uVK1eKWX4NDQ1GjhxJZGRkC4EiaVfbunVr8V0bGRlhYmJCU1OTMMFasmQJe/fuRUNDQ6izVlVVCREoU1NTAJKTk3FwcODevXsMHjyYkydP4uDgQH5+Pn5+fqLh09/fn5CQEDIzM1m4cCHu7u5CcO+FF15g7969GBsbk5aWRmpqqkhnSjdyQ0ODaNacNGkSkZGRpKamigyKTCbD2NiYV155hY8++kjoCUgCSJKxoVTjLS4upn///tjZ2dGtWzcWL15MYWEhGhoaGBgYiD4oSWZa8S8TNIVCQUFBAT179qS6upq4uDixYx85ciRHjx4VplvSbtTGxkakoiWzvLS0NCwtLYWAndRU6ejoKByO9+zZQ48ePXjllVfYvXs3kydPZvny5YSEhBAZGSms4A0NDSkuLkZVVVWkkZOSkoTmgaTAK5fLhcaD1LAoZRdKS0vp0aOHCHRLSkoYNGgQp0+fZvDgwSQlJZGSkiJq+WpqapSUlIj3vHnzJm3bthV/29jYKBp8pSDpxo0bFBcXM3/+fDQ1NcWC/dFHH7Vwxq6uriY7O5uKigrq6uqwtbWluLiYTp06iZ2m1CMTEBDAhg0bGDt2LNeuXQMgJiYGVVVV4YCtpaVFRUUFdnZ2ZGRkYGVlRU5ODu+++65Q2+3UqZMQiLxy5QoJCQl4enri5OREbGysyMQ5ODhw/vx54uLi6NevH127duXEiRMtxLJkMhk6OjrMmzeP1atX079/f44dO0ZDQwMdOnQQ/V9GRkbExsaK7K/Ue6aqqkpYWBgnTpwQcurZ2dlCeE4y7ZTk76Vm1Z49e3L+/HlycnKora0lMDCQs2fPoqOjIxrJvb29UVVVZevWrbzzzjvs3bsXFxcXzp07R8+ePZHL5YSHhxMeHk6nTp2IjY3lxRdf5NtvvxW9fFLzs9TgLJkDSvosDQ0NjBkzBoVCIQQ6JbVT6TkhNfJLzdBSZtXMzIzq6mr27NnD6tWruXPnDvr6+mhqahIbG0tDQ4PQjurVqxdHjhwRitz379/n0qVL+Pj4cPz4cYyNjRk4cCA7duygpqZGlFWuX7+OpaUl9fX1rFq1ig0bNtCvXz+hf+Tm5sbChQsJDAxky5YtLfRhamtr8fPzIz8/H2traxoaGoiMjGTevHmsX7+eqVOncvr0aaysrITJqbGxMfb29shkMi5fvkyXLl3Yv38/AKWlpeL8SZmOuro6evXqRVlZGXK5XBjENs9kSht1KcCRNKCk/qA+ffpgbm5O7969mTZt2lM/95/rjAk8cGDV0tLC19eXzMxM4YhYWlqKk5MTGhoauLu7c/v2berq6hg2bJiYOnF0dBSLdXZ2Ng4ODqSnp/PDDz8wcuRIevToQW1tLZ6enkRGRlJVVUVaWhra2tpUVVURFBREU1MTdnZ2FBUV0bNnT7FDfvHFFxk7dizXr18nOjpaCNckJSWJdKCOjg65ubmUlJQIYTAptd2hQwfMzMwICAhg6NChGBgYiGyEpqamKJlIu2+pzipJLUvql1IzqjS1JJfLRRrTx8eHrl27YmNjw4YNG7C0tGTUqFGcOXNGSLW/9NJLyGQy+vbtS11dnZDDP3bsGAkJCaSmpqJQKCgqKuLEiROMGjUKDQ0NMjMzuXz5MhcvXiQ9PZ358+fz3Xff8d5771FcXMzVq1fF5z127BiHDh0iKCiIoUOHihKc5EwqOdHm5uZSXFyMiooK69atY8aMGXzyySf4+Phw8uRJrK2tMTAwoH379gwcOJA+ffpQUlIigjipZnrmzBnRbFlQUMCRI0fo3Lkzcrmc6OhoFi1ahJaWFo6OjqSnp4udl1wux9DQkLCwMDIzM+nXrx8jR44UwY8k1CWllb/55hvRuKihoSEWjMzMTFauXElNTQ1Tp07FwMAANzc3srOzadOmDerq6mzduhUzMzPs7e2F/P78+fOFk6+Wlhb3798XzWldunThyy+/FD0gin95u8TFxREfH0+bNm2wsbGhsrJSiECNHz+e4OBg9PT06N+/v2iIKysrIy0tTUyZ3b9/H3Nzc7ELlsprKioqXLt2jVWrVmFqaoqtrS0NDQ2Ym5vj7u7OP//5T8LDw/H39+fll19GS0uLuXPnoq6uTmFhofAuUVdXJyQkBD8/P9E86eXlJSZYjIyMGD9+PEuXLqWmpoaYmBh0dXVp3bo1Ojo6hIaG0rVrV6Kjo7G3tycgIABdXV0hkKinp4elpSXm5uZoa2uLXqrZs2ezadMm4c48c+ZMMjMz6dSpE6qqqpw5c4aTJ0+ycOFCFi9eTFNTE506dcLCwoJ33nkHuVxOZWUlnTp14s0332T06NFMmjSJq1evMmDAAMaNGyfUSy0sLMTDWlNTk5EjR+Lo6Ii7uztyuRxTU1PRkH/v3j26dOnC+PHjCQoKIioqipycHKZMmYK6ujoXLlzg7t27pKam8sorr7B8+XKWLl2KQqEgPT1dZDnr6+sZMmQI1tbWbNmyhWHDhhEcHIyFhQVBQUEsXboUc3Nz1qxZg4aGhvib0NBQ8ZCVpv+k8p2dnR0DBw4kKioKCwsLvvvuOxQKBU5OTqSmpvLVV1/h7++PhYUFzs7OeHp6Mm7cOMzNzTE3N6dPnz5ERETg4ODAl19+iZGREd999x3FxcWsWLGCvLw8DA0NGTduHAUFBURFRZGXl4eFhQUpKSmiZGdhYcGkSZOwsrLCxMQEFRUVoqOj0dXVZdmyZaKBPTc3l4KCAoyMjMjLy8PX15dTp04RFBREr169qKmpIS8vT6g+FxUVYWBgQO/evXF2dkZbWxs7OzuMjIyYOnUq+vr69OvXT5QPjx49KpRuCwoKuHXrFgYGBlhZWREWFka/fv2Ij48XgxWbN28mKSmJefPmsX37dgICAjAzM+PmzZv4+fkxbtw41q9fLyxQ1NXV0dbWZtWqVVy7do0TJ05w5MgRCgoKmDRpEsbGxpw8eVLYXUgZ/AEDBqClpUV6ejqXLl0iJyeHIUOG8M033wgxPScnJ9TU1Dhy5Ag9evQQauIRERH4+/sTERFB7969+fLLL9HU1BQVgg0bNhAYGEjnzp05c+YM8fHxojXC0NAQmUyGvb09GzZsQE9Pj759+6Kvr8+KFStEeb9z58707duXiooKOnXq9LuCEvgLBCZ3795l1apVAMKuXF9fnwEDBpCdnc2VK1dISkoSI6QNDQ0MGjSI0tJSsVO1s7PDysoKNTU1rKysWLt2Lb/88ouQ+S4tLcXV1VXUgaW63vHjx5HJZNy5c4eXXnqJ8+fP07ZtW6ysrDhy5AgODg4sXrxYNPpJvjr379+nrKxMlJakSRRNTU20tbVFL0tVVRXXrl3j6tWrlJSUkJGRIXpdZDKZ8FSRSljZ2dncv39fRKcaGhpoaGjg6uoqPDtUVVXR1dVFT0+PN998k7CwMDp37sy6desoKCggOjqarl27smnTJrEYaWpqcuXKFVRUVCgtLRVp461bt2JgYCDMomxsbNixYwe1tbV88MEHdO/eHV9fX4KCgmhsbMTLywtzc3PMzMyEnLyNjQ0VFRUMGjSIhIQEbt++jaurK6NHj6ZTp07o6OiIHVqrVq1EY6tcLhc3goeHBz169KCwsJB58+ahra2NQqHg2rVrWFtbi4ZdY2NjLC0t0dbWxtLSEmNjYzw9PfH09KR9+/bo6enh4uJCYmIi2dnZrFu3jqtXr3LmzBkiIiLYunWrGBG1tramoqKCrKwsYeInZXQUCgU+Pj5C1TYxMREbGxvRuNy6dWtGjRqFTCYjMjKSvn37Mnz4cDQ0NERgeurUKVq3bi12O5L6rWQm6ePjI5qfvb29yc3NZfv27dja2qKjoyOa0mpra3F2dhb1bema0NHR4cCBA4wfPx6FQkF5ebmQoK6pqaGwsFCU5lxdXUVpytXVVSh6Xr9+HRMTE+RyOSNHjiQnJwd3d3dycnIYOXIkjY2NwrtISvtK2QV9fX2CgoIwMjJCJpMJa4D09HTRhKiuro6XlxcGBgacPXuWhQsXoqamRm5urugjUigUzJ8/nzFjxoh0tY6ODiNGjEAulyOXyzExMeH27dvExMRQXl7OxYsXKS8vJzIykvnz55OZmUlqaipbtmyhtLSUF198kQ4dOrBv3z4mTpwo+jsMDAxISUlBRUWFiRMn0tjYiLW1tShJrVq1ilWrVhEVFYW2tjYHDx7k3r17uLm5iZ6DHTt2kJWVRUlJifDpCQwMpKmpiYEDBwrTvhs3buDo6MjSpUvx8PAgNDQUTU1NfvzxR2xtbRk5ciRaWlps2LCB27dvs3r1aiH5vWjRIpydnVm6dCmGhoaMHDmSGTNmoKKiwqlTp4QU+2effSYar/38/IRHjqRS+v3334vpt7CwMFq3bk1eXh4XLlzg1VdfpbCwUPSDSQqoW7dupaqqSpQlz5w5Q3JyMmpqagQGBnLkyBGCg4O5cuUKy5Yt4/79+wQEBKCurs7w4cMZPnw4L7/8MhkZGeKaSk1NZefOnXh5edG3b18xavzzzz+L9XTAgAHU1NTg7OzMV199JZowpb6m/Px8mpqaGDJkCJcvXxZZrMrKSqZNmyaaV7OzswkPD+ell14SGe033ngDgD179ohgv7S0FA8PD7Kzsxk8eDCXL1+mY8eO9OvXT2yU/fz88PPzQ0dHh5CQEHr27ImzszP37t1jypQpvPnmm+IZkpKSQmVlJdeuXRNZnuLiYlGOk3qP3Nzc0NDQICEhgaysLDw8PNDQ0KBLly7CAiE3N5eYmBi0tLTEc2/u3LlMmDABFxcXZsyYgZ+fH5WVlRQUFHDq1CkRBEZFRQEPehB/+ukn0U+oUCjEJvTNN9+koaGBnJwchg8fjpqamsh2JScn8/rrrzNkyBBOnTqFh4cHBQUFdOjQgYULF5KSksL58+fZtGkTZmZmyOVyYYT4e3juSzmSCZ69vT1btmxh1KhRIp2k8i/DPckHpU2bNpSWlmJjY0NMTIx4DWnRlBoJLS0tAVi3bh3du3cX9cvq6mratWsnzNWys7PR09OjtLQULS0tbG1tWblyJXPnzuWll17i0qVLdO7cmYyMDAoLC4mNjRXz3817KB53iqXUvJTqVvzLGOzh35dqfs0b0FRUVMTDV3L8bd26dYtGPX19fTw9PQkNDWXv3r1MmDCBTz/9FFNTU/Lz8+nduzfHjx+nrq6Oqqoq0dhkYGAgMjwqKirMmDGDn376iVu3bjF8+HDRBJmfn8/Ro0dRV1cX35OlpSUGBgZiusDU1JSsrCzRs5Oamoq1tTUTJ07Ey8uLzZs34+fnR0xMDMOHDycqKoru3bvz2muviebn5cuXExcXR01NDXfv3sXHx4f+/ftTWlrKnDlzUFFRYdCgQY88xzk5OaxYsQI9PT3S09NFvVdFRYXIyEjy8vJa/P6uXbsA2L17N+Xl5bi5ufHPf/4TJycnrKys2LRpkziu0NBQDA0N6datG9HR0RQUFKCpqSlKMVJJJCUlBX19faqrq5kyZQp79uzBw8ND+DxlZGQwefJkNm3axIABA/j555/JzMwUo+yTJk3Czc2N8+fPExkZybvvviv0c5YvXy7k1KWx4zVr1rB3716uXLlCt27dxASGqqoqGRkZ2NjYAA+8O2bMmCF6aLKzszEzMxNd/NL38Dik8zB27Fh0dXXJzc0VvjL6+voYGxuTnZ0tpuV2797dYqRZS0uLzz//nNWrV2NpaUl5eTlxcXHcv38fCwsLCgoK8PT0xMjIiFmzZrF582YOHTr0q+MICAjA2dmZpKQkbt++TWFhIYMHDyYrK4spU6YQEhLCrl272L59O6+++irwwIcrKSmJXr16oampKcbQ9+7dS1xcHO3atROvb21tzejRowkJCWnxvtu2bWP58uVkZmaip6eHpqYmQUFBbN++naCgIAwNDTl8+LAItqRmbykI9/T0pLCwUNhIREREcP/+fVJSUti9ezd79uxh7NixjB07FoVC0eLz29raoq6uLsw/S0tLSUlJYfv27axZs0aY182cOZPCwkLkcjndunWjpKSEgIAA3nnnHW7fvs2sWbOIjo7m/v37qKqqkp2dLWTeLS0tcXZ2FplMBwcHrl69io2NDVpaWiKTaGVlBUDPnj2Jjo5my5YtzJkzh6+++oqGhgbmzp3L5s2bsbGxwcjICEtLSw4ePEhJSYn4/mbPns3bb7+Njo4O2dnZDB06FBcXF+Li4oiMjKR3796cOHECLy8vMjIyCAsLY8mSJWzfvp3333+fjIwMmpqaGDRokBiDVVFRESWZyZMn8+KLL1JcXMzbb79NREQEc+fOpVWrVmzevFlsHj08PJDJZGRkZODk5ERSUpJwLpfJZBQVFYlAMTw8HG9vbywsLEhISMDBwYGSkhLu3LnDxo0bhT+aQqHg5MmTbN68mbCwMGGgaW9vz/nz5+nbt6/wVpIk3k1NTfH29mbfvn0MHToUHR0dYmNj8fDw4PDhw7Rr145vv/2WN954Az09PZE9T0tLw8/Pj9raWgoLCxk/fnyLa1ZFRYWQkBDq6+uJiIggMTGRY8eOiYy9lZUVp0+fBh4ELzNnzmTfvn2iXFpVVUWPHj1wc3PD3d2dkJAQwsPD+e677zA2Nqaqqor8/HyR+XV0dCQwMBAfH5/HrtGPQu2pf/MZIX3RpqamXL9+HXNzc0xMTDA1NaV9+/YA9OrVC4VCwbFjxzh48KAQd+revTvBwcEAzJ49m759+7YQeZG8RcrKyrh69arwejAyMhLunpIzYvfu3fnyyy/56KOPsLKyIjIykuTkZCZNmkSbNm1YvHixqAu2b9+egwcP0tDQgIeHh5hHB0RgIekJ1NXVYWVlJZw5pYcmILxipIuuZ8+eaGpqirq3ZDAleURI0xEWFhbU1tZiY2NDjx49qK+vFwFKc6TR0n79+gk34Dt37hAbGyv+/frrr+Pq6oqrqyuHDh1izZo1ANTU1PDdd98RHBzM/v37SU9P56233qKwsFA027Zr146kpCReeOEFYUgn2WvL5XIsLCwIDAzk9OnTfP311/To0YMff/xRPJCl5uchQ4bQtm1bEhISWLt2LZaWllRXV1NeXi4EsR6Hs7MzI0aMEIFmbm4uCxcuJCoqivfff/9Xvy+do7q6Onr06EFUVBSLFy9m4cKFDB48uMVxBQQEcOfOHVJSUlBVVWXUqFHCA+frr78Wn9fe3h5zc3POnTuHu7u7CAjt7Oy4d+8ePj4+XL58WYxujh49mrZt27J582by8vK4du0abdq0YfXq1Zw7d65FsHDp0iXeeecdBg4ciKWlJenp6bz77rvo6elRVlYmzmVzbty4we3bt3F2diYsLIyZM2cyY8YMBgwYIFLHx44de+TfNkc6D507dyY3N1eUWZuamrCysiImJgYLCwvOnj3L2bNn6d+/P8OGDePMmTN07twZgLKyMhH8jR49mqysLAwNDdHS0kKhUODs7ExaWhq5ubkEBgY+8jiCgoLYs2cPEydO5L333mPdunXExsZibGwsgonQ0FBiYmLE9yuTyRg8eDBnz54FHizY+/btY8qUKeLc5OTksGPHDoYPHy6OtzmTJ08WOg8xMTF8++23xMbG4uTkJJxUJRfqoKAgIiMjad26NSUlJdTV1ZGSkoKOjg4BAQG0bdsWeLBObd26lY0bN2Jubs6iRYvo1asXAIGBgaxdu5aYmBiGDRtGVVUVu3btahFMr1mzhmvXrrFgwQKOHj3KZ599hqGhodi8FBUV0b9/f+rr63n99ddZu3atyAIOGzYMNTU1CgsLuXz5Mnl5edTW1tK6dWuxTkrrqYqKCj169BAPOokrV66gqqqKTCZj7dq1rFmzhuzsbEaOHImKigo9e/YkJCREaHFI39+0adOwtrbG19eXVq1aMXDgQAYNGoS/vz9jx46lY8eOxMfHi2mZ9evX4+7uTnh4OK1ataK2thYfHx9++OEHdu7cyblz54AHwdK+fftaTBJJjav79+9n+vTp9OrVi3v37pGVlUVBQQF+fn7Y29uTnZ2NXC5n/Pjx4vtUV1fHycmJ48ePc+fOHU6fPs3HH3/M7du3yc7ORkdHh8zMTCZMmCDEzgAWLVpETk4O1tbW3LhxA319fWQyGRMmTODkyZOUlZVhYGCApaVli/6MrVu3cv/+fSErIPUiJicn4+3tjZGREYMGDSI0NJR169YxYMAAPvvsM+DBpv7hNV9CJpOJdaRLly7s2bMHe3t7zp07x5gxY0R1IDs7m5KSElRUVHB3d6dz587i3yEhIUyYMIEJEyawcuVK2rZti6amJsuWLSMmJoYzZ85QWlpKamqqCF6fluc+MJFSd0OHDqWmpobQ0FAWL14sfj5v3jyOHz+Oq6srMTEx9O3bV1hXS4JLVlZWTJw4kX79+hEdHY2dnR3z58/Hx8eHqqoq1q5dy/79+yksLCQ1NVWMdUnqe9bW1kybNo3S0lKxs582bRqurq5UVFTQtm1b1q9fj5GREbt27UIul/Puu+9ib2/PqlWr0NbWpmvXriQnJ1NRUYGxsTEqKiqkp6eLEoGdnR35+flCqEcat2tqaqJDhw5CTCsrKwsnJyf09fXZsWOHeJirqqqyf/9+duzYIWzZExISGDFiBGZmZkyePBmgxYUaGhpKdHQ0JiYmrFmzhvDwcL744gsaGhq4cuUKQ4cOJSkpiSFDhoixSglNTU0mTpzIxIkTgX8/HDZu3Mhbb71FmzZtSEhIEDcN8NiI+cCBA0RFRWFqakp0dLT43ebHWl5ezpYtW9i1a5eoF+/atUuoCz6O+Ph4lixZQmpqKunp6fz0008EBwe3CIAexZQpUwDEyO1XX33V4rw9fC6nT5/Ot99+i7W1NcXFxfj4+KCurs706dM5cOAA69ev56233iI0NJRffvmFiooKXn75ZdE3Ik20BAcHc/z4ceLj42lsbCQgIICLFy+ybds27t69i7GxsUjHKhQKSktLuXz5MgYGBmzcuJFXXnmF/Px8KioqePXVV8UDV2LDhg3k5eXh4+PDN998Q0JCAi4uLpw+fZpRo0ZRXl5Ox44dW3wPj+NR5yEnJ4f27dtTWFgozomhoSFubm7iPMXHx7NgwYIWrzN79myWLVvG0qVLmTlzJqqqqjg5OfHLL7+IPpMnHce1a9cIDQ1l+fLlnDx5UvQ4denShW3btnH58uUWD8MVK1YI8a/Dhw+LEV14kEndvHkzy5YtIzQ0lA8//PCxo47ScXXq1Am5XE5jYyNhYWGi3BoSEsK4cePo2rUrDg4OLFiwAGdnZ8rLyzE0NOT48eMigwUPRLNmzJjB3LlziYuLo7KyUgT1AC+99JIwSpSCGekYTp48iZmZGampqWKiqH379hQVFQnBrxdeeIGtW7cSFBTEa6+9hrq6Ovn5+cyfP5+oqChRlp49ezZXrlwR+k36+vqYmZmxYcOGx34P8GCXDbB27VpUVFS4evUqGzZsENcsQFRUFJmZmb86h82RpjgMDAyQy+WMGjUKBwcH3nrrLVJTU/n0009paGjA2NgYuVzOzJkzuXnzJtu3b8fc3JyIiAjxWgcOHMDMzIyoqKgW61i3bt2wtrYmPT2dd955h2nTpgnT0MLCQlJSUggODiYxMZFly5ZRUFDA4cOHadu2LTt37qS8vFxMwlRXV+Ps7Mw//vEPbt68SWpqKt7e3qxYsQIfHx9Gjx4tNh/r1q1j5cqV5ObmcvLkSezs7Lh06RJFRUWiPGppacn9+/fp2bMns2bNYs6cObRr144ZM2awfft2MQEjNUW/8cYbtGnTho8//lh8bkkE7bcIDQ0lMjKScePGYWpqyvLly0UGHB4Ed81pHoy6uroybNgwtm7dyvjx4/n555/55JNP8PT0RFVVle3bt9O9e/enOo7mPPeBSfNdrZR+ak5paakIXCRZ6MTERDHudPDgQZFu3717N7dv3xZCSLNmzWLv3r388ssvJCcn4+Pjw6VLl/Dz8+PmzZusXLmSgwcPkpeXh0Kh4NatW2Jnn5OTw8mTJzlz5gw2NjbMnz+fK1euYGJiIsbGBg4cSEVFBba2tly/fp0BAwYITQwANzc3odbo5uaGiooK69evJzo6WjQbSW7E3bp1o6CgAEDIPj/8MJeaCTMzM6msrGTq1KmkpaXx3nvvPfLcSjstQFhzX758mSNHjtCuXTucnJz4+eefOXXqFMBjd6zw4OL+4Ycf8Pf3R1dXFzMzMxITE2nfvj2nTp3C3d39sX8r7bylmvOjMiCzZs0iLS2NI0eOYG9vD8DMmTN/MzBZvnw5n3zyCYaGhgwbNozp06f/KgD6M3BychKNYEeOHMHIyEjU1+3s7JgwYYJQjS0uLiYqKop//vOfREVFsWHDBo4ePSo63H19fYmNjWXAgAFUVVUJB2p4oFCck5Mj/jsuLk4EtpMnT6a6uprdu3ezc+dOgoODRbpcIjExkc8//xyAUaNG0alTJ3766Sd8fX1xdnYmNjaWCRMmPPZ7+C2cnZ3p1q0bZmZmXLt2jczMTPT19UlMTKSkpISwsDBu3rwpHlTS+W/VqhVvvPEGdnZ23LlzB4VCwY0bN9i/f/9TKUZKNfPevXszfvx4cR/Cgwdhx44dRXAOD7Rd4N/X++zZs8XPpDJsU1MTXbt2FQHA4/D19cXR0ZEpU6aIEczm15WbmxvTp08XZRgTExM+/fRTPvzwQ5YvXy6ajzdt2iRKByNHjmTDhg2cO3euxWt9/fXXotdHoVAIxWJp5PPMmTOUlZXx+eefU1JSgr29vbAScHJy4sqVK+jr69O/f39u374tSkKenp4cOHBATBcaGRlRU1PDuHHjWLRoEZcvX+bYsWNP/T1IiqnS5GHzaxZgzJgxLf77cYGnhYUFP/74I1u2bCEpKYnPP/8cFRUV5syZw5dffomHhwfR0dHiel+3bp2QGIAH15e0vkhI65iJiQmhoaEsXLhQ2BP079+f69ev8+OPP/LWW2+xadMmXn/9dY4dO0arVq04cuSIuB5nzZqFv78/bdu2ZebMmdTU1BAQEMCJEydEb+Dt27cZOHAgCxcuxM3NTWScMzMzUVdXx9LSkjVr1jB8+HDs7e2pqqpizJgxbN68mSlTpvDuu+/i7+/P+PHjCQgIICEhQahE37t3jw8//JBevXoxe/ZsFAoF586dw9fXl6tXr7bQzPotlixZQlRUlNgwSNnh5t9NbW0t+/fv5/DhwyIwkQLM2NhYVFRUyM7OJiQkRGxQ9uzZ878ZmMybN4+srCxcXV1JSkpiwoQJLX5+7949tm7dKkYwi4qK6NChA4cOHSI3NxdjY2M+/vhjZs+eLWbaBw8ezKlTpwgNDSU4OJiIiAi+//57OnfuTGhoKKtWrcLExIT169czduxYtLW1OXXqFEVFRSIYuHTpEnl5eaJhzNfXV4wxDhs2jO3bt/PFF1+gqakpdB2kG1uhUGBubo6Xlxeurq5YWVnxwgsvUF9fLx4krVq1EmWAAQMGsGvXLry8vFiyZAna2tr8+OOPLR7m//jHPygtLRVjjZqampSUlHD37t1fPQQeRX5+Pj/++KNYhO3s7HBzcxM73YdTto/itddeo2fPnlhaWtK9e3eMjIxapP0ex5N2wxI7d+5k1qxZIiiBfwvPPQnJihxg3Lhx7Nu374kB0B9l9OjRvP/++3z22WdiFFoSd3Nzc2Pr1q1iQZMWSnt7e3JycsQYeocOHX6VhXj4oezq6vrI87Vx40YcHBzIyckhNjaWa9eucf36ddauXdvi95pnDQD8/f25ffs2Bw4cwM3NjRUrViCTyf7weYiPjycwMJCysjLU1NTo2rWrUMK8c+cOcrmczp07/+r8r1mzRjQpu7u7ExAQwPr16zE2Nm6RrXoca9asISIigoSEBL755hsqKytFwKGrq0tmZiZbt25l6tSpwK+vdylohAcqv1LfA9DC4+pRvP7668CDslRZWdmvrqt79+7R0NDA1KlTCQgIYPLkyZw9exYrKysSEhKor6/Hw8MDePAANzIyYuXKlUL6/eHPGR4eLkSrKisrWb16NTt37hTCY5IAoDTFI5V8u3btKvqlvvrqK/r27Qs8CKwKCgr4+OOP0dXVZcOGDZiYmAjV3Ycfqk/zPUjXk/Q3T3OPP4pPP/2UOXPmsHnzZmbNmoWfnx8JCQlYW1vj7OzcYo2qq6sjOzsbOzu7FtfX4947PT2dqKgompqahIz+2LFj8fHxabGRyM/Pp6amhk6dOgEIaYCdO3cCcOjQIaFjcvDgQc6ePYu9vT3u7u7IZDJcXFzYuHHjr977tdde48cff2Ty5MkUFhby5ptvcvHiRQICAoiKimpxH6qrq1NVVUVeXh4xMTGYmpqira1NQUEBq1evpqmpibVr14pz7+Li0iJ78lu4uLiI7DA86B+TiImJYc+ePcLS4/Dhw786h+np6Tg6OtKmTRuGDh3Kvn37hL3BH+G5D0yaZ0R++OGHX930OTk5dOjQgUWLFgEPHqC9e/cW6acvv/yS3Nxc/P396dGjh/hv6XWkWlufPn2IiooSu6rx48fj5eXVYrFpvrOfPn06JiYmzJ8/H21tbVRVVVssLo6OjgDcuXNHqKdKSKN6knqlFJgoFArOnj1LaWmpUDDMzs4mNTVVHL90PI962J87d447d+4ACP0RFxeXp3oIPxwxz5kzp8XF+TTU19fj7u5O//79MTc3x9vb+6kCmt+DtBt72h2BXC4XgZmZmRlaWlp/eJF8EsuXL2f37t0EBgaKBrvmN3DzB6z0/lLZIT09nfHjxwv9i4d50meWXsvX15eoqCi++uor9PX1WblyJRcuXPjVzvTSpUsEBweLBSMtLe030/O/h+XLl1NZWcm2bdvEzn3FihX07t37iX8n1dvDw8NRU1Ojvr7+N3tcHv771157jTVr1tC9e3dCQkJo3769KEM+zJN2iMuWLWvxu791vfzWz6UMmTQZYW5uzvnz53n11Vfp3r17i2Dxt15LymxWVlYyc+ZM3nvvPUaMGMG2bduYP38+b7zxBocOHUJDQwNnZ2cqKyuFQmxYWJjw5AFEv93rr7/Onj17qKysFJMjenp6YnwcYMaMGU88LonmvQt/Burq6mJsuaCggPDwcDHB9vHHHwufJniQAVmyZAmurq4Av7nmBQUFkZOTI7xhhg4dir+/P/7+/hw9epS4uDiCg4PFSPqjrhV4sNFoaGjg9OnTfPbZZ9jZ2dGrVy+mT5/+yN+X3vv+/ftUVlZibW1N//792bFjB/DgHpJKghK+vr6oqqpibm5Ov379RCP1Z599hqWlJREREairqzNz5sw/cJafzIsvvsjixYt5+eWXmTNnTotzLp3DoKAg4EH/mpQlXrx4MVlZWX/oPZ/7wKR5RkQq40i7HnhwMSYlJbX4m969e4sb/OrVq1hZWfHRRx9x9epVnJ2dH3nzPxwxfvrpp091fE8KFAAsLS0ZNmxYi6ZbaaY7Ly+Pbdu2ceHCBeG7YWZm9lTv+zChoaH/rwfukyLmp+Xzzz9n9+7dLF68uEUPwZ/Fxx9//Lt3BI9L4/7Z6OrqcurUKQIDA1s02D2JKVOmtOhzeNQ5f9rPLH1/FhYWHDt2jDZt2gjH1OZcuHBB/Fuawvkz0dHRwdzcXNyT1dXVJCUl/WZgAg8eulJdXCaTsXfvXvz8/H7X+3/77bfs2bOHX375RTicNt8USPzW9Z6bm8vOnTs5dOgQY8aMEQ2of4RHlSpv377N3bt3uXv3rvi95uva45AyPXfv3qW8vJy+ffsK/ySAV155RZhNfvzxx+LhPGPGDHr37v3I7yE0NJTLly9jYWEhMlxpaWm/2uU/K7Zv386iRYvYs2ePmM56FKqqqsJjSvrcT7rupPVy/PjxqKurt1g/u3XrxoEDB8S9aWdnJ+7ph68V6e8mTZr0yPd51H39qLX6hx9+EC7Ohw8fJiUlRfxMysoB/5EN35P49NNPqa6uJjg4mDt37ghxNvj155CGPB7OEv9envvAZOjQobzzzjvs3LkTXV3dX/1cMgh7XMrojzzMnoanDQLkcjmHDh3i7NmzWFtbM3/+fPGz8PBw5s+fj62trRApa9738VfD1NRU+M1IF+ijHgp/lD+yI/hPZEceRUpKClevXuXWrVsoFAqSk5PJy8t7oj/Ek/ocJH7vZ5bGS7Oysti+fbsYK5ZYsGABH3zwAatXrxb+INu2bfsdn/TJyOVyYfAoCc41byx9Eg9fK09TqnsYX19ffH19qamp4auvvhJTY7+H4OBgrKysGD9+PGlpacybN+93H0dzHnUNfv/9909ctx6HlOk5fPgw33//vdBwkqbIHB0dW5SfpfLtb220PvnkEyIiIjh06BA3btwQfWXPA82Dzfz8fKFC+zBpaWkcOnRI9P29/fbbT/X6j3qth+/NP/qA/S2aXxtXrlwhJCSEadOm/eo9/1vr2KM4ceIEmzZt4vTp06xdu5adO3eKwORhpKEB4KnLsI/iuQ9MEhMTKSgoYOrUqWzduvVXP7ewsHhiVPyfSm89Lerq6tjb23P16lXRRCtF85K0PTzYwf1/avvPA1VVVUyfPh2FQsH06dMfqRPyv8ro0aMpLy8XC0hhYeFv9iY8qc/h/4ulpSWLFi36VSpV6nu5d+8eW7ZsYe7cuX/ae8KD87Br1y6MjY2BBwtV82zhk1AoFL+7VPc4NDU1CQsLa5EOf1q8vLy4fv06v/zyy29+h3+U31q3HkfzTI+/v3+Lsf0n9br81gah+TrZvn17wsLCRND08GTXf5vmwWZziYKHeXit/f/w8L35KHPWP5tn8Z5Pg9RHkpGRwa1bt4Rf1n+S515gTRJx+vzzz0XNrnkT5+Oi5+cFqWNf2sU079iXFEfhwc3/6quvioa0vyKrV69+pFDa34Vly5bRoUMHDh8+TEJCAtu2bRMp9kchTadMnjwZHR2dx5Zzfg/NxzIVCgWfffYZR48eFf9v3LhxDBw4kIKCAhYsWMCMGTMemxr/o/zRe7Kuro6IiAhu3ryJi4sLkydPblHP/m9x8eJFvL29uXr1Ktu2bSMzM5OTJ0/+qe/xPK9bza+JjIwMdu7cKXr4nmeetNb+Xv4T9+bz+J5Pw8NZ/P9GGem5D0yexUn5M/k9x9/U1CR0AP6KhIWFtdBW+T3Ni/8L2Nvbs2jRIkJCQp5YC/9P8qhSYPM08IwZM7hw4QIBAQHIZDLOnz/fIphR8kBrRRqpBnj11VfZvn37Mzyi/y6/N8B+XvirPyuU/JvnvpTzLGtrfwa/5/hff/11IWrzV6T5OPWfrRPyV+Bpa+H/SX7reps7dy5z587lH//4B++//z4vvfTSf+nI/jo8/J09TSPz/xJSs6k0ov9XCErgr/+sUPJvnvvA5H8Zb29v0RgomdL9lXkaobT/ZZ62Fv4ska43AwODp25K/bvxZ/a6/BV5HgJsJX9vnvtSzv8yD6eM33vvPZYvX/4Mj0jJ/zqSLPeGDRvEFNDfKav1NDwvvS7PGinAPnz48HMXYCv530YZmChR8jdCWYdXokTJ844yMHmGSDbsR44cISIigt69exMeHv6sD0uJEiVKlCh5Zvx1R0D+B4iJiQHg6NGjHD16tIV7rxIlSpQoUfJ3RBmYPEPu379PcnIy+vr6AH/pUWElSpQoUaLkz0D5JHyGvPbaa2zcuFFIXje361aiRIkSJUr+jigDk2dIfX09jY2NlJaWAo83gVKiRIkSJUr+LigDk2fI559/zsKFC/noo4+e9aEoUaJEiRIlzwXKwOQZ8rAbr+TIq0SJEiVKlPxdUY4LP0PCwsJQUVERjsN/JzdeJUqUKFGi5FEoMybPkI4dO1JVVUW/fv2or6/n/ffff9aHpESJEiVKlDxTlF45z5Bbt27x9ddf/23deJUoUaJEiZKHUWZMniEPu/Eq7eeVKFGiRMnfHWWPyTNE6VuiRIkSJUqUtEQZmChRokSJEiVKnhuUpRwlSpQoUaJEyXODMjBRokSJEiVKlDw3KAMTJUqUKFGiRMlzgzIwUaJEiRIlSpQ8NygDEyVKlChRokTJc4MyMFGiRIkSJUqUPDcoAxMlSpQoUaJEyXPD/wERzDZn2rgt6wAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import scipy\n",
    "\n",
    "den = scipy.cluster.hierarchy.dendrogram(\n",
    "    clustermap.dendrogram_col.linkage,\n",
    "    labels=distance_matrix.index,\n",
    "    color_threshold=0.25,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "c63a3f07-044f-44e7-bbb3-26890d565536",
   "metadata": {},
   "outputs": [],
   "source": [
    "# extract clusters and perform gprofiler\n",
    "from collections import defaultdict\n",
    "\n",
    "\n",
    "def get_cluster_classes(den, label=\"ivl\"):\n",
    "    cluster_idxs = defaultdict(list)\n",
    "    for c, pi in zip(den[\"color_list\"], den[\"icoord\"]):\n",
    "        for leg in pi[1:3]:\n",
    "            i = (leg - 5.0) / 10.0\n",
    "            if abs(i - int(i)) < 1e-5:\n",
