bacteriaDataStats.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "f3e35340-86af-4851-9a1d-bbae3b77221d",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "6ac6a4e2-2f20-4312-b689-c3811fc12e5c",
   "metadata": {},
   "outputs": [],
   "source": [
    "path = '/n/holyscratch01/wadduwage_lab/D2NN_QPM_classification/datasets/bacteria_np/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "06ffb329-60c2-41c1-a860-f560005821b7",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_dir = path"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "895654ad-5af2-45e5-97f5-e1f194e0a448",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train \n",
      "   0  12480 \n",
      "   1  11228 \n",
      "   2  13343 \n",
      "   3  15493 \n",
      "   4  24912 \n",
      "   5  11164 \n",
      "   6  16239 \n",
      "   7  23151 \n",
      "   8  15432 \n",
      "   9  17076 \n",
      "   10  10744 \n",
      "   11  7808 \n",
      "   12  27857 \n",
      "   13  21111 \n",
      "   14  18246 \n",
      "   15  29990 \n",
      "   16  24910 \n",
      "   17  17706 \n",
      "   18  18648 \n",
      "   19  16452 \n",
      "   20  21453 \n",
      "val \n",
      "   0  1560 \n",
      "   1  1403 \n",
      "   2  1668 \n",
      "   3  1937 \n",
      "   4  3114 \n",
      "   5  1396 \n",
      "   6  2030 \n",
      "   7  2894 \n",
      "   8  1929 \n",
      "   9  2135 \n",
      "   10  1343 \n",
      "   11  976 \n",
      "   12  3482 \n",
      "   13  2639 \n",
      "   14  2281 \n",
      "   15  4640 \n",
      "   16  3853 \n",
      "   17  2213 \n",
      "   18  2331 \n",
      "   19  2056 \n",
      "   20  2682 \n",
      "test \n",
      "   0  1561 \n",
      "   1  1404 \n",
      "   2  1668 \n",
      "   3  1937 \n",
      "   4  3114 \n",
      "   5  1396 \n",
      "   6  2030 \n",
      "   7  2894 \n",
      "   8  1929 \n",
      "   9  2135 \n",
      "   10  1343 \n",
      "   11  976 \n",
      "   12  3483 \n",
      "   13  2639 \n",
      "   14  2281 \n",
      "   15  4861 \n",
      "   16  3854 \n",
      "   17  2214 \n",
      "   18  2332 \n",
      "   19  2057 \n",
      "   20  2682 \n"
     ]
    }
   ],
   "source": [
    "lis = {}\n",
    "\n",
    "for k in ['train','val','test']:\n",
    "    \n",
    "    print(f\"{k} \")\n",
    "    folders = sorted(os.listdir(path + k), key=lambda x: int(x))\n",
    "    \n",
    "    # folders = os.listdir(path + k)\n",
    "    \n",
    "    lis[k] = []\n",
    "    for folder in folders:\n",
    "        print(f\"   {folder}\", end=\" \")\n",
    "        files = os.listdir(path + '/' + k + '/' + folder)\n",
    "        #count += len(files)\n",
    "\n",
    "\n",
    "        \n",
    "        lis[k].append(len(files))\n",
    "        \n",
    "        print(f\" {len(files)} \")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "4b5f2eb4-fde2-46b2-bd7c-9f688cb4b242",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "56316.45"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "375443*0.15"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "fd0d80cf-0a6d-4cc3-bf53-16bbffd9e46f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7318.5"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "48790*0.15"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "050b1605-d532-496f-8df6-351100a02774",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[12480,\n",
       " 11228,\n",
       " 13343,\n",
       " 15493,\n",
       " 24912,\n",
       " 11164,\n",
       " 16239,\n",
       " 23151,\n",
       " 15432,\n",
       " 17076,\n",
       " 10744,\n",
       " 7808,\n",
       " 27857,\n",
       " 21111,\n",
       " 18246,\n",
       " 29990,\n",
       " 24910,\n",
       " 17706,\n",
       " 18648,\n",
       " 16452,\n",
