Skip to content
Snippets Groups Projects
Normal view Permalink
MachineLearning_inMolecularBiology2022_GroupProject.ipynb 158 KiB
Newer Older
jpronkko's avatar
initial python notebook
jpronkko committed
1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>C&gt;A</td>\n",
       "      <td>ACA</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>C&gt;A</td>\n",
       "      <td>ACC</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 9693 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "  Mutation type Trinucleotide  ALL::TARGET-10-PAIXPH-03A-01D  \\\n",
       "0           C>A           ACA                              0   \n",
       "1           C>A           ACC                              0   \n",
       "\n",
       "   ALL::TARGET-10-PAKHZT-03A-01R  ALL::TARGET-10-PAKMVD-09A-01D  \\\n",
       "0                              0                              0   \n",
       "1                              0                              0   \n",
       "\n",
       "   ALL::TARGET-10-PAKSWW-03A-01D  ALL::TARGET-10-PALETF-03A-01D  \\\n",
       "0                              1                              0   \n",
       "1                              1                              0   \n",
       "\n",
       "   ALL::TARGET-10-PALLSD-09A-01D  ALL::TARGET-10-PAMDKS-03A-01D  \\\n",
       "0                              0                              0   \n",
       "1                              0                              0   \n",
       "\n",
       "   ALL::TARGET-10-PAPJIB-04A-01D  ...  Head-SCC::V-109  Head-SCC::V-112  \\\n",
       "0                              2  ...                0                0   \n",
       "1                              0  ...                1                0   \n",
       "\n",
       "   Head-SCC::V-116  Head-SCC::V-119  Head-SCC::V-123  Head-SCC::V-124  \\\n",
       "0                0                0                0                0   \n",
       "1                0                0                0                0   \n",
       "\n",
       "   Head-SCC::V-125  Head-SCC::V-14  Head-SCC::V-29  Head-SCC::V-98  \n",
       "0                0               0               0               1  \n",
       "1                0               1               0               0  \n",
       "\n",
       "[2 rows x 9693 columns]"
      ]
     },
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1088 
     "execution_count": 10,
jpronkko's avatar
initial python notebook
jpronkko committed
1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "other_wes_mut = pd.read_csv(\"./project_data/catalogs/WES/WES_Other.96.csv\")\n",
    "other_wes_mut.head(2)"
   ]
  },
  {
   "cell_type": "code",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1100 
   "execution_count": 11,
jpronkko's avatar
initial python notebook
jpronkko committed
1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Cancer Types</th>\n",
       "      <th>Sample Names</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>SBS1</th>\n",
       "      <th>SBS2</th>\n",
       "      <th>SBS3</th>\n",
       "      <th>SBS4</th>\n",
       "      <th>SBS5</th>\n",
       "      <th>SBS6</th>\n",
       "      <th>SBS7a</th>\n",
       "      <th>...</th>\n",
       "      <th>SBS51</th>\n",
       "      <th>SBS52</th>\n",
       "      <th>SBS53</th>\n",
       "      <th>SBS54</th>\n",
       "      <th>SBS55</th>\n",
       "      <th>SBS56</th>\n",
       "      <th>SBS57</th>\n",
       "      <th>SBS58</th>\n",
       "      <th>SBS59</th>\n",
       "      <th>SBS60</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ALL</td>\n",