    "                cluster_idxs[c].append(int(i))\n",
    "    cluster_classes = {}\n",
    "    for c, l in cluster_idxs.items():  # noqa\n",
    "        i_l = [den[label][i] for i in l]\n",
    "        cluster_classes[c] = i_l\n",
    "    return cluster_classes\n",
    "\n",
    "\n",
    "clusters = get_cluster_classes(den)\n",
    "# extract functions for clusters\n",
    "cluster_process = {}\n",
    "for c in clusters:\n",
    "    cluster_df = pd.DataFrame(clusters[c], columns=[\"gene_symbol\"])\n",
    "    res = get_gprofiler(cluster_df)\n",
    "    cluster_process[c] = res[res[\"p_value\"] <= 0.05]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "c4149852-b136-46a4-bdec-f065425455b6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# save grpofiler results\n",
    "with open(\n",
    "    \"figures/pert_embedding_cluster_gprofiler/pertspectra_replogle_pert_emb_gprofiler.pickle\",\n",
    "    \"wb\",\n",
    ") as handle:\n",
    "    pickle.dump(cluster_process, handle, protocol=pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "84864ce6-2724-4d11-a955-1ec4585c7876",
   "metadata": {},
   "source": [
    "## PR Curve+AUC\n",
    "- AUCPR using prior graph as binary label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "d281c6a7-767a-44c0-aeb1-3c5bf11e27a5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "               AUC\n",
      "StringDB  0.215162\n",
      "CORUM     0.262926\n",
      "                F1  Threshold\n",
      "StringDB  0.246123   0.869706\n",
      "CORUM     0.306326   0.884541\n"
     ]
    }
   ],
   "source": [
    "aucpr_df, f1_df, pr_dict = auprc(distance_matrix)\n",
    "print(aucpr_df)\n",
    "print(f1_df)\n",
    "aucpr_df.to_csv(\"figures/pert_embedding_recall/pertspectra_replogle_aucpr.csv\")\n",
    "f1_df.to_csv(\"figures/pert_embedding_recall/pertspectra_replogle_f1.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "cd6af182-b82d-4fe1-a5a8-fccb34556130",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the precision-recall curve\n",
    "prior = \"StringDB\"\n",
    "plt.figure(figsize=(8, 6))\n",
    "plt.plot(\n",
    "    pr_dict[prior][\"recall\"],\n",
    "    pr_dict[prior][\"precision\"],\n",
    "    marker=\".\",\n",
    "    label=\"Precision-Recall Curve\",\n",
    ")\n",
    "# Adding labels and title\n",
    "plt.xlabel(\"Recall\")\n",
    "plt.ylabel(\"Precision\")\n",
    "plt.title(\"Precision-Recall Curve\")\n",
    "plt.legend()\n",
    "# Save the plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8fdff66a-8f20-4268-8520-d33a0fa4969a",
   "metadata": {},
   "source": [
    "# Interpretability of our learned latent space"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1ec54f80-8a0f-41f0-8c55-edd2b1fbfb3c",
   "metadata": {},
   "source": [
    "## Factor Enrichment\n",
    "- GSEA on log transformed latent-by-gene factors\n",
    "- Then associate each perturbation to its top latent factors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "35eb8169-15b1-4b5b-b926-d5b7238f372a",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "factor_to_go = factor_enrichment_gsea(adata, adata.uns[\"SPECTRA_factors\"], fdr=5e-2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "814189ce-53f3-4716-834a-a4f0ea6efcf9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# filter and add description to processes\n",
    "go_df = read_aws_csv(\"s3://pert-spectra/references/GO_to_Description.txt\")\n",
    "go_df.set_index(\"Term\", inplace=True)\n",
    "go_dict = go_df.to_dict()[\"Description\"]\n",
    "\n",
    "filtered_factor_to_go = {}\n",
    "for i in factor_to_go:\n",
    "    proc = factor_to_go[i]\n",
    "    proc[\"descr\"] = [go_dict[x] for x in proc[\"GO_ID\"]]\n",
    "    filtered_factor_to_go[i] = proc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4e82f91a-70e9-44c2-b802-058348b77052",
   "metadata": {},
   "outputs": [],
   "source": [
    "filtered_factor_to_go"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8b91d73c-87de-4197-8f7c-2554ee3d0491",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\n",
    "    \"figures/factor_enrichments/MODEL_replogle_factor_enrichment.pickle\", \"wb\"\n",
    ") as handle:\n",
    "    pickle.dump(filtered_factor_to_go, handle, protocol=pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b01b3f40-7d97-481a-8e57-01627cd627ad",
   "metadata": {},
   "source": [
    "## Overlap with prior knowledge/ground truth\n",
    "- Group A: a set of GO terms associated with a perturbation (either drivers from msigdb, or from literature) and its neighbors in stringdb\n",
    "- Group B: a set of GO terms from the interpretability analysis\n",
    "- Hypergeometric test on the overlap of the two groups"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "96895bc7-560b-4893-9f3f-40a8f733779f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# load precomputed factor_to_go dict if available\n",
    "filtered_factor_to_go = read_aws_pickle(\n",
    "    \"s3://pert-spectra/figures/factor_enrichments/pertspectra_replogle_factor_enrichment.pickle\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "e9eca6ce-4fc7-4f6a-b140-0b2276d9abe4",
   "metadata": {},
   "outputs": [],
   "source": [
    "# get neighbors for each perturbation\n",
    "pert_neighbors = retrieve_stringdb_neighbors(adata.uns[\"Spectra_pert_labels\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "ce927aaa-f5b4-4403-ab21-beeb1202a4f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# construct group A - known processes for each perturbation\n",