       " 21453]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lis['train']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "784d6bcd-7097-40d8-a901-30d44e61536f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2ba5274859e8>]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAD4CAYAAADsKpHdAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deXzc1Xno/88ZraN9pJFtWZItL/KOdxs72CwBgsEQkjYLNA380rQkhKRJk9yUtPeG9N7yayklaZMWmlByQwIl0CwFbEhCsxOwjRcs20jeJFmSNVpmtI+W0cyc+8d8Rx7LI2kkzXxn0fN+vfTy+DvznTkay3rme57zPEdprRFCCCEs8R6AEEKIxCABQQghBCABQQghhEECghBCCEACghBCCEN6vAcwU3a7XVdVVcV7GEIIkVSOHDni1FqXhrsvaQNCVVUVhw8fjvcwhBAiqSilLkx0n0wZCSGEACQgCCGEMEhAEEIIAUhAEEIIYZCAIIQQAoggICilspVSh5RSx5VSp5RSf2McL1ZKvaaUOmv8aQs558tKqXNKqdNKqVtCjm9RSp0w7vuGUkoZx7OUUs8bxw8qpaqi/60KIYSYTCRXCCPAu7XWG4CNwB6l1A7gQeAXWutq4BfG31FKrQHuAtYCe4DHlVJpxnM9AdwHVBtfe4zjHwe6tdbLga8Dj0ThexNCCDENUwYEHTBg/DXD+NLAncDTxvGngfcZt+8EfqC1HtFaNwDngO1KqTKgQGv9pg703P7euHOCz/VD4Mbg1YMQInX5/ZrnDjUx5PHFeyiCCHMISqk0pdTbQAfwmtb6IDBfa+0AMP6cZzy8HGgOOb3FOFZu3B5//LJztNZeoBcoCTOO+5RSh5VShzs7OyP7DoUQCetYczdf/vEJnj04Ya2UMFFEAUFr7dNabwQqCHzaXzfJw8N9steTHJ/snPHj+LbWeqvWemtpadjKayFEEjnf6QZg/wlHnEciYJqrjLTWPcCvCcz9txvTQBh/dhgPawEqQ06rAFqN4xVhjl92jlIqHSgEuqYzNiFE8ml0BgLCsaYeWroH4zwaEckqo1KlVJFx2wrcBNQBLwH3Gg+7F3jRuP0ScJexcmgJgeTxIWNaqV8ptcPID9wz7pzgc30A+KWWvT2FSHmNLjcF2YGWaq+eaIvzaEQkze3KgKeNlUIW4AWt9T6l1JvAC0qpjwNNwAcBtNanlFIvAO8AXuABrXUwY3Q/8F3ACrxqfAE8BXxfKXWOwJXBXdH45oQQia3BOcjWqmI6+0fYd8LBn127NN5DmtOmDAha6xpgU5jjLuDGCc55GHg4zPHDwBX5B631MEZAEULMDVprGp1udi4tYfuSYv7+1TqauwapLM6J99DmLKlUFkLERXvfCEOjPpbYc9h7VRkgyeV4k4AghIiLBiOhXGXPpbI4hw0VheyvkYAQTxIQhBBx0egKBIQl9lwA9q4v48TFXi4Yx4X5JCAIIeKi0ekmM93CwkIrALfJtFHcSUAQQsRFvdPN4uIcLJZAXWqFLYeNlUUybRRHEhCEEHHR6HRTZUwXBd2+voxTrX1j+QVhLgkIQgjT+f2aC12DLB0XEG41po1ekWmjuJCAIIQwXWvvEB6v/4orhPIiK5sXFbFPpo3iQgKCEMJ0Y0tOS3KvuO/29QupdfRxvnPgivtEbElAEEKYLtjUbon9yoAQXG30ilwlmE4CghDCdA3OQawZacwvyLrivgWF2Wyrssny0ziQgCCEMF2jK7DCaKKNEfdeVUZdWz/nOvpNHtncJgFBiBRxsN7F6bbk+AXa6HSzxD5xE7tbrypDKdhfIy2xzSQBQYgU8Zc/quGhl07GexhT8vr8NHUNhk0oB80vyGZbVTH7T7RO+BgRfRIQhEgRnf0jHG3qYXg0sTesb+kewuvXVyw5He/29WWcaR/gTHtyXPWkAgkIQqSAIY8Pt8eHx+vnaFN3vIczqQajed34orTx9qxbgFJITYKJJCAIkQJc7pGx2wfqE3s78saQtteTmZefzdVLitlf04rsqGsOCQhCpADXgAcApeDAeVecRzO5Bqeb/Kx0SnIzp3zs3vULOd/p5rRMG5lCAoIQKaDLHQgI26uKOdbczZAncfMIDc7Jl5yG2rN2ARaFdEA1iQQEIVKAcyAwZXT7hoWM+jRHLiRuHiFYgxCJ0vwsdiwtYX+NQ6aNTCABQYgU4DKuEG5ZO580i+LNemecRxTeiNfHxe6hsC0rJrJ3fRn1Tje1Dpk2ijUJCEKkANfACNkZFkrzslhfUZiwieXmrkH8mkmL0sbbs3YBaRYlNQkmkIAgRApwDXgoyc1CKcXOpSUcb+7BPeKN97Cu0OAcBMJ3OZ1ISV4W71om00ZmkIAgRApwuT3Y8wKrdnYuK8Hr1xxOwDzCZF1OJ7P3qjIaXYOcau2LxbCEQQKCECnA5R6h2FjGuXVxMRlpijcTcPlpg8uNLSeDopypl5yGumVs2khWG8WSBAQhUoBrwENJXqCVtDUzjY2VRbxZn3gBIdw+ypGw5WZyzXK7TBvFmAQEIZKc1toICJc+de9YWsLJi730D4/GcWRXanC6WTKN/EGo268qo6lrkJMXZdooViQgCJHk+ke8eHx+7LmXNpvZubQEn19zuDFx8ghDHh+O3uEZXSEAvGftfNItin2y2ihmJCAIkeS6jLYVoVcImxfbyEyzJNS00YWumSWUg4pyMtlVLdNGsSQBQYgkF2xsVxzSGyg7I41Ni4oSKrE80xVGofZeVUZL9xA1Lb3RGpYIIQFBiCTnNK4Q7HmX70+8c1kJp1p76R1KjDxCfYRdTifznjULyEiT1UaxIgFBiCTnCjNlBIHEsl/DWw2JUbXc6HRjz8siLyt9xs9RmJPB7upSmTaKEQkIQiQ518CVU0YAmxYVkZWeOHmERufgtFpWTGTvVWVc7BniWHNPFEYlQklAECLJudwe8rPTyUpPu+x4VnoaWxbbEiaP0OByzyp/EHTTmvlkplmkJXYMSEAQIsm53J4JN5vZubSE2rY+egY9Jo/qcgMjXjr7R2aVPwgqtGZw7Qo7r5xw4PfLtFE0SUAQIsm5BkbGqpTH27msBK3jv63m2AqjGRaljbd3fRmO3mGONSdOnUUqmDIgKKUqlVK/UkrVKqVOKaU+axz/qlLqolLqbePrtpBzvqyUOqeUOq2UuiXk+Bal1Anjvm8oY8skpVSWUup54/hBpVRV9L9VIVJToNNp+CuE9RVFWDPSOBDnPEJDFFYYhbpp9Xwy0y3sk2mjqIrkCsELfEFrvRrYATyglFpj3Pd1rfVG4+sVAOO+u4C1wB7gcaVUcHLzCeA+oNr42mMc/zjQrbVeDnwdeGT235oQc4PLPfEVQma6ha1VtrgHhOAVwnTaXk8mPzuD61eUyrRRlE0ZELTWDq31UeN2P1ALlE9yyp3AD7TWI1rrBuAcsF0pVQYUaK3f1IH1Yt8D3hdyztPG7R8CN6pINlwVERn0eOnoG473MEQM+P2arpDW1+HsWFpCXVv/2GqkeGhwuSkrzMaamTb1gyO0d30Z7X0jHGmSaaNomVYOwZjK2QQcNA59WilVo5T6jlLKZhwrB5pDTmsxjpUbt8cfv+wcrbUX6AVKwrz+fUqpw0qpw52dndMZ+pz2jz87w/sffyPewxAx0DM0il9fueQ01M5lgf9KB+NYj9DgdEft6iDoRmPa6Gcn26L6vHNZxAFBKZUH/Aj4nNa6j8D0zzJgI+AAHgs+NMzpepLjk51z+QGtv6213qq13lpaWhrp0Oe8Y83dXOwZSsgdtMTsBD/1TzRlBHBVeSE5mWlxXX4607bXk8nLSmdjZRFvNSZG4V0qiCggKKUyCASDZ7XWPwbQWrdrrX1aaz/wJLDdeHgLUBlyegXQahyvCHP8snOUUulAISD/ylHg92tOtwU2J7/YMxTn0YhoG2tbMckVQkaahW1VxXErUOsdHKV7cDQqRWnjbauycbK1Tz7sREkkq4wU8BRQq7X+WsjxspCHvR84adx+CbjLWDm0hEDy+JDW2gH0K6V2GM95D/BiyDn3Grc/APxSS116VDR1DTLo8QHQ0j0Y59GIaAs2tpvsCgEC00bnOgbo7Dc/j9DgCja1y4v6c2+rKsbn1xxrkqrlaIjkCuEa4KPAu8ctMf0HYwlpDXAD8BcAWutTwAvAO8BPgQe01j7jue4H/p1Aovk88Kpx/CmgRCl1Dvg88GBUvjtBrePSZiIt3XKFkGom6mM03s6lgTxCPFYbXepyGv0rhC2LbVgUMm0UJVN2mdJav074Of5XJjnnYeDhMMcPA+vCHB8GPjjVWMT01bb1Y1GQbrFIQEhBLrcHpcA2xR7FaxcWkJ+Vzpv1Lu7YsNCk0QXUO91YFFQWRz8g5GdnsGpBgQSEKJFK5RRX6+hjiT2XCpuVixIQUo5rYARbTiZplslXaaenWdi2pJgDcUgsNzrdLCyyXtFrKVq2LynmWFMPoz5/TJ5/LpGAkOLq2vpYXVZAuc0qOYQUNFmV8ng7l5ZQ73TTbnJNSmOUmtpNZFtVMUOjPk61yl7LsyUBIYX1D4/S3DXE6rICKmw5MmWUggJVyhEGhGXm5xG01jQ4Yx0QAiVQibLvQzKTgJDCgstNV5flU2Gz4nJ7GPTI8rxU4hrwTLnCKGh1WQEF2emm1iO43B76h71RL0oLNa8gm8UlORySPMKsSUBIYbVGQFi1oIAKmxVA8ggpZrLW1+OlWRRXLy0xtR4hGvsoR2JbVTGHG7tkF7VZkoCQwmodfRRkp1NWmD0WEFqkOC1leLx+eodGKcmN7AoBAn2NLrgGaTXp5yDaXU4nsq3KRvfgKOc7B2L6OqlOAkIKq3MEEspKKSpsgSV/kkdIHd2DkdUghArWI5g1bdTocpNuUWMfSGJlW1UxAIcapNHdbEhASFF+v6aurZ/VZQUAlOZlkZlmkZVGKcRp9DGarNPpeKsW5GPLyTAtsdzgdFNZnENGWmx/1Syx52LPy+Sw5BFmRQJCimruDrSsWF2WD4DFooylp3KFkCouVSlHPmVksSiuXmJeHqHBOUhVSfQL0sZTSrGtqlgSy7MkASFF1TouJZSDKpIsILx0vJUz7f3xHkbC6nIbASHCpHLQzmUltHQP0dwV26tFrTUXXNHvcjqRrVXFtHQP4ehNnp/xRCMBIUXVOvqwKFgxP3/sWHlR8lQrD4x4+Yvn3+aff3E23kNJWMEpo+kklSGQWAZifpXQ0T/CoMfHUpMCwnYjj/BWo+QRZkoCQoqqa+ujyp572Q5VFTYrzoERhkd9k5yZGI5c6Mbn17wtXSwn5HJ7SLcoCqxTtiS7zIr5eZTkZsa8jYVZK4yCVpflk5uZJgVqsyABIUXVOvpZHTJdBCTVSqNg0vNiz5DprRaShWsgUKU83d1mlVLsWFrCgXpXTNftN0R5H+WppKdZ2LzYJo3uZkECQgoaGPHS1DU4llAOGqtFSIKVRgfrXeRnBz75HpM9c8MK9DGa3nRR0I5lJbT2DtMUwzxCo9NNZpqFhUWxXXIaaltVMafb++kdHDXtNVOJBIQUdLot0ORrVZJeIQx6vNS09PKhrZVkplk4KtNGYbncnmnVIIQyox6hwelmUUnOlJ1Yo2lbVTFaw5EmuUqYCQkIKWhshdG4K4R5+VlkpKmE30rz6IUevH7N7mo768oLOHpBrhDCcblHpr3CKGhZaS6l+VkxTSzHustpOBsri8hIU1KgNkMSEFJQXVsf+dnplI+7VLdYFAuLEn/p6cEGF2kWxdaqYjYtsnHiYi8er/S6H286je3GC+YR3jwfmzyC369pdA2aHhCsmWmsKy+UArUZkoCQgoIJ5XDJxook2BfhQL2LdQsLyMtKZ/MiGyNe/2VbgYrAtNqgxzfjKSMITBt19I+MJX+jqbV3CI/Xb1pCOdT2qmJqWnqTYjVdopGAkGL8fs3ptv4rEspBFUWJvS/C8KiP4829Y2vlNy0qAuCoJJYvE6xSts8wqQyX9keIxbRRozPwoaMqBvsoT2VrVTEen5/jzZJ7mi4JCCmmpXuIgREvq8oKwt5fYbPS2Z+4tQhHm7rx+PxcvTRQZLSwyMqCgmyOSWL5MmNVyrO4QqgqyWFBQXZMEssNrsBVx1J7XtSfeypbFwc2zDksuadpk4CQYmqNFUarJwgI5cbSU7PaH0/XwfouLCrwKS9o8+IiuUIYx+UOVCkXzzCpDME8QjEH6qO/j0Cj0401I435BTO/gpkpW24mK+bncUgK1KZNAkKKqXX0oVSgGjWcRF96eqDexZqFBRRkZ4wd21Rpo6V7iI5+KVALcganjGaYVA7auawE58AI5zqiu49Ag9PN4pKcaRfNRcvWqmKOGtXuInISEFJMnaOfqpJccjLDtzO4VJyWeAFheNTHseYediwpuez45sWBPIJMG11yqdPpzK8QAHYutQPR32e5Mcb7KE9le1Ux/SNe6tpkMcJ0SEBIMXVtfRMmlAHmF2STblEJudLoeHMPHq+fq5deHhDWLiwkI03JtFEI18AI1oy0CQN/pCqLrZQXWaOaWPb6/DR1DZrWwyicbUuMRncybTQtEhBSiHvEy4WuwSsqlEOlJXAtwsGGLpS61LUyKDsjjTULC+UKIUTXLKqUQ13qa9SFP0rTKxd7hvD6dVyvEMqLAoEu1Tqfaq3511+di9lWoRIQUsjp9n60njihHFReZE3IauUD9S5WLSigMCfjivs2LyqipqWHUZ8UqAE43Z4ZVymPt2NpMV1uD2c6orP3RL1R1xDPgACwtSrQ6C6WDfzMduJiL4/+7HTMCu8kIKSQYPHWqgUTTxlBYhanebx+jjZ1s2Npcdj7Ny2yMTzqp84hG+ZAsNNpdFbwjNUjRGn5aaPJXU4nsq2qmI7+kZg28DPbvhoH6RbFLWsXxOT5JSCkkDpHP/lZ6VNuaF5hy6G9b4QRb+LUItS09DA86ufqcQnloM1Ggdqx5tSaApipQKfT6FwhVNhyqCy2Ri2x3Oh0k5eVPq29nmNhu5FHSJXlp36/Zt/xVq5dUUpRTmzeWwkIKaSurY9VZflTLvWrGKtFSJxlnAeN/7TB/8TjlRdZmZefJY3uCMwju9zRu0IA2F1dym/OdEblyrHB6GEUryWnQctL8yi0ZqTM/gjHmrtp7R3m9vVlMXsNCQgpQmtNnaN/0oRyUCLui3Cg3sXK+fkTFloppdi0qIhj0o6A/hEvoz4d1U/gD9ywHIXib/fVzvq5Gp3m7aM8GYtFsa3KxuEUSSy/fNxBZrqFm9fMj9lrSEBIES3dQ/SPeKdMKMOlauVE2V951OfnyIWJ8wdBmxfZuOAaHNtLeK4K1iDMpkp5vPIiK59+93J+eqqN357pnPHzeLx+WroHWVJifg+jcLZVFVPvdNPZn9w/Mz6/Zv8JBzesLCU/+8pFF9EiASFFjCWUJ6lBCFpQkE2aRSXM0tMTF3sZ9PiuqD8Yb7PRo2auLz91GQExmlNGAH+6ewlVJTl89eVTM2433tQ1iF+bt4/yVIItUJK9Hfahhi46+0e4ff3CmL6OBIQUUdfWj1Kwcv7UASE9zUJZYXbCTBkdrJ88fxB0VXkh6RY157fUDLatiFZSOSgrPY2H3ruW+k43T73eMKPnaEyQJadBV5UXkp1hSfp6hJdrWrFmpHHj6nkxfR0JCCmi1tHH4uIccrMiq1wNLD1NjCuEA/Uuls/Lm7IvT6BArWDOVywHG9vNto9RODesnMfNa+bzzV+exdE7/Z+PRldiBYTMdAsbK4uSOrHs9fn56ck2blw9b9aV6VORgJAi6toiSygHVdgSY18Er8/P4cauKfMHQZsqizje3It3DheodcUghxDqK7evwefXPLx/+gnmeqebopyMmC2LnIltVcWcau1lYMQb76HMyBvnXXS5PdyxIbbTRRBBQFBKVSqlfqWUqlVKnVJKfdY4XqyUek0pddb40xZyzpeVUueUUqeVUreEHN+ilDph3PcNZaxLU0plKaWeN44fVEpVRf9bTV2DHi+NLndECeWg8iIr7f3Dcd+a8lRrH26Pb8L6g/E2L7YxNOrjdPvcLVBzuT3kZ6eTmR6bz3OVxTncf/0y9tU4eOO8c1rnNjrdcS9IG29bVTF+TdJONb58vJX8rHSuW1Ea89eK5CfKC3xBa70a2AE8oJRaAzwI/EJrXQ38wvg7xn13AWuBPcDjSqk047meAO4Dqo2vPcbxjwPdWuvlwNeBR6Lwvc0Zp9sCLSsiSSgHVdisaM2MpgWi6WBDoBjq6givEDYvCnzuODqHE8vOgZGYTBeF+uR1y6gstvLQi6em1S4k3l1Ow9m82IZFJWejuxGvj5+dauPmNfPJzkib+oRZmjIgaK0dWuujxu1+oBYoB+4EnjYe9jTwPuP2ncAPtNYjWusG4BywXSlVBhRord/UgeYi3xt3TvC5fgjcqOJd1ZJE6toCn5ZXT3PKCOLfBvtAfRdL7bnMy8+O6PEVNiv2vMyk/bQXDdGsUp5IdkYaX7l9LWc7Bnj6jcaIzhke9dHaO5xwASEvK501Cws4lIR5hN+dcdI37DVlugimmUMwpnI2AQeB+VprBwSCBhBMf5cDzSGntRjHyo3b449fdo7W2gv0AlfMISil7lNKHVZKHe7snPla6VRT6+gjL4KWFaESoTjN59e81dA15XLTUIECNducXnoaqFKO/Rz9TavnccPKUv7pv8/S0Td1VXswoZwoS05Dbasq5m2jvXoy2VfTSqE1g2uW2015vYgDglIqD/gR8Dmt9WS7ToT7ZK8nOT7ZOZcf0PrbWuutWuutpaWxn09LFnWOflYuyMdiifyiqqww/rUItY4++ke8ESeUgzYvstHgdI/tKzzXBFpfx35rSqUUD92xFo/Xz9+9Wjfl48eWnCZYDgECLdWHR/2cbO2N91AiNjzq47V32rl13YKY5YvGi+hVlFIZBILBs1rrHxuH241pIIw/O4zjLUBlyOkVQKtxvCLM8cvOUUqlA4VA8l3fxYHWmtopNsUJJz3NwoKC7LhWKwebqUWaUA7aZDS6e3sONrrz+XUgIMR4yiioyp7Lfdcu5SfHLk7ZJK7BOWickxhVyqGCBWrJlEf4VV0Hbo8v5sVooSJZZaSAp4BarfXXQu56CbjXuH0v8GLI8buMlUNLCCSPDxnTSv1KqR3Gc94z7pzgc30A+KVOpSbmMXSxZ4j+Ye+0lpwGlce5FuFAfReLS3JYUBhZ/iBofUUhaRbF0Qtzb9qoZ9CDX0e/KG0yD9ywnPIiK1958eSky30bnW7seVkxba0wU6X5WSyx5yZVgdq+Ggf2vMxpX0HPRiRXCNcAHwXerZR62/i6Dfh74Gal1FngZuPvaK1PAS8A7wA/BR7QWgf7LN8P/DuBRPN54FXj+FNAiVLqHPB5jBVLYmrB/QGms+Q0KJ77Ivj9mrcau67YPzkSOZnprC7Ln5MFai53cC/l2E8ZBVkz0/ife1dT19bPMwcuTPi4BpebJQl4dRC0rcrG4QvR2xkultwjXn5R186t68pITzOvXGzKsjet9euEn+MHuHGCcx4GHg5z/DCwLszxYeCDU41FXCnYw2jlFJvihFNhy6Gt7yIer9+0OcqgurZ+eodGI15uOt6mShs/PtqCz69Jm0buJNk5x/oYmVv4tWfdAnZX23nstTPcvmFh2GWvDU4315uwVn6mtlUV88LhFs51DrAighYv8fTfte0Mj/pj2uo6HKlUTnJ1bf0sKs4hL8KWFaEqbFb8Gtp6zd8X4VL9wfSvEAA2Ly7C7fFxZo4VqAUT6bGuQxgvmGAeHvXxSJgE88CIl87+kYRcYRS0LZhHSILlp/tqHMwvyBobs1kkICS5Wsf0E8pBFUXG0tMe86eNDtS7qLAFNkKfiU2Vc7PzqStGje0isXxeHn+yawn/eaSFI+M2Kkq0pnbhLC7JoTQ/K+ETy71Do/zmdCd7r1o4rZWD0SABIYkNeXw0uNwzSihD/IrT/H7NoYYudszw6gAC/7mLczPnXB7BNTCCUsStV9Cfv7ua+QVZPPTSSXwhc/GJ1tQuHKUU26uKEz6x/No77Xh8fu7YYO50EUhASGqn2wMtK2aSUAZYUJiNRZkfEM52DNA9OMrVU7S7noxSis2LiuZcQHC6PRTnZMYtb5Kblc5f713DyYt9PHeoaex4Q6dRlJaANQihtlbZuNgzxMWe+Dd2nMjLx1upsFnZWFlk+mtLQEhidUZCeaZTRpnpgVoEs1caBfMHs7lCANi0yEZ9p5uewblToOYaMKdKeTJ3rC9jx9JiHv3Z6bGcRoPLzYKCbKyZse+3MxvbEnzDnG63h9+fc7J3fVlc9qSWgJDE6tr6yc1Mo9I286V+8WiDfaDexcLC7Gm12ggnWKA2l/ZZDvQxMjehPJ5Sir957zoGRrw8+rPTQHAf5cRdchq0uqyAvKz0KYvspnKuY4DfzGKr0Yn89FQbXr/mDhOL0UJJQEhi7zj6pt2yYrxym9XUamWtL+UPZvsJaENFERYFxy7MnWmjQNuK+O81sHJBPv/fu6r4wVtN1LT00OgaZIk9L97DmlKaRbF5sY3DM8wjdPQP81c/OcEt//Rb7v3OIX56si2q43v5eCtL7LmsXTizaeDZkoCQpLTW1Dn6WDXD/EFQhc1KW9+waRvOnO8cwDngmXH9QajcrHRWLiiYU1cIzoGRuKwwCudzN1VTkpvFl35YQ5fbk9BFaaG2V9k43d4/ranGQY+Xf/7vs1z/6K954a1mPrpjMRsqi/jifx6nvnMgKuPq7B/hQL2LO+I0XQQSEJJWa+8wfcPeGSeUgypsVnx+jcOkWoQDxv7J0+1fNJHNi4p4u6knKapPZ8vj9dM37DW1Snky+dkZ/NVtq8baryd6QjnoUh5h6qsEn1/zg0NNXP/or/n6f5/huhWlvPb56/jqe9fyxEc2k5lu4ZPPHGHQM/vd2F496cCv4XaTWl2HIwEhSY0llGdQoRzK7KWnB+pdzC/IYnFJdD5Nblpko3/Ey7kofUpLZF1jbSsS4woB4P2bytlWFagJSeQlp6E2VBaRkaZ468LEeQStNb863cFt//w7HvzxCSpsVn50/06e+OMtY9/nwiIr37hrE+c6BnjwRyeYbfu1l4+3sonPIsYAAB8TSURBVHJ+flyrqCUgJKngp7KZtKwIZea+CFprDkYpfxC02UgsH50DeYSxthVxTiqHUkrx6Ac28MANy1hWmvg5BAhs/rO+omjCArWTF3v546cO8rH/+xbDXh+Pf2QzP7r/XWxZfOU0565qO194z0peOt4a8UZC4Th6h3irsdv0VhXjTb/fgUgI7zj6qCy2zrqzZFmhFaUwZV12g9NNZ/9I1KaLIPCptCgng6NN3dy1fVHUnjcRXWpbkThXCBBokf0/blkV72FMy7aqYp56vZ7hUd/Y1pStPUP8489P85NjFymyZvDQHWv4yNWLp+zzdf91yzjW1M3f7q/lqorCsIFjKvtrHEB8p4tArhCSVp2jb8YVyqEy0y3Mz882ZcrooPGJLBoJ5SClFJsqi+ZECwuXO3CFUJwgSeVktn2JjVGf5u3mHvqGR3nkp3Xc8I+/Zl+Ng09cu4xf/48b+Ng1SyJq+mixKB770EbKbVY+9exROvtHpj2el2scrCsviPu0mwSEJDQ86qPB6Z51QjnIrDbYB+pd2POyWBrlH/rNi2yc7Rigd2g0qs+baMb6GCVIUjmZbVlUjFKMrRx64tfnue2qMn75het48NZVFFqnd+VdaM3giY9soXdolM88d3Raq/aauwY53txj6kY4E5GAkITOtPfj17NPKAdVmLBRjtaag/Vd7FhaHPUldZsXB5Kab6f48lPngIeMNEVBtsz0zlZhTgYr5+fzZr2LVQvy2feZXXz9wxvHFlnMxJqFBTz8vqs4UN/Foz8/HfF5L9cENo7ce1V88wcgOYSkFNwDYbY1CEEVthxernHg9fljthlHU9cgbX3DM253PZn1FYUoBceaurkugfvxz5ZrYISS3Ky4rVFPNY99aAO9g6PsXBa9RQ5/uKWCo03dfOs39WyqtLFn3YIpz9l33MGmRUVUFse/jkOuEJJQraMfa0Yai6P0A1Ru1CK0z2DuM1IHjfqDHbNoaDeR/OzAp72jKZ5HSJQq5VSxdmEh71puj3qA/codayIuWjvfOcA7jr6EmC4CCQhJqa5t9i0rQo0tPe2KXR7hQL2LktxMls+LzdLETYuKeLupO6UL1JxujySUk0BWehqPf2QzGWlqyqK1fccdKJUY00UgASHpaK2pdfRHLaEM5hSnHWzo4uoY5A+CNi2y0Tfspd6ZugVqroER03dKEzNTXmTlG3dv4uwkRWtaa16uaWVbVTELCrPjMMorSUBIMm19w/QOjc645XU4C4sCP4yxCgjNXYNc7BmKav3BeJsXBRLLRy/Mftqos39k1t0wYyHQ6VSuEJLF7upSvnDzigmL1k6393OuY4A74lx7EEoCQpIZSyhHoQYhKCs9jfkFWTFbehqL+oPxltpzKchO51jz7CqWL7jcvO9ff8+Hv/3mjNaTx8qgx8vQqE+WnCaZT12/nJtWz+Nv99dyZFyrjH3HHVgU3BpB4tksEhCSTK0j0LJiVRSvECBwiRurauUD9S6KcjJYMS92PVosFsWmRbZZXSGc6xjgQ996E5d7BK3hjfPOKI5wdi7VIMgVQjIJFq0tLLq8aC04XXTNcntCTQNKQEgytY4+yousFMyyZcV4sdwo52CDi6uXFMd8w/DNi2yc6einb3j6BWq1jj4+/K038fnhx/dfQ6E1g9+dTaCAEGxsJ1NGSafQmsETf7yZnsFLRWsnL/ZxwTUY995F40lASDJ1bdFNKAdV2Ky09gxdtnF6NLT2DNHcFdv8QdCmRUVoDTXNvdM6r6alh7ufPEBGmoXnP7GDNQsLuGZ5Ca+fdc66g2W0uIKN7RLo06SI3NqFhTz8/ktFa/tqWkm3KG5ZmzjTRSABIakMj/qo7xyIakI5qMKWg9evae+L7r4Iwf2TY5k/CNq4qAil4GhT5HmEIxe6+MiTB8nLSueFT+wc69i5a3kpbX3DnE+QttpjU0ZyhZC0PrClgj+6ehHf+k09zx5s4toVpRTlJNa/pwSEJHK2fSDQsiJGVwgQ/ZVGB853UZCdHtUk+EQKsjNYXprHsQgDwhvnnXz0qUPY87N44RM7WRSyR8PuajtAwkwbOd3BK4TE+gUipuehO9awoaKQgRFvwk0XgQSEpFLbFlxhFP0rhHIjIFzsie5Ko4MNLrYvKSYtxvmDoM2LbBxr7plyqufXpzv42P99iwqblec/sYOFRdbL7q8szmFxSQ6vJ0hA6BrwkJOZRk6mdJtJZlnpaXzro1v5ws0ruC1BitFCSUBIIrWOPrIzLCyOwVaF5UXBauXoXSE4eododA2akj8I2ry4iJ7BUeqd7gkf87NTbfzZ9w6zrDSPH9y3k3n54YuCdi23c6DexahJ+01PxiVVyiljQWE2n7mxemwfhkQiASGJ1Dn6WbmgICaftrMz0ijNz4rqlNFPT7YBcMOqeVF7zqlsMgrUJtof4eXjrXzq2aOsXVjIc3+2Y9Jfsrur7bg9voTYa8E5MCIJZRFzEhCShNaa2ra+qLW8DqfCZqUlilNG+2scrFqQH7P+ReEsL80jPzs9bGL5h0da+OwPjrFlsY1n/vRqCnMmX7q7c5kdi4LXz3bGargRcw14sMsVgogxCQhJor1vhJ7B0ZgklIOiWYvQ2jPE4Qvm7xFrsSg2htlB7ZkDF/jifx7nmuV2nv7YdvKypp6LL7RmsL6iiN+di38eweUekYSyiDkJCEkilgnloPKiQC1CNDqGvnIisEdsPBJnmxbZON3Wx8BIoMvkU6838D//6yQ3rprHk/dsxZoZ+dzt7mo7x5t74robm9baaH0tU0YitiQgJIlY9DAar8JmZdSn6YhCD5/9JxysKStgaal500VBmxcV4ddQ09zDv/7qHP9n3zvcum4BT/zxlmkn8nYtt+PX8OZ5V4xGO7W+YS+jPi01CCLmJCAkiTpHP+VF1innvWfjUi3C7PIILd2DHGvq4fYN8VlWt6kykFj+Xy+e5NGfneZ9Gxfyzbs3RbRh+hXPtchGTmYar5+LXx7hUpWyBAQRWxIQksAFl5tfne5gfUVhTF8nWvsivHoisLooXpt+FOZksKw0l/Odbu7aVsljH9o4461BM9Mt7FhaEtd6hEt9jGTKSMSWVLkkuEGPl098/wgWpfjyratj+lrRukLYd8LBVeWFMamXiNTnblrBxZ4h7tu9dNZN9XZX2/llXQfNXYNx2fdWrhCEWab82KSU+o5SqkMpdTLk2FeVUheVUm8bX7eF3PdlpdQ5pdRppdQtIce3KKVOGPd9QxlbZymlspRSzxvHDyqlqqL7LSYvrTVf+mENZ9r7+ebdmy5rrRAL2Rlp2PMyZ9UGu7lrkOPNPeyNc1n+HRsW8snrlkWlw2qwjcXrcVptFLxCSKQ2ySI1RXId/V1gT5jjX9dabzS+XgFQSq0B7gLWGuc8rpQKZvGeAO4Dqo2v4HN+HOjWWi8Hvg48MsPvJeX8++8a2Ffj4Iu3rOTaFaWmvGb5LJee7jdWFyXKHrHRsKw0jwUF2XGbNgo2trMlWCM0kXqmDAha698Cke4neCfwA631iNa6ATgHbFdKlQEFWus3daDJzPeA94Wc87Rx+4fAjSpWG+8mkd+fc/J3r9Zy67oF3H/dMtNet8JmnV1AqHGwoaIwLlMrsaKUYle1nd+fd0a9PXgkXAMjFGSnzygpLsR0zOYn7NNKqRpjSslmHCsHmkMe02IcKzdujz9+2Tlaay/QC4RtfqOUuk8pdVgpdbizM/7Vo7HS0j3Ip//jKMtK83j0gxtitjF9OBU2Kxe7Z1aLcMHl5sTFXm5fnzh7xEbL7mo7PYOjnGqd3l4L0eB0e2S6SJhipgHhCWAZsBFwAI8Zx8P95tKTHJ/snCsPav1trfVWrfXW0lJzplDMNjzq45PPHMHr03zro1siqqiNpgpbDh6fn86B6dciBKeLbr0qsTb9iIZrlsevHbZrQKqUhTlmFBC01u1aa5/W2g88CWw37moBKkMeWgG0Gscrwhy/7BylVDpQSORTVClFa81f/eQEJy/28U93bYxLUVdF0cz3Rdhf42DToqKx5aupxJ6XxeqygrjkEbrcHllyKkwxo4Bg5ASC3g8EVyC9BNxlrBxaQiB5fEhr7QD6lVI7jPzAPcCLIefca9z+APBLnSj7Fprse29e4MdHL/K5m6q5cfX8uIxhpktPG5xuTrX2pVQyebzd1XaOXOhmyOMz9XVdAx65QhCmiGTZ6XPAm8BKpVSLUurjwD8YS0hrgBuAvwDQWp8CXgDeAX4KPKC1Dv7vuR/4dwKJ5vPAq8bxp4ASpdQ54PPAg9H65pLJoYYu/s++d7hp9Tz+/N3VcRtH+Qx3TttfE7jgS8RNP6Jl13I7Hp9/bFtQM/j8mq5Bj7StEKaYcoJaa313mMNPTfL4h4GHwxw/DKwLc3wY+OBU40hlbb3DfOrZI1QW5/C1D2+Mytr5mcrJTKckN3PaAWFfjYMti21X7DyWSrYvKSYz3cLrZ51cv9KcPR66Bz1ojTS2E6aQdWxxNuINJJEHPT6+9dEtFGTHrldRpAJLTyOfMjrXMUBdW39C7hEbTdkZaWyrspmaWA7WIMiUkTCDBIQ4++pLp3i7uYfHPriBFfNj19p6Ospt1mlVK79ywoFScOu61A4IALuWl3K6vZ+OvmFTXs/lNtpWSFJZmEACQhw9d6iJ5w41c//1y7g1gebeK2w5XOwemnKj+qD9NQ62LS5mQWH4vYlTidltLIJXCHa5QhAmkIAQJ0ebunnoxVPsrrbzxfesjPdwLlNhszLijawW4Wx7P6fb++Peu8gsa8oKKM7NNG35abCx3WR7PwsRLRIQ4qCjf5j7nznC/MIsvnn3JtLimEQOp2IaK432j00XpV4xWjgWi+Jdy0p4/Zwz4iuo2XC5PVgUFEkfI2ECCQgmG/X5+fSzx+gdGuVbf7w1If+jR7ovgtaafTUOtlcVM68g9aeLgnZX2+noH+FM+0DMX8s54KE4NzPhPjSI1CQBwWQP76/lUGMXj/zhetYsjN12mLNRbiwdvThFQDjTPsC5joGUX1003q7qQNuU352NfT+tLveIJJSFaSQgmOhHR1r47huN/Mk1S7hzY/nUJ8RJblY6tpyMKZee7q9pxaJgzxxYXRSqvMjKUnuuKYllqVIWZpKAYJITLb381U9OsGNpMV++bVW8hzOliin2RdBas++Egx1LSyjNn3ufYHdV2zlY38WIN7ZtLFxujySUhWkkIJigs3+E+75/mJLcTP7ljzaTMcP9fc00VXFaXVs/9Z3uObO6aLxdy+0Mjfo4eqEnpq/jHBiR1tfCNIn/mynJebx+PvXsEboHPXz7nq1J85+7wihOm2glzf4aR2C6aO3cWF003o5lJaRZFK+fi10eYcTro3/YK32MhGkkIMTY/953ircau3nkD9ezrrww3sOJWHmRleFR/9h+vqECq4taedcy+5ztsVOQncHGyqKY1iN0u0cB6WMkzCMBIYaeO9TEMwea+MS1SxM6iRzOZEtPT7X20eganLPTRUG7ltupudhLz+CVQTManEZRmiSVhVkkIMTI4cYuvvLiSXZX2/nSnsRPIo9XUTzxvgj7TzhIs6g5O10UtLvajtbwxvnYtMMOXp3JlJEwiwSEGGjrHeaTzxxlYZGVf7l7c1IWFZVPsHOa1pr9NQ6uWW7HNsd/UW2oLCIvKz1m3U9dY1cIMmUkzDHnAkJz1yD/8suz+GawiXwkhkd9fOL7hxnyeHnynq0U5sS/nfVM5GdnUBSmFuHkxT6auga5PYGa8cVLRpqFHUtLYpZYltbXwmxzLiC8dLyVf/z5Ge5+8gCO3unvGzwZrTV//ZOTHG/p5Wsf3pgw7axnqrzIekW18r4TraRbFO9ZG58tPhPN7mo7zV1DXHC5o/7cTvcImWkW8rOm3MdKiKiYcwHhgRuW89gHN3DyYi+3/vPv+O932qP23N99o5EfHW3hszdWc0sKzK8HahEuBYTgdNGuantC9mCKh2A77FhMG3UZVcqBbciFiL05FxAA/nBLBfs+s4vyIit/+r3DfPWlU7OuOH3jnJO/3V/LzWvm89kb47cncjQFq5WDtQjHW3pp6R5ir0wXjVliz6W8yBqT5adSpSzMNicDAsDS0jx+/Kl38bFrqvjuG438weNvUN85s+6VzV2DPPAfR1liz+VrH9oQ1z2Ro6nCZmVo1EeXsdplf00rGWmK96TA1U+0KKXYtdzOG+edUc9LuQZGJKEsTDVnAwJAVnoaD92xlifv2crFniFu/+br/OhIy7SeY9Dj5b7vH8Hn1zx5z1byE2BP5GgJrUUIThddW11KoTV1vsdo2FVtp2/YS01LdNtYOAc82OUKQZhoTgeEoJvXzOfVz+5mXXkhX/jP43z++bcZGPFOeZ7Wmi/9sIa6tj6+cfcmlthzTRitecbaYPcMcay5h9be4TlfjBbONcvtKEXUp41c7hFZYSRMJQHBUFZo5bk/28Hnbqrmv96+yB3ffJ2TF3snPeeJ35xnX42DL92yiutXzjNppOYpt10qTttf4yAzzcJNa2R10XjFuZmsXVjA76LYDnvQ42V41C9TRsJUEhBCpFkUn7tpBf/xZzsY8vh4/+O/5zuvN4Rt8Par0x08+rPT3LFhIZ+8bmkcRht7hdYMCrLTaeoa5JUTDq5dUUpBCk2JRdOu5aUca+rGHcGVZSTGahBkykiYSAJCGDuWlvDqZ3dz3YpS/ve+d/jTpw+PJVYB6jsH+PPnjrF6QQH/8IfrU3pZYIUth5+dasfROzzndkabjt3VdkZ9moMN0WljIX2MRDxIQJiALTeTJ+/ZykN3rOF3Z53c+s+/5UC9i/7hUe77/hEy0ix8+54tWDPT4j3UmKqwWensHyEz3cKNq1NvWixatiy2kZVuiVo9wqUrBJkyEuaREshJKKX42DVL2FZVzGeeO8YfPXmAZaV5NDjdPPPxq8dW4aSyYB7hhpWlKbWCKtqyM9LYvqQ4aolll1uuEIT55AohAuvKC3n5M7t436ZyznYM8L/2rmbnspJ4D8sUwaC3d/3COI8k8e2utnO2YyAqLVEudTqVKwRhHrlCiFBeVjpf+9BGHrx1FfPys+M9HNPcuGoep1p7uXm1rC6ayq7lpUAdr5918sGtlbN6LteAh9zMtJSfkhSJRa4QpmkuBQOAKnsuX/vQRvnFFIFVC/Kx52XyehSWn7oGRiiW6SJhMgkIQkSJxaK4Zrmd359z4p9lGwuX2yPTRcJ0EhCEiKJdy+04Bzwcbeqe1fM4BzzY5QpBmEwCghBRtGfdAux5Wfz/r9SGLWiMVJd7RK4QhOkkIAgRRfnZGXxpz0qONvXwX29fnNFzaK1xGXshCGEmCQhCRNkHNlewoaKQv3+1bkatLPqGvHj9WvZCEKabMiAopb6jlOpQSp0MOVaslHpNKXXW+NMWct+XlVLnlFKnlVK3hBzfopQ6Ydz3DWX0e1BKZSmlnjeOH1RKVUX3WxTCXBaL4qH3rqW9b4THf31u2uc7jaI0uzS2EyaL5Arhu8CeccceBH6hta4GfmH8HaXUGuAuYK1xzuNKqeB6xSeA+4Bq4yv4nB8HurXWy4GvA4/M9JsRIlFsXmTjDzaV8+TvGmhyDU7r3LG2FTJlJEw2ZUDQWv8W6Bp3+E7gaeP208D7Qo7/QGs9orVuAM4B25VSZUCB1vpNHci0fW/cOcHn+iFwo0rlbnFizvjLW1eRblH87f53pnVeV7BthSSVhclmmkOYr7V2ABh/BruelQPNIY9rMY6VG7fHH7/sHK21F+gFwvaFUErdp5Q6rJQ63NnZOcOhC2GO+QXZfPrdy/n5O+3T6nHkNK4QZNmpMFu0k8rhPtnrSY5Pds6VB7X+ttZ6q9Z6a2lp6QyHKIR5/uSaJSwqzuFvXj7FqM8f0TnBKSObJJWFyWYaENqNaSCMPzuM4y1AaBOXCqDVOF4R5vhl5yil0oFCrpyiEiIpZWek8T/3ruZsxwDPHLgQ0Tku9wiF1gwy0mQRoDDXTH/iXgLuNW7fC7wYcvwuY+XQEgLJ40PGtFK/UmqHkR+4Z9w5wef6APBLPZuKHiESzM1r5rO72s7XXzuDy9j4ZjJSgyDiJZJlp88BbwIrlVItSqmPA38P3KyUOgvcbPwdrfUp4AXgHeCnwANaa5/xVPcD/04g0XweeNU4/hRQopQ6B3weY8WSEKlCKcVXbl+D2+PjsdfOTPl4l3sEuySURRxM2f5aa333BHfdOMHjHwYeDnP8MLAuzPFh4INTjUOIZFY9P597di7mu2808pGrF7F2YeGEj3UNeFg+L8/E0QkRIJOUQpjkczetwJaTyd+89M6kfY5cbo9UKYu4kIAghEkKrRl88T0rOdTYxb4aR9jHeH1+ugc9lEiVsogDCQhCmOjD2ypZU1bA371Sy5DHd8X93YOjaC01CCI+JCAIYaI0i+Kr711La+8w//ab81fc3yV7KYs4koAghMm2Lynm9vVl/NtvztPSfXmfo+CyVFl2KuJBAoIQcfBXt61GKfi7V+ouO+4cu0KQgCDMJwFBiDhYWGTl/uuWs/+EgzfPu8aOX7pCkCkjYT4JCELEySeuW0p5kZW/efkUXqPPkWvAg0VBkTUjzqMTc5EEBCHiJDsjjb/eu5q6tn6eeyvQJDhQg5CFxSId4IX5JCAIEUe3rlvAjqXFPPbz0/QMenANjMiSUxE3EhCEiCOlFA/dsZa+odFA8zu3NLYT8SMBQYg4W11WwB9dvYhnDjZxpr2fYqlBEHEiAUGIBPCFm1eSl5VO/7BXlpyKuJGAIEQCsOVm8vmbVwBSgyDiZ8r210IIc3zk6kW09w1z2/qyeA9FzFESEIRIEOlpFr60Z1W8hyHmMJkyEkIIAUhAEEIIYZCAIIQQApCAIIQQwiABQQghBCABQQghhEECghBCCEACghBCCIPSWsd7DDOilOoELszwdDvgjOJwokXGNT0yrulL1LHJuKZnNuNarLUuDXdH0gaE2VBKHdZab433OMaTcU2PjGv6EnVsMq7pidW4ZMpICCEEIAFBCCGEYa4GhG/HewATkHFNj4xr+hJ1bDKu6YnJuOZkDkEIIcSV5uoVghBCiHEkIAghhABSPCAopfYopU4rpc4ppR4Mc79SSn3DuL9GKbXZhDFVKqV+pZSqVUqdUkp9NsxjrldK9Sql3ja+vhLrcRmv26iUOmG85uEw98fj/VoZ8j68rZTqU0p9btxjTHm/lFLfUUp1KKVOhhwrVkq9ppQ6a/xpm+DcSX8WYzCuR5VSdca/00+UUkUTnDvpv3mMxvZVpdTFkH+v2yY41+z37PmQMTUqpd6e4NyYvGcT/W4w9WdMa52SX0AacB5YCmQCx4E14x5zG/AqoIAdwEETxlUGbDZu5wNnwozremBfHN6zRsA+yf2mv19h/k3bCBTWmP5+AdcCm4GTIcf+AXjQuP0g8MhMfhZjMK73AOnG7UfCjSuSf/MYje2rwBcj+Lc29T0bd/9jwFfMfM8m+t1g5s9YKl8hbAfOaa3rtdYe4AfAneMecyfwPR1wAChSSsV0Q1uttUNrfdS43Q/UAuWxfM0oMv39GudG4LzWeqYV6rOitf4t0DXu8J3A08btp4H3hTk1kp/FqI5La/1zrbXX+OsBoCJarzcdE7xnkTD9PQtSSingQ8Bz0Xq9CMc00e8G037GUjkglAPNIX9v4cpfvJE8JmaUUlXAJuBgmLt3KqWOK6VeVUqtNWlIGvi5UuqIUuq+MPfH9f0C7mLi/6TxeL8A5mutHRD4Dw3MC/OYeL9vf0Lgyi6cqf7NY+XTxnTWdyaYAonne7YbaNdan53g/pi/Z+N+N5j2M5bKAUGFOTZ+jW0kj4kJpVQe8CPgc1rrvnF3HyUwLbIB+CbwX2aMCbhGa70ZuBV4QCl17bj74/l+ZQLvBf4zzN3xer8iFc/37a8BL/DsBA+Z6t88Fp4AlgEbAQeB6Znx4vaeAXcz+dVBTN+zKX43THhamGPTfr9SOSC0AJUhf68AWmfwmKhTSmUQ+Ad/Vmv94/H3a637tNYDxu1XgAyllD3W49Jatxp/dgA/IXAZGiou75fhVuCo1rp9/B3xer8M7cFpM+PPjjCPidfP2b3A7cBHtDHRPF4E/+ZRp7Vu11r7tNZ+4MkJXjNe71k68AfA8xM9Jpbv2QS/G0z7GUvlgPAWUK2UWmJ8urwLeGncY14C7jFWz+wAeoOXZrFizE8+BdRqrb82wWMWGI9DKbWdwL+TK8bjylVK5QdvE0hKnhz3MNPfrxATfmqLx/sV4iXgXuP2vcCLYR4Tyc9iVCml9gB/CbxXaz04wWMi+TePxdhC807vn+A1TX/PDDcBdVrrlnB3xvI9m+R3g3k/Y9HOlCfSF4FVMWcIZN//2jj2SeCTxm0F/Ktx/wlgqwlj2kXgUq4GeNv4um3cuD4NnCKwUuAA8C4TxrXUeL3jxmsnxPtlvG4OgV/whSHHTH+/CAQkBzBK4BPZx4ES4BfAWePPYuOxC4FXJvtZjPG4zhGYUw7+jP3b+HFN9G9uwti+b/z81BD4pVWWCO+Zcfy7wZ+rkMea8p5N8rvBtJ8xaV0hhBACSO0pIyGEENMgAUEIIQQgAUEIIYRBAoIQQghAAoIQQgiDBAQhhBCABAQhhBCG/wcQ37xW8qTLvAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(lis['train'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "3d4fe1cf-24a7-4290-9413-55a3094aaa6a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2ba52737c5f8>]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(lis['val'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "20573ad4-fcf3-41c6-ac29-e2720774e026",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2ba5273c50b8>]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(lis['test'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7ff73a93-45cf-4197-8fdf-bd04b09773c2",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8963b471-aa60-4983-984f-67bd3b6b60d4",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "258917e5-7634-489c-8676-ebd7d30e4f1d",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "631c913b-ccbb-45f9-a560-50c6359e5476",