       "      <td>TARGET-10-PAIXPH-03A-01D</td>\n",
       "      <td>0.529</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ALL</td>\n",
       "      <td>TARGET-10-PAKHZT-03A-01R</td>\n",
       "      <td>0.696</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 68 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "  Cancer Types              Sample Names  Accuracy  SBS1  SBS2  SBS3  SBS4  \\\n",
       "0          ALL  TARGET-10-PAIXPH-03A-01D     0.529     0     0     0     0   \n",
       "1          ALL  TARGET-10-PAKHZT-03A-01R     0.696     0     0     0     0   \n",
       "\n",
       "   SBS5  SBS6  SBS7a  ...  SBS51  SBS52  SBS53  SBS54  SBS55  SBS56  SBS57  \\\n",
       "0     0     0      0  ...      0      0      0      1      0      0      0   \n",
       "1     0     0      0  ...      0      0      0      1      0      0      0   \n",
       "\n",
       "   SBS58  SBS59  SBS60  \n",
       "0      0      0      0  \n",
       "1      0      0      0  \n",
       "\n",
       "[2 rows x 68 columns]"
      ]
     },
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1217 
     "execution_count": 11,
jpronkko's avatar
initial python notebook
jpronkko committed
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "other_wes_act = pd.read_csv(\"./project_data/activities/WES/WES_Other.activities.csv\")\n",
    "other_wes_act.head(2)"
   ]
  },
  {
   "cell_type": "code",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1229 
   "execution_count": 50,
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import sklearn\n",
    "from sklearn.decomposition import PCA\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import torch \n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.model_selection import train_test_split\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1242 
    "from sklearn.model_selection import KFold\n",
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1243 
    "from sklearn.ensemble import RandomForestClassifier\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1244 
    "from sklearn.metrics import accuracy_score\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import roc_curve\n",
    "\n",
    "# These ones are work in progress\n",
    "def plot_roc_auc(y_pred, y_test):\n",
    "    auc = roc_auc_score(y_test, y_pred)\n",
    "\n",
    "    fp_rate, tp_rate, thresholds = roc_curve(y_test, y_pred)\n",
    "    plt.figure(figsize=(7,6))\n",
    "    plt.axis('scaled')\n",
    "    plt.xlim([0,1])\n",
    "    plt.ylim([0,1])\n",
    "    plt.title(\"AUC & ROC\")\n",
    "    plt.plot(fp_rate, tp_rate, 'g')\n",
    "    plt.fill_between(fp_rate, tp_rate, facecolor = \"green\", alpha = 0.7)\n",
    "    plt.text(0.95, 0.05, f'AUC = {auc}', ha='right', fontsize=12, weight='bold', color='blue')\n",
    "    plt.xlabel(\"False Positive Rate\")\n",
    "    plt.ylabel(\"True Positive Rate\")\n",
    "\n",
    "def plot_confusion_mat(y_pred, y_test):\n",
    "    cm = sklearn.metrics.confusion_matrix(y_pred, y_test)\n",
    "    plt.figure(figsize=(7,6))\n",
    "    sns.heatmap(cm, annot=True, fmt='d', cmap='viridis')\n",
    "    plt.xlabel('Predicted')\n",
    "    plt.ylabel('True')\n",
    "    plt.ylim(0, 2)\n"
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1271 1272 1273 1274 1275 1276 
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1277 1278 1279 
    "### Dataset preprocess, combine profile data to a single data frame\n",
    "\n",
    "From all profile sets, a combined data frame is made, which has samples in the rows and features in the columns."