    "gene_sets = read_aws_pickle(\"s3://pert-spectra/references/GO_to_Gene.pickle\")\n",
    "# BP only go terms\n",
    "go_reference = read_aws_csv(\"s3://pert-spectra/references/GO_terms.txt.gz\", zipped=True)\n",
    "go_bp = go_reference[go_reference[\"go_category\"] == \"biological_process\"]\n",
    "go_bp_ids = set(go_bp[\"go_id\"].values)\n",
    "filtered_go_terms = {key: gene_sets[key] for key in go_bp_ids if key in gene_sets}\n",
    "\n",
    "# GO terms per perturbation AND its neighbors in stringdb\n",
    "pert_to_go = {key: set() for key in adata.uns[\"Spectra_pert_labels\"]}\n",
    "for goterm in filtered_go_terms:\n",
    "    for pert in adata.uns[\"Spectra_pert_labels\"]:\n",
    "        if pert in filtered_go_terms[goterm] and set(\n",
    "            filtered_go_terms[goterm]\n",
    "        ).intersection(pert_neighbors[pert]):\n",
    "            pert_to_go[pert].add(goterm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "e9433684-9108-4d8b-8c0f-f9c3f79155e5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# construct set B - model identified processes for each perturbation\n",
    "n = 10  # number of top factors to get processes from\n",
    "model_pert_to_go = {}\n",
    "for pert in adata.uns[\"Spectra_pert_labels\"]:\n",
    "    if pert in [\"ctrl\", \"intergenic\", \"basal\"]:\n",
    "        continue\n",
    "    # get top factors\n",
    "    pert_embeddings_df = pd.DataFrame(\n",
    "        adata.uns[\"SPECTRA_pert_scores\"], index=adata.uns[\"Spectra_pert_labels\"]\n",
    "    )\n",
    "    pert_loading = pert_embeddings_df.loc[pert]\n",
    "    ctrl_loading = pert_embeddings_df.loc[\"ctrl\"]\n",
    "    delta_loading = np.abs(np.log(pert_loading) - np.log(ctrl_loading))\n",
    "    top_n_factors = np.argpartition(np.array(delta_loading), -n)[-n:]\n",
    "    # get processes\n",
    "    model_processes = set()\n",
    "    for f in top_n_factors:\n",
    "        proc = filtered_factor_to_go[f]\n",
    "        model_processes = model_processes.union(set(proc[\"Term\"]))\n",
    "    model_pert_to_go[pert] = model_processes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "739fb773-f08a-486c-b70b-3317bc749cd4",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Overlap for ATP6AP1: 0 out of 9 in researchDB\n",
      "P-value for ATP6AP1: 1.0\n",
      "Overlap for RPL30: 0 out of 5 in researchDB\n",
      "P-value for RPL30: 1.0\n",
      "Overlap for DDX21: 3 out of 13 in researchDB\n",
      "P-value for DDX21: 5.980456287677466e-05\n",
      "Overlap for PHB: 2 out of 24 in researchDB\n",
      "P-value for PHB: 0.010741839670463106\n",
      "Overlap for DMAP1: 1 out of 14 in researchDB\n",
      "P-value for DMAP1: 0.06354093376899977\n",
      "Overlap for MCRS1: 1 out of 19 in researchDB\n",
      "P-value for MCRS1: 0.10580925901854948\n",
      "Overlap for CCNH: 3 out of 7 in researchDB\n",
      "P-value for CCNH: 3.0005424766078534e-05\n",
      "Overlap for NUMA1: 1 out of 6 in researchDB\n",
      "P-value for NUMA1: 0.03329641937620341\n",
      "Overlap for ALDOA: 0 out of 6 in researchDB\n",
      "P-value for ALDOA: 1.0\n",
      "Overlap for SNAPC3: 0 out of 5 in researchDB\n",
      "P-value for SNAPC3: 1.0\n",
      "Overlap for SEC61A1: 0 out of 8 in researchDB\n",
      "P-value for SEC61A1: 1.0\n",
      "Overlap for CCT3: 3 out of 10 in researchDB\n",
      "P-value for CCT3: 5.402090913326375e-05\n",
      "Overlap for EIF6: 2 out of 6 in researchDB\n",
      "P-value for EIF6: 0.000615303446286681\n",
      "Overlap for UTP20: 1 out of 5 in researchDB\n",
      "P-value for UTP20: 0.04628097253128953\n",
      "Overlap for FBL: 1 out of 6 in researchDB\n",
      "P-value for FBL: 0.028199168104588085\n",
      "Overlap for RPS7: 3 out of 9 in researchDB\n",
      "P-value for RPS7: 8.84599979430884e-06\n",
      "Overlap for BUB1B: 2 out of 7 in researchDB\n",
      "P-value for BUB1B: 0.00040051787251466374\n",
      "Overlap for PES1: 0 out of 6 in researchDB\n",
      "P-value for PES1: 1.0\n",
      "Overlap for ATP6V1A: 2 out of 11 in researchDB\n",
      "P-value for ATP6V1A: 0.0007310971425003066\n",
      "Overlap for NFRKB: 1 out of 13 in researchDB\n",
      "P-value for NFRKB: 0.02948322206040407\n",
      "Overlap for MIOS: 0 out of 5 in researchDB\n",
      "P-value for MIOS: 1.0\n",
      "Overlap for NEDD8: 1 out of 8 in researchDB\n",
      "P-value for NEDD8: 0.05022968114062165\n",
      "Overlap for TAF1: 0 out of 23 in researchDB\n",
      "P-value for TAF1: 1.0\n",
      "Overlap for RAD21: 1 out of 11 in researchDB\n",
      "P-value for RAD21: 0.015586554804595597\n",
      "Overlap for RPS6: 4 out of 9 in researchDB\n",
      "P-value for RPS6: 3.850507725903189e-07\n",
      "Overlap for MAX: 0 out of 7 in researchDB\n",
      "P-value for MAX: 1.0\n",
      "Overlap for VPS72: 1 out of 10 in researchDB\n",
      "P-value for VPS72: 0.08166368798369315\n",
      "Overlap for GATA1: 2 out of 20 in researchDB\n",
      "P-value for GATA1: 0.008055568024662566\n",
      "Overlap for SUPT4H1: 1 out of 8 in researchDB\n",
      "P-value for SUPT4H1: 0.009473365146555657\n",
      "Overlap for C1QBP: 0 out of 12 in researchDB\n",
      "P-value for C1QBP: 1.0\n",
      "Overlap for RPL26: 4 out of 7 in researchDB\n",
      "P-value for RPL26: 4.910516979866623e-08\n",
      "Overlap for UBA3: 0 out of 5 in researchDB\n",
      "P-value for UBA3: 1.0\n",
      "Overlap for MED17: 1 out of 9 in researchDB\n",
      "P-value for MED17: 0.058361519318946814\n",
      "Overlap for SNRPG: 1 out of 6 in researchDB\n",