   "metadata": {},
   "outputs": [],
   "source": [
    "image = np.load('/n/holyscratch01/wadduwage_lab/D2NN_QPM_classification/datasets/bacteria_np/train/16/16_34.npy',allow_pickle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "2b078a64-097a-45c4-9057-078469a3aa5c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x2b5559b59160>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(image[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "abc98ec8-8565-4521-ad26-fd93d4e9cc1a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "image[1][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "7775789e-d0a1-448f-9854-cf7ed13ddf1b",
   "metadata": {},
   "outputs": [],
   "source": [
    "import glob"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "4d735f85-7422-4440-aeb6-b1e66c9b756f",
   "metadata": {},
   "outputs": [],
   "source": [
    "type_ = 'val'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "dfee51d8-2aca-4e55-80cf-edec0ddee3c3",
   "metadata": {},
   "outputs": [],
   "source": [
    "img_dirs = sorted(glob.glob(f'{data_dir}/{type_}/*/*'), key= lambda x: int(x.split('/')[-1][:-4]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "55274143-74d6-40b9-b849-ef7ccad205c2",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "052b2aed-a241-460d-9e82-c7faaab66724",
   "metadata": {},
   "outputs": [],
   "source": [
    "dirs = {}\n",
    "\n",
    "for i,x in enumerate(img_dirs):\n",
    "    \n",
    "    class_ = int(x.split('/')[-2])\n",
    "    \n",
    "    if(class_ in dirs.keys()):\n",
    "        dirs[class_].append(x)\n",
    "    else:\n",
    "        dirs[class_] = [x]\n",
    "\n",
    "        \n",
    "img_dirs_filtered = []\n",
    "\n",
    "for i in range(0,21):\n",
    "    count = len(dirs[i])\n",
    "    img_dirs_filtered.append(dirs[i][:int(count*0.05)])\n",
    "    \n",
    "flat_list = [item for sublist in img_dirs_filtered for item in sublist]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1e880a7f-6428-4333-961e-528370d79281",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d994e706-ab02-4bf5-8067-4a762bc33185",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bcbc7f1d-fdc6-484a-8530-f4664f40d6de",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "ddf94cef-f051-4721-8d9b-aaddbbee1105",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2418"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(flat_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ddc59156-eb98-489b-a46d-d03e12778a71",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "from modules.helpers import *\n",
    "from modules.datasets import *\n",
    "from modules.dataloaders import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9bf1d1d3-e3fc-4b4b-890e-8dad545a186f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using balanced mode\n",
      "Dataset type train label type: strain -> "
     ]
    }
   ],
   "source": [
    "dataloaders = {}\n",
    "dataloaders['train'], dataloaders['val'], dataloaders['test'] =  get_bacteria_dataloaders(64, 16 ,10, \"strain\", balanced_mode = True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5cfcafbc-8e8f-4283-a950-7f8c7849e931",
   "metadata": {},
   "outputs": [],
   "source": [
    "cat = [0]*21\n",
    "\n",
    "for i, (inputs, labels) in enumerate(dataloaders['test']):\n",
    "    \n",
    "    for k in labels:\n",
    "        cat[k] += 1\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "898cf7cf-6581-463a-b418-557562099a50",
   "metadata": {},
   "outputs": [],
   "source": [
    "cat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b66ab772-d3e5-41be-82be-843d75819745",
   "metadata": {},
   "outputs": [],
   "source": [
    "np.sum(np.array(cat))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c83ae5f0-6720-494d-aad5-d8b818a2b18e",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "c44e37fa-a0db-4701-943d-344af272a71d",
   "metadata": {},
   "source": [
    "### Save class types in a single .npy array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "a3756fab-6557-499c-b60a-12b29adc1e56",