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1280 1281 1282 1283 
   ]
  },
  {
   "cell_type": "code",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1284 
   "execution_count": 47,
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1285 
   "metadata": {},
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1286 1287 1288 1289 1290 
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1291 1292 1293 1294 1295 1296 1297 1298 
      "Missing entries in profile mutations: 0\n",
      "The shape of the all mutations data frame (23829, 97)\n",
      "Checking normalization: sum of some rows:\n",
      " Biliary-AdenoCA::SP117655    1.0\n",
      "Biliary-AdenoCA::SP117556    1.0\n",
      "Biliary-AdenoCA::SP117627    1.0\n",
      "Biliary-AdenoCA::SP117775    1.0\n",
      "Biliary-AdenoCA::SP117332    1.0\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1299 1300 
      "dtype: float64\n",
      "Some tumor counts:\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1301 1302 1303 1304 1305 
      " Breast-cancer       1637\n",
      "Liver-HCC           1318\n",
      "ColoRect-AdenoCA    1185\n",
      "Prost-AdenoCA       1091\n",
      "Skin-Melanoma       1070\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1306 
      "Name: tumor_types, dtype: int64\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1307 1308 1309 
      "Unique tumor types\n",
      "['ALL', 'AML', 'Adrenal-neoplasm', 'Biliary-AdenoCA', 'Bladder-TCC', 'Blood-CMDI', 'Bone-Benign', 'Bone-Epith', 'Bone-Osteosarc', 'Bone-cancer', 'Breast-AdenoCA', 'Breast-DCIS', 'Breast-Fibroadenoma', 'Breast-LobularCA', 'Breast-cancer', 'CNS-GBM', 'CNS-LGG', 'CNS-Medullo', 'CNS-Oligo', 'CNS-PiloAstro', 'CNS-glioma-NOS', 'Cervix-AdenoCA', 'Cervix-CA', 'Cervix-SCC', 'ColoRect-AdenoCA', 'ColoRect-Adenoma', 'DLBC', 'Eso-AdenoCA', 'Eso-SCC', 'Ewings', 'Eye-Melanoma', 'Eye-RB', 'Head-SCC', 'Kidney-ChRCC', 'Kidney-NOS', 'Kidney-Papillary', 'Kidney-RCC', 'Kidney-Wilms', 'Liver-Benign', 'Liver-HCC', 'Lung-AdenoCA', 'Lung-CArcinoid', 'Lung-NOS', 'Lung-SCC', 'Lung-Small', 'Lymph-BNHL', 'Lymph-CLL', 'Lymph-NOS', 'Lymph-TNHL', 'Lymph-cHL', 'Meninges-Meningioma', 'Mesothelium-Mesothelioma', 'Myeloid-AML', 'Myeloid-MDS', 'Myeloid-MPN', 'Neuroblastoma', 'Oral-SCC', 'Ovary-AdenoCA', 'Panc-AdenoCA', 'Panc-Endocrine', 'Panc-Other', 'Para-AdenoCA', 'Para-Adenoma', 'Pheochromocytoma', 'Pit-All', 'Prost-AdenoCA', 'Prost-Adenoma', 'Sarcoma', 'Sarcoma-bone', 'Skin-BCC', 'Skin-Melanoma', 'Skin-SCC', 'Small-Intestine-carcinoid', 'SoftTissue-Leiomyo', 'SoftTissue-Liposarc', 'Stomach-AdenoCA', 'Testis-CA', 'Thy-AdenoCA', 'Thymoma', 'Transitional-cell-carcinoma', 'UCS', 'Uterus-AdenoCA']\n",
      "Some content from the full set\n"
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1310 1311 1312 1313 
     ]
    },
    {
     "data": {
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>mut_tri</th>\n",
       "      <th>C&gt;A_ACA</th>\n",
       "      <th>C&gt;A_ACC</th>\n",
       "      <th>C&gt;A_ACG</th>\n",
       "      <th>C&gt;A_ACT</th>\n",
       "      <th>C&gt;A_CCA</th>\n",
       "      <th>C&gt;A_CCC</th>\n",
       "      <th>C&gt;A_CCG</th>\n",
       "      <th>C&gt;A_CCT</th>\n",
       "      <th>C&gt;A_GCA</th>\n",
       "      <th>C&gt;A_GCC</th>\n",
       "      <th>...</th>\n",
       "      <th>T&gt;G_CTT</th>\n",
       "      <th>T&gt;G_GTA</th>\n",
       "      <th>T&gt;G_GTC</th>\n",
       "      <th>T&gt;G_GTG</th>\n",
       "      <th>T&gt;G_GTT</th>\n",
       "      <th>T&gt;G_TTA</th>\n",
       "      <th>T&gt;G_TTC</th>\n",
       "      <th>T&gt;G_TTG</th>\n",
       "      <th>T&gt;G_TTT</th>\n",
       "      <th>tumor_types</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 
       "      <th>Biliary-AdenoCA::SP117655</th>\n",
       "      <td>0.018054</td>\n",
       "      <td>0.009933</td>\n",
       "      <td>0.001678</td>\n",
       "      <td>0.010336</td>\n",
       "      <td>0.014430</td>\n",
       "      <td>0.009933</td>\n",
       "      <td>0.001812</td>\n",
       "      <td>0.012081</td>\n",
       "      <td>0.011074</td>\n",
       "      <td>0.005101</td>\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1369 