      "P-value for SNRPG: 0.02866347810549369\n",
      "Overlap for PRPF19: 1 out of 6 in researchDB\n",
      "P-value for PRPF19: 0.044798272846770126\n",
      "Overlap for NAF1: 0 out of 6 in researchDB\n",
      "P-value for NAF1: 1.0\n",
      "Overlap for TXNL4A: 2 out of 5 in researchDB\n",
      "P-value for TXNL4A: 0.000529721111247926\n",
      "Overlap for RPS19: 2 out of 11 in researchDB\n",
      "P-value for RPS19: 0.0009645903158693221\n",
      "Overlap for AATF: 1 out of 6 in researchDB\n",
      "P-value for AATF: 0.03791093137572726\n",
      "Overlap for PHB2: 2 out of 19 in researchDB\n",
      "P-value for PHB2: 0.009185594870537925\n",
      "Overlap for VPS28: 0 out of 9 in researchDB\n",
      "P-value for VPS28: 1.0\n",
      "Overlap for RPS28: 1 out of 7 in researchDB\n",
      "P-value for RPS28: 0.03336150600238275\n",
      "Overlap for EXOSC7: 2 out of 13 in researchDB\n",
      "P-value for EXOSC7: 0.002471059938780703\n",
      "Overlap for WDR12: 0 out of 5 in researchDB\n",
      "P-value for WDR12: 1.0\n",
      "Overlap for POLD1: 2 out of 12 in researchDB\n",
      "P-value for POLD1: 0.0035245791373806645\n",
      "Overlap for COPS3: 0 out of 6 in researchDB\n",
      "P-value for COPS3: 1.0\n",
      "Overlap for NVL: 3 out of 5 in researchDB\n",
      "P-value for NVL: 2.4319712243376875e-06\n",
      "Overlap for PAF1: 0 out of 13 in researchDB\n",
      "P-value for PAF1: 1.0\n",
      "Overlap for RAD51: 5 out of 28 in researchDB\n",
      "P-value for RAD51: 1.0101324774451766e-06\n",
      "Overlap for CCT5: 3 out of 10 in researchDB\n",
      "P-value for CCT5: 4.480528947125347e-05\n",
      "Overlap for MED1: 1 out of 35 in researchDB\n",
      "P-value for MED1: 0.1771416321151657\n",
      "Overlap for EXOC1: 0 out of 9 in researchDB\n",
      "P-value for EXOC1: 1.0\n",
      "Overlap for CCNK: 1 out of 10 in researchDB\n",
      "P-value for CCNK: 0.03587615783321504\n",
      "Overlap for RPL10: 2 out of 6 in researchDB\n",
      "P-value for RPL10: 0.000615303446286681\n",
      "Overlap for ACTR8: 5 out of 16 in researchDB\n",
      "P-value for ACTR8: 5.384424999101365e-08\n",
      "Overlap for SNRPD3: 2 out of 6 in researchDB\n",
      "P-value for SNRPD3: 0.0005149057259652727\n",
      "Overlap for LAMTOR4: 0 out of 5 in researchDB\n",
      "P-value for LAMTOR4: 1.0\n",
      "Overlap for TAF7: 1 out of 17 in researchDB\n",
      "P-value for TAF7: 0.10010863013860945\n",
      "Overlap for RFC4: 2 out of 5 in researchDB\n",
      "P-value for RFC4: 0.00016526499161894154\n",
      "Overlap for RPTOR: 1 out of 27 in researchDB\n",
      "P-value for RPTOR: 0.27462065680998554\n",
      "Overlap for RPL23: 1 out of 10 in researchDB\n",
      "P-value for RPL23: 0.041995859043389136\n",
      "Overlap for SMC1A: 4 out of 10 in researchDB\n",
      "P-value for SMC1A: 2.4186483678243333e-08\n",
      "Overlap for SF3A2: 1 out of 5 in researchDB\n",
      "P-value for SF3A2: 0.014186403172397456\n",
      "Overlap for EIF2B3: 1 out of 6 in researchDB\n",
      "P-value for EIF2B3: 0.03652850935432338\n",
      "Overlap for POLD3: 1 out of 8 in researchDB\n",
      "P-value for POLD3: 0.05868377093103718\n",
      "Overlap for INTS3: 1 out of 7 in researchDB\n",
      "P-value for INTS3: 0.04888428544377561\n",
      "Overlap for MNAT1: 2 out of 11 in researchDB\n",
      "P-value for MNAT1: 0.00139474279814442\n",
      "Overlap for GRB2: 2 out of 16 in researchDB\n",
      "P-value for GRB2: 0.0011432815400414342\n",
      "Overlap for INO80B: 3 out of 13 in researchDB\n",
      "P-value for INO80B: 0.00012280139562832808\n",
      "Overlap for CCT8: 2 out of 10 in researchDB\n",
      "P-value for CCT8: 0.0030106276764372197\n",
      "Overlap for BOP1: 1 out of 6 in researchDB\n",
      "P-value for BOP1: 0.03329641937620341\n",
      "Overlap for COPS8: 0 out of 6 in researchDB\n",
      "P-value for COPS8: 1.0\n",
      "Overlap for INO80: 1 out of 20 in researchDB\n",
      "P-value for INO80: 0.031251091286424816\n",
      "Overlap for UBE2N: 2 out of 16 in researchDB\n",
      "P-value for UBE2N: 0.007349623192881736\n",
      "Overlap for TAF12: 1 out of 17 in researchDB\n",
      "P-value for TAF12: 0.12302325506021794\n",
      "Overlap for EIF2B2: 2 out of 9 in researchDB\n",
      "P-value for EIF2B2: 0.0009190149284216131\n",
      "Overlap for EIF2B4: 0 out of 8 in researchDB\n",
      "P-value for EIF2B4: 1.0\n",
      "Overlap for TELO2: 1 out of 5 in researchDB\n",
      "P-value for TELO2: 0.037472286923089536\n",
      "Overlap for BCR: 2 out of 14 in researchDB\n",
      "P-value for BCR: 0.002795217526457602\n",
      "Overlap for RPL11: 4 out of 15 in researchDB\n",
      "P-value for RPL11: 4.932969797549702e-06\n",
      "Overlap for MED30: 0 out of 6 in researchDB\n",
      "P-value for MED30: 1.0\n",
      "Overlap for PPP1CA: 1 out of 18 in researchDB\n",
      "P-value for PPP1CA: 0.036471485600401074\n",
      "Overlap for MED10: 0 out of 5 in researchDB\n",
      "P-value for MED10: 1.0\n",
      "Overlap for EXOSC4: 2 out of 15 in researchDB\n",
      "P-value for EXOSC4: 0.007603333550124456\n",
      "Overlap for KANSL2: 0 out of 7 in researchDB\n",
      "P-value for KANSL2: 1.0\n",
      "Overlap for MED14: 0 out of 7 in researchDB\n",
      "P-value for MED14: 1.0\n",
      "Overlap for NCBP1: 5 out of 24 in researchDB\n",
      "P-value for NCBP1: 3.952539644538647e-07\n",
      "Overlap for EXOSC3: 1 out of 14 in researchDB\n",
      "P-value for EXOSC3: 0.08421492165545252\n",