   "metadata": {},
   "outputs": [],
   "source": [
    "import glob\n",
    "data_dir = '/n/holyscratch01/wadduwage_lab/D2NN_QPM_classification/datasets/bacteria_np'\n",
    "type_ = 'test'\n",
    "\n",
    "all_dirs = sorted(glob.glob(f'{data_dir}/{type_}/*/*'), key= lambda x: int(x.split('/')[-1][:-4]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "d666d2f4-139c-48dd-b37f-ee7ed67d8a5d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset type test -> All files = 48790\n"
     ]
    }
   ],
   "source": [
    "print(f\"Dataset type {type_}\", end = \" -> \")\n",
    "print(f\"All files = {len(all_dirs)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "830885b7-26e3-4d73-b87f-9a8dadf89622",
   "metadata": {},
   "outputs": [],
   "source": [
    "species_mapping_dict  = {\n",
    "    0: 0,\n",
    "    1: 1,\n",
    "    2: 2,\n",
    "    3: 4,\n",
    "    4: 2,\n",
    "    5: 2,\n",
    "    6: 2,\n",
    "    7: 3,\n",
    "    8: 4,\n",
    "    9: 2,\n",
    "    10: 2,\n",
    "    11: 2,\n",
    "    12: 3,\n",
    "    13: 3,\n",
    "    14: 3,\n",
    "    15: 3,\n",
    "    16: 0,\n",
    "    17: 0,\n",
    "    18: 0,\n",
    "    19: 0,\n",
    "    20: 0,\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "3183c280-d078-40a0-833b-3bfdd0807f6e",
   "metadata": {},
   "outputs": [],
   "source": [
    "def __getclass_(meta_data, label_type):\n",
    "    if(label_type == 'class'):\n",
    "        return meta_data[0]\n",
    "\n",
    "    elif(label_type == 'antibiotic_resistant'):\n",
    "        '''\n",
    "            Dataset wild_type equals to class 1\n",
    "\n",
    "            antibiotic_resistance is when class is => not wild_type\n",
    "        '''\n",
    "        antibiotic_resistance = int(not(meta_data[1]))\n",
    "        return antibiotic_resistance\n",
    "\n",
    "    elif(label_type == 'gram_strain'):\n",
    "        return meta_data[2]\n",
    "\n",
    "    elif(label_type == 'species'):\n",
    "        return species_mapping_dict[meta_data[0]] #map class to species\n",
    "\n",
    "    elif(label_type == 'all'):\n",
    "        antibiotic_resistance = int(not(meta_data[1]))\n",
    "        \n",
    "        return {'class' : meta_data[0], 'antibiotic_resistant' : antibiotic_resistance, \n",
    "                'gram_strain' : meta_data[2],\n",
    "                'species' : species_mapping_dict[meta_data[0]] }\n",
    "    else:\n",
    "        raise Exception(\"Invalid label type\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "a3a16a98-0220-4e1b-b9de-09aad91a136f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n",
      "10000\n",
      "20000\n",
      "30000\n",
      "40000\n"
     ]
    }
   ],
   "source": [
    "global_dict = {}\n",
    "\n",
    "img_dirs_filtered = []\n",
    "\n",
    "for i,x in enumerate(all_dirs):\n",
    "\n",
    "    data  = np.load(x, allow_pickle=True)[1]\n",
    "    metadata = __getclass_(data, 'all')\n",
    "\n",
    "    global_dict[x.split('/')[-1]] = metadata\n",
    "\n",
    "\n",
    "    if(i % 10000 == 0):\n",
    "        print(i)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "acd149fe-6185-4474-94b1-5c674c35735d",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "self.img_dirs = img_dirs_filtered\n",
    "\n",
    "print(f\"Loaded {len(self.img_dirs)} images only from class {isolate_class}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "486e5361-1a98-4c15-8425-c7410196694d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "global_dict['0_15472.npy']['class']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "0d19131a-c4e1-4e9d-ba70-143ba55b9b96",
   "metadata": {},
   "outputs": [],
   "source": [
    "# import pickle\n",
    "# with open('saved_dictionary.pkl', 'wb') as f:\n",
    "#     pickle.dump(global_dict, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "94348543-4631-4aed-a9e4-870f81d94b48",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/n/home12/ramith/FYP/bacteria-classification\n"
     ]
    }
   ],
   "source": [
    "!pwd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "40e84c61-2caf-4290-99a6-6b38bc4ba2f9",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "fyp_env",
   "language": "python",
   "name": "fyp_env"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}