       "      <td>...</td>\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 
       "      <td>0.017987</td>\n",
       "      <td>0.001275</td>\n",
       "      <td>0.001141</td>\n",
       "      <td>0.002886</td>\n",
       "      <td>0.005973</td>\n",
       "      <td>0.005570</td>\n",
       "      <td>0.003221</td>\n",
       "      <td>0.004228</td>\n",
       "      <td>0.013221</td>\n",
       "      <td>Biliary-AdenoCA</td>\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1380 1381 
       "    </tr>\n",
       "    <tr>\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 
       "      <th>Biliary-AdenoCA::SP117556</th>\n",
       "      <td>0.022406</td>\n",
       "      <td>0.011006</td>\n",
       "      <td>0.002555</td>\n",
       "      <td>0.013758</td>\n",
       "      <td>0.012382</td>\n",
       "      <td>0.009631</td>\n",
       "      <td>0.001376</td>\n",
       "      <td>0.013561</td>\n",
       "      <td>0.015920</td>\n",
       "      <td>0.007272</td>\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1393 
       "      <td>...</td>\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 
       "      <td>0.010417</td>\n",
       "      <td>0.001572</td>\n",
       "      <td>0.000983</td>\n",
       "      <td>0.003931</td>\n",
       "      <td>0.002948</td>\n",
       "      <td>0.005110</td>\n",
       "      <td>0.001965</td>\n",
       "      <td>0.006093</td>\n",
       "      <td>0.012579</td>\n",
       "      <td>Biliary-AdenoCA</td>\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1404 1405 
       "    </tr>\n",
       "    <tr>\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 
       "      <th>Biliary-AdenoCA::SP117627</th>\n",
       "      <td>0.017728</td>\n",
       "      <td>0.011987</td>\n",
       "      <td>0.002195</td>\n",
       "      <td>0.012325</td>\n",
       "      <td>0.011987</td>\n",
       "      <td>0.009286</td>\n",
       "      <td>0.001351</td>\n",
       "      <td>0.010299</td>\n",
       "      <td>0.010299</td>\n",
       "      <td>0.008442</td>\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1417 
       "      <td>...</td>\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 
       "      <td>0.007429</td>\n",
       "      <td>0.001182</td>\n",
       "      <td>0.001013</td>\n",
       "      <td>0.002364</td>\n",
       "      <td>0.002364</td>\n",
       "      <td>0.002701</td>\n",
       "      <td>0.001351</td>\n",
       "      <td>0.002533</td>\n",
       "      <td>0.008779</td>\n",
       "      <td>Biliary-AdenoCA</td>\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1428 1429 
       "    </tr>\n",
       "    <tr>\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 
       "      <th>Biliary-AdenoCA::SP117775</th>\n",
       "      <td>0.017484</td>\n",
       "      <td>0.009911</td>\n",
       "      <td>0.002337</td>\n",
       "      <td>0.010152</td>\n",
       "      <td>0.010394</td>\n",
       "      <td>0.006607</td>\n",
       "      <td>0.002095</td>\n",
       "      <td>0.011522</td>\n",
       "      <td>0.013053</td>\n",
       "      <td>0.006768</td>\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1441 
       "      <td>...</td>\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 
       "      <td>0.010636</td>\n",
       "      <td>0.001692</td>\n",
       "      <td>0.000564</td>\n",
       "      <td>0.002256</td>\n",
       "      <td>0.002740</td>\n",
       "      <td>0.003062</td>\n",
       "      <td>0.002417</td>\n",
       "      <td>0.004270</td>\n",
       "      <td>0.009830</td>\n",
       "      <td>Biliary-AdenoCA</td>\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1452 1453 
       "    </tr>\n",
       "    <tr>\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 
       "      <th>Biliary-AdenoCA::SP117332</th>\n",
       "      <td>0.015046</td>\n",
       "      <td>0.010417</td>\n",
       "      <td>0.002315</td>\n",
       "      <td>0.008970</td>\n",
       "      <td>0.008681</td>\n",
       "      <td>0.006366</td>\n",
       "      <td>0.002894</td>\n",
       "      <td>0.010995</td>\n",
       "      <td>0.006076</td>\n",
       "      <td>0.007234</td>\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1465 