      "Overlap for SUPT6H: 2 out of 9 in researchDB\n",
      "P-value for SUPT6H: 0.0011577054718808673\n",
      "Overlap for COPS5: 2 out of 12 in researchDB\n",
      "P-value for COPS5: 0.0016176646331365141\n",
      "Overlap for UPF1: 4 out of 11 in researchDB\n",
      "P-value for UPF1: 1.4839056717361259e-06\n",
      "Overlap for SDHA: 0 out of 6 in researchDB\n",
      "P-value for SDHA: 1.0\n",
      "Overlap for DAD1: 1 out of 6 in researchDB\n",
      "P-value for DAD1: 0.045713472662023885\n",
      "Overlap for TRRAP: 3 out of 18 in researchDB\n",
      "P-value for TRRAP: 0.0003614867115463507\n",
      "Overlap for CPSF6: 1 out of 8 in researchDB\n",
      "P-value for CPSF6: 0.03742397846149575\n",
      "Overlap for TAF6: 2 out of 14 in researchDB\n",
      "P-value for TAF6: 0.005215080250213804\n",
      "Overlap for GINS4: 1 out of 5 in researchDB\n",
      "P-value for GINS4: 0.028598634586795634\n",
      "Overlap for METTL3: 3 out of 13 in researchDB\n",
      "P-value for METTL3: 0.00012656348190672007\n",
      "Overlap for MED7: 0 out of 7 in researchDB\n",
      "P-value for MED7: 1.0\n",
      "Overlap for TTK: 3 out of 11 in researchDB\n",
      "P-value for TTK: 2.9651787336133862e-05\n",
      "Overlap for GTF2H1: 1 out of 7 in researchDB\n",
      "P-value for GTF2H1: 0.014338369809308697\n",
      "Overlap for RPS9: 0 out of 5 in researchDB\n",
      "P-value for RPS9: 1.0\n",
      "Overlap for CHMP6: 1 out of 25 in researchDB\n",
      "P-value for CHMP6: 0.08187584716296124\n",
      "Overlap for SEH1L: 2 out of 11 in researchDB\n",
      "P-value for SEH1L: 0.0006118761218801126\n",
      "Overlap for MED8: 0 out of 5 in researchDB\n",
      "P-value for MED8: 1.0\n",
      "Overlap for MTOR: 5 out of 63 in researchDB\n",
      "P-value for MTOR: 0.00022537551709343121\n",
      "Overlap for TAF2: 0 out of 12 in researchDB\n",
      "P-value for TAF2: 1.0\n",
      "Overlap for MED6: 0 out of 6 in researchDB\n",
      "P-value for MED6: 1.0\n",
      "Overlap for ENO1: 0 out of 6 in researchDB\n",
      "P-value for ENO1: 1.0\n",
      "Overlap for MED20: 1 out of 6 in researchDB\n",
      "P-value for MED20: 0.045255964139129114\n",
      "Overlap for ATP6V1B2: 0 out of 5 in researchDB\n",
      "P-value for ATP6V1B2: 1.0\n",
      "Overlap for GAB2: 0 out of 6 in researchDB\n",
      "P-value for GAB2: 1.0\n",
      "Overlap for MYBBP1A: 4 out of 11 in researchDB\n",
      "P-value for MYBBP1A: 1.2223399245264868e-06\n",
      "Overlap for KANSL3: 0 out of 8 in researchDB\n",
      "P-value for KANSL3: 1.0\n",
      "Overlap for MCM3: 2 out of 6 in researchDB\n",
      "P-value for MCM3: 0.00027671399181365935\n",
      "Overlap for RFC5: 3 out of 5 in researchDB\n",
      "P-value for RFC5: 9.646951764033886e-07\n",
      "Overlap for NUP62: 3 out of 12 in researchDB\n",
      "P-value for NUP62: 5.5790530145847814e-05\n",
      "Overlap for GLRX3: 1 out of 6 in researchDB\n",
      "P-value for GLRX3: 0.031908474742726685\n",
      "Overlap for TADA3: 1 out of 19 in researchDB\n",
      "P-value for TADA3: 0.08802395013148663\n",
      "Overlap for METTL14: 2 out of 9 in researchDB\n",
      "P-value for METTL14: 0.0012215579257578778\n",
      "Overlap for MCM5: 3 out of 6 in researchDB\n",
      "P-value for MCM5: 1.3088944199027693e-06\n",
      "Overlap for SMG5: 2 out of 5 in researchDB\n",
      "P-value for SMG5: 0.0009853709543044674\n",
      "Overlap for DKC1: 2 out of 12 in researchDB\n",
      "P-value for DKC1: 0.0033095825799992597\n",
      "Overlap for SKP2: 0 out of 6 in researchDB\n",
      "P-value for SKP2: 1.0\n",
      "Overlap for RPS3A: 3 out of 5 in researchDB\n",
      "P-value for RPS3A: 6.1926356356671515e-06\n",
      "Overlap for NCBP2: 2 out of 19 in researchDB\n",
      "P-value for NCBP2: 0.001539819932302026\n",
      "Overlap for CHMP3: 0 out of 26 in researchDB\n",
      "P-value for CHMP3: 1.0\n",
      "Overlap for KAT8: 0 out of 10 in researchDB\n",
      "P-value for KAT8: 1.0\n",
      "Overlap for HSPA9: 3 out of 8 in researchDB\n",
      "P-value for HSPA9: 2.5499492628815173e-05\n",
      "Overlap for POLD2: 1 out of 5 in researchDB\n",
      "P-value for POLD2: 0.028598634586795634\n",
      "Overlap for CASP8AP2: 0 out of 5 in researchDB\n",
      "P-value for CASP8AP2: 1.0\n",
      "Overlap for UBE2I: 2 out of 11 in researchDB\n",
      "P-value for UBE2I: 0.00045257392877956064\n",
      "Overlap for EXOSC8: 1 out of 12 in researchDB\n",
      "P-value for EXOSC8: 0.03556147437814442\n",
      "Overlap for EIF2B1: 1 out of 6 in researchDB\n",
      "P-value for EIF2B1: 0.040670816947906745\n",
      "Overlap for YEATS4: 1 out of 13 in researchDB\n",
      "P-value for YEATS4: 0.0736482173215087\n",
      "Overlap for RPL5: 4 out of 12 in researchDB\n",
      "P-value for RPL5: 1.8029092140895893e-07\n",
      "Overlap for KIF20A: 0 out of 6 in researchDB\n",
      "P-value for KIF20A: 1.0\n",
      "Overlap for EIF2B5: 1 out of 11 in researchDB\n",
      "P-value for EIF2B5: 0.05442069701975445\n",
      "Overlap for WDR61: 0 out of 6 in researchDB\n",
      "P-value for WDR61: 1.0\n",
      "Overlap for ATF5: 2 out of 6 in researchDB\n",
      "P-value for ATF5: 0.0003405680502523435\n",
      "Overlap for HINFP: 0 out of 7 in researchDB\n",
      "P-value for HINFP: 1.0\n",
      "Overlap for EXOSC6: 2 out of 11 in researchDB\n",
      "P-value for EXOSC6: 0.0018034167554763156\n",
      "Overlap for MAGOH: 2 out of 8 in researchDB\n",
      "P-value for MAGOH: 0.001194722130001048\n",
      "Overlap for MAD2L1: 5 out of 9 in researchDB\n",