       "      <td>...</td>\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 
       "      <td>0.005208</td>\n",
       "      <td>0.000579</td>\n",
       "      <td>0.001157</td>\n",
       "      <td>0.002025</td>\n",
       "      <td>0.002025</td>\n",
       "      <td>0.004340</td>\n",
       "      <td>0.002315</td>\n",
       "      <td>0.004630</td>\n",
       "      <td>0.010995</td>\n",
       "      <td>Biliary-AdenoCA</td>\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1476 1477 1478 1479 1480 1481 
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 97 columns</p>\n",
       "</div>"
      ],
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1482 
      "text/plain": [
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 
       "mut_tri                     C>A_ACA   C>A_ACC   C>A_ACG   C>A_ACT   C>A_CCA  \\\n",
       "Biliary-AdenoCA::SP117655  0.018054  0.009933  0.001678  0.010336  0.014430   \n",
       "Biliary-AdenoCA::SP117556  0.022406  0.011006  0.002555  0.013758  0.012382   \n",
       "Biliary-AdenoCA::SP117627  0.017728  0.011987  0.002195  0.012325  0.011987   \n",
       "Biliary-AdenoCA::SP117775  0.017484  0.009911  0.002337  0.010152  0.010394   \n",
       "Biliary-AdenoCA::SP117332  0.015046  0.010417  0.002315  0.008970  0.008681   \n",
       "\n",
       "mut_tri                     C>A_CCC   C>A_CCG   C>A_CCT   C>A_GCA   C>A_GCC  \\\n",
       "Biliary-AdenoCA::SP117655  0.009933  0.001812  0.012081  0.011074  0.005101   \n",
       "Biliary-AdenoCA::SP117556  0.009631  0.001376  0.013561  0.015920  0.007272   \n",
       "Biliary-AdenoCA::SP117627  0.009286  0.001351  0.010299  0.010299  0.008442   \n",
       "Biliary-AdenoCA::SP117775  0.006607  0.002095  0.011522  0.013053  0.006768   \n",
       "Biliary-AdenoCA::SP117332  0.006366  0.002894  0.010995  0.006076  0.007234   \n",
       "\n",
       "mut_tri                    ...   T>G_CTT   T>G_GTA   T>G_GTC   T>G_GTG  \\\n",
       "Biliary-AdenoCA::SP117655  ...  0.017987  0.001275  0.001141  0.002886   \n",
       "Biliary-AdenoCA::SP117556  ...  0.010417  0.001572  0.000983  0.003931   \n",
       "Biliary-AdenoCA::SP117627  ...  0.007429  0.001182  0.001013  0.002364   \n",
       "Biliary-AdenoCA::SP117775  ...  0.010636  0.001692  0.000564  0.002256   \n",
       "Biliary-AdenoCA::SP117332  ...  0.005208  0.000579  0.001157  0.002025   \n",
       "\n",
       "mut_tri                     T>G_GTT   T>G_TTA   T>G_TTC   T>G_TTG   T>G_TTT  \\\n",
       "Biliary-AdenoCA::SP117655  0.005973  0.005570  0.003221  0.004228  0.013221   \n",
       "Biliary-AdenoCA::SP117556  0.002948  0.005110  0.001965  0.006093  0.012579   \n",
       "Biliary-AdenoCA::SP117627  0.002364  0.002701  0.001351  0.002533  0.008779   \n",
       "Biliary-AdenoCA::SP117775  0.002740  0.003062  0.002417  0.004270  0.009830   \n",
       "Biliary-AdenoCA::SP117332  0.002025  0.004340  0.002315  0.004630  0.010995   \n",
       "\n",
       "mut_tri                        tumor_types  \n",
       "Biliary-AdenoCA::SP117655  Biliary-AdenoCA  \n",
       "Biliary-AdenoCA::SP117556  Biliary-AdenoCA  \n",
       "Biliary-AdenoCA::SP117627  Biliary-AdenoCA  \n",
       "Biliary-AdenoCA::SP117775  Biliary-AdenoCA  \n",
       "Biliary-AdenoCA::SP117332  Biliary-AdenoCA  \n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1517 1518 1519 1520 
       "\n",
       "[5 rows x 97 columns]"
      ]
     },
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1521 
     "execution_count": 47,
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1522 1523 1524 1525 1526 1527 
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1528 1529 1530 1531 1532 1533 1534 
    "def prepare_profile_mut_df(raw_mutation_dfs):\n",
    "\n",
    "    mutations_all = pd.DataFrame()\n",
    "\n",
    "    for df in raw_mutation_dfs:\n",
    "        # Make a copy of the original data frame and start processing from there\n",
    "        mutations  = df.copy()\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1535 
    "    \n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1536 1537 1538 
    "        mutations['mut_tri'] = mutations.apply(lambda a: '{}_{}'.format(a['Mutation type'], a['Trinucleotide']), axis=1)\n",
    "        mutations = mutations.set_index('mut_tri').drop(['Mutation type', 'Trinucleotide'], axis=1)\n",
    "        mutations = mutations.T\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1539 
    "\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1540 1541 1542 
    "        # Rename some index names\n",
    "        renamed_items = list(mutations.index)\n",
    "        index_items = list(mutations.index)\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1543 
    "\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1544 1545 1546 
    "        for i in range(len(index_items)):\n",
    "            #result = re.sub(r'BC::', \"CA::\", items[i])\n",
    "            renamed_items[i] = index_items[i].replace('Ca', 'CA')\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1547 
    "       \n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1548 
    "        mutations.rename(index=dict(zip(index_items, renamed_items)), inplace = True)\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1549 
    "   \n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1550 1551 1552 
    "        # Normalize \n",
    "        row_sums = mutations.sum(axis=1)\n",
    "        mutations = mutations.divide(row_sums, axis = 0)\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1553 
    "\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1554 1555 1556 1557 
    "        mutations_all = pd.concat([mutations_all, mutations])\n",
    "\n",
    "    # Do we need to renormalize after obtaining the full dataframe?\n",
    "  \n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1558 
    "    # Figure out tumor types based on the first part of the index\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1559 1560 1561 
    "    tumor_types = [a.split(':')[0] for a in mutations_all.index]\n",
    "\n",
    "    unique_tumor_types = list(sorted(set(mutations_all)))\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1562 1563 
    "\n",
    "    # Attach this back to the frame\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1564 
    "    mutations_all[\"tumor_types\"] = tumor_types\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1565 1566 1567 
    "\n",
    "    # Prepare a list with all the types appearing only once\n",
    "    unique_tumor_types = sorted(list(set(tumor_types)))\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1568 1569 1570 1571 1572 
    "    return (mutations_all, unique_tumor_types)\n",
    "\n",
    "profile_raw_data_sets = [PCAWG_wgs_mut, TCGA_wes_mut, nonPCAWG_wgs_mut, other_wes_mut]\n",
    "\n",
    "profile_mut_all, unique_tumor_types = prepare_profile_mut_df(profile_raw_data_sets)\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1573 1574 1575 
    "\n",
    "\n",
    "# Check if the data frame is ok\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1576 1577 
    "print(\"Missing entries in profile mutations:\", profile_mut_all.isnull().sum().sum())\n",
    "print(\"The shape of the all mutations data frame\", profile_mut_all.shape)\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1578 1579 
    "\n",
    "# Check to see if the rows are normalized to one\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1580 
    "print(\"Checking normalization: sum of some rows:\\n\", profile_mut_all.iloc[:,0:-1].sum(axis=1).head(5))\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1581 1582 
    "\n",
    "# Check some counts of tumor types\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1583 
    "print(\"Some tumor counts:\\n\", profile_mut_all[\"tumor_types\"].value_counts().head(5))\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1584 1585 
    "\n",
    "# Tumor types\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1586 1587 1588 1589 1590 1591 1592 1593 1594 
    "print(\"Unique tumor types\")\n",
    "print(unique_tumor_types)\n",
    "\n",
    "# Data matrix X for fitting, omit the tumor labeling from there, use that information in constructing true y\n",
    "# Note: this contains profile data only\n",
    "X_prf = profile_mut_all.drop(\"tumor_types\", axis=1)\n",
    "\n",
    "print(\"Some content from the full set:\")\n",
    "profile_mut_all.head(5)\n"