      "P-value for MAD2L1: 2.5363032932725105e-09\n",
      "Overlap for HSPD1: 3 out of 18 in researchDB\n",
      "P-value for HSPD1: 0.00020118530173201323\n",
      "Overlap for XPO1: 2 out of 8 in researchDB\n",
      "P-value for XPO1: 0.0010311513996934835\n",
      "Overlap for EIF4G2: 1 out of 5 in researchDB\n",
      "P-value for EIF4G2: 0.030146560935722915\n",
      "Overlap for RPL7: 0 out of 5 in researchDB\n",
      "P-value for RPL7: 1.0\n",
      "Overlap for BRD8: 0 out of 10 in researchDB\n",
      "P-value for BRD8: 1.0\n",
      "Overlap for CLTC: 2 out of 13 in researchDB\n",
      "P-value for CLTC: 0.0017894707781930812\n",
      "Overlap for RBM4: 1 out of 8 in researchDB\n",
      "P-value for RBM4: 0.030656133269285887\n",
      "Overlap for DNAJA3: 2 out of 8 in researchDB\n",
      "P-value for DNAJA3: 0.0009038201917254252\n",
      "Overlap for SUPT16H: 3 out of 7 in researchDB\n",
      "P-value for SUPT16H: 1.5546548801441513e-05\n",
      "Overlap for MCM6: 3 out of 5 in researchDB\n",
      "P-value for MCM6: 5.374919107367473e-06\n",
      "Overlap for TSG101: 1 out of 24 in researchDB\n",
      "P-value for TSG101: 0.10133881660330367\n",
      "Overlap for TAF8: 0 out of 7 in researchDB\n",
      "P-value for TAF8: 1.0\n",
      "Overlap for MCM2: 4 out of 10 in researchDB\n",
      "P-value for MCM2: 2.618018946162168e-07\n",
      "Overlap for PRPF6: 2 out of 6 in researchDB\n",
      "P-value for PRPF6: 0.0005857221455239715\n",
      "Overlap for CDC73: 2 out of 20 in researchDB\n",
      "P-value for CDC73: 0.005107020540025154\n",
      "Overlap for EXOSC2: 2 out of 12 in researchDB\n",
      "P-value for EXOSC2: 0.002042317444150196\n",
      "Overlap for EP400: 1 out of 10 in researchDB\n",
      "P-value for EP400: 0.06984722382984401\n",
      "Overlap for CHAF1B: 3 out of 5 in researchDB\n",
      "P-value for CHAF1B: 3.6691488048685227e-07\n",
      "Overlap for RPS24: 2 out of 5 in researchDB\n",
      "P-value for RPS24: 0.0006620439075372245\n",
      "Overlap for EXOC4: 0 out of 9 in researchDB\n",
      "P-value for EXOC4: 1.0\n",
      "Overlap for MED12: 0 out of 8 in researchDB\n",
      "P-value for MED12: 1.0\n",
      "Overlap for ATP6V1H: 1 out of 9 in researchDB\n",
      "P-value for ATP6V1H: 0.06173790884147587\n",
      "Overlap for NUDT21: 1 out of 9 in researchDB\n",
      "P-value for NUDT21: 0.046116910946927304\n",
      "Overlap for CCT7: 2 out of 9 in researchDB\n",
      "P-value for CCT7: 0.001602393812389033\n",
      "Overlap for CUL1: 0 out of 7 in researchDB\n",
      "P-value for CUL1: 1.0\n",
      "Overlap for RRS1: 1 out of 6 in researchDB\n",
      "P-value for RRS1: 0.024943816481445866\n",
      "Overlap for YEATS2: 0 out of 13 in researchDB\n",
      "P-value for YEATS2: 1.0\n",
      "Overlap for AP2S1: 0 out of 7 in researchDB\n",
      "P-value for AP2S1: 1.0\n",
      "Overlap for NCKAP1: 1 out of 10 in researchDB\n",
      "P-value for NCKAP1: 0.09189207950588776\n",
      "Overlap for XRCC5: 2 out of 25 in researchDB\n",
      "P-value for XRCC5: 0.007925137444318536\n",
      "Overlap for DICER1: 1 out of 10 in researchDB\n",
      "P-value for DICER1: 0.02972129585261667\n",
      "Overlap for CCT4: 1 out of 11 in researchDB\n",
      "P-value for CCT4: 0.03433672065092231\n",
      "Overlap for SMC3: 5 out of 11 in researchDB\n",
      "P-value for SMC3: 1.2919646600125852e-09\n",
      "Overlap for ERCC3: 1 out of 16 in researchDB\n",
      "P-value for ERCC3: 0.07818335296874425\n",
      "Overlap for MCM4: 2 out of 6 in researchDB\n",
      "P-value for MCM4: 0.00038676590840153185\n",
      "Overlap for CPSF3: 0 out of 6 in researchDB\n",
      "P-value for CPSF3: 1.0\n",
      "Overlap for CCT2: 3 out of 12 in researchDB\n",
      "P-value for CCT2: 7.369246161295306e-05\n",
      "Overlap for TCP1: 2 out of 12 in researchDB\n",
      "P-value for TCP1: 0.002331790613204371\n",
      "Overlap for NUP54: 0 out of 6 in researchDB\n",
      "P-value for NUP54: 1.0\n",
      "Overlap for MIS12: 3 out of 5 in researchDB\n",
      "P-value for MIS12: 3.134875270533881e-06\n",
      "Overlap for CTR9: 0 out of 15 in researchDB\n",
      "P-value for CTR9: 1.0\n",
      "Overlap for ERCC2: 0 out of 14 in researchDB\n",
      "P-value for ERCC2: 1.0\n",
      "Overlap for TRMT112: 0 out of 5 in researchDB\n",
      "P-value for TRMT112: 1.0\n",
      "Overlap for PNPT1: 0 out of 7 in researchDB\n",
      "P-value for PNPT1: 1.0\n",
      "Overlap for MED21: 0 out of 6 in researchDB\n",
      "P-value for MED21: 1.0\n",
      "Overlap for RPS3: 7 out of 15 in researchDB\n",
      "P-value for RPS3: 4.927470217586745e-12\n"
     ]
    }
   ],
   "source": [
    "pvals = perturbation_signal_recovery(\n",
    "    pert_to_go,\n",
    "    model_pert_to_go,\n",
    "    list(filtered_go_terms.keys()),\n",
    "    list(adata.uns[\"Spectra_pert_labels\"]),\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "98cb9f87-58e7-46fb-ab9c-bfc04fb7209a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# save as csv for visualization\n",
    "pd.DataFrame.from_dict(data=pvals, orient=\"index\").to_csv(\n",
    "    \"figures/process_recovery_hypergeo_pvals/pertspectra_replogle_hypergeo_neighbors_recovery_pvalues.csv\",\n",
    "    header=False,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "39264204-40a8-46b4-b24f-5e88fb3280ce",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "pertspectra",
   "language": "python",
   "name": "pertspectra"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}