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1595 1596 1597 1598 1599 1600 1601 1602 1603 
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Check data content"
   ]
  },
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1604 1605 1606 1607 1608 1609 1610 
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the plots below, we see that at the moment there are tumor types with not that big sample size. How to combine them?"
   ]
  },
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1611 1612 
  {
   "cell_type": "code",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1613 
   "execution_count": 27,
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1614 1615 1616 1617 
   "metadata": {},
   "outputs": [
    {
     "data": {
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1618 
      "image/png": "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",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1619 
      "text/plain": [
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1620 
       "<Figure size 1800x360 with 1 Axes>"
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1621 1622 1623 1624 1625 1626 1627 1628 1629 
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1630 
    "\"\"\" fig, ax = plt.subplots(3,1)\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1631 
    "fig.set_size_inches(10,12)\n",
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1632 
    "\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1633 1634 1635 1636 
    "sns.countplot(x=mutations_TCGA[\"tumor_types\"], \n",
    "              palette=sns.hls_palette(2),\n",
    "              ax=ax[0])\n",
    "ax[0].tick_params(axis=\"x\", rotation=90)\n",
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1637 
    "\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1638 1639 1640 1641 
    "sns.countplot(x=mutations_PCWGA[\"tumor_types\"], \n",
    "              palette=sns.hls_palette(2),\n",
    "              ax=ax[1])\n",
    "ax[1].tick_params(axis=\"x\", rotation=90)\n",
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1642 
    "\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1643 1644 1645 1646 1647 
    "sns.countplot(x=mutations_all[\"tumor_types\"], \n",
    "              palette=sns.hls_palette(2),\n",
    "              ax=ax[2])\n",
    "ax[2].tick_params(axis=\"x\", rotation=90)\n",
    "\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1648 
    "fig.tight_layout() \"\"\"\n",
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1649 
    "\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1650 1651 1652 
    "plt.figure(figsize=(25, 5))\n",
    "sns.countplot(x=profile_mut_all[\"tumor_types\"], palette=sns.hls_palette(2))\n",
    "plt.xticks(rotation=90);\n"
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1653 1654 
   ]
  },
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1655 1656 1657 1658 
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1659 
    "### Testing with a single RandomForest binary classifier"
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1660 1661 
   ]
  },
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1662 1663 
  {
   "cell_type": "code",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1664 
   "execution_count": 48,
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1665 1666 
   "metadata": {},
   "outputs": [
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1667 1668 1669 1670 
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1671 1672 
      "Dimension of the training data (16680, 96) and test data (7149, 96)\n",
      "Accuracy: 0.9795775632955658\n"
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1673 1674 
     ]
    },
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1675 1676 
    {
     "data": {
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1677 
      "image/png": "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",
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1678 
      "text/plain": [
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1679 
       "<Figure size 504x432 with 2 Axes>"
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1680 1681 1682 1683 1684 1685 
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1686 1687 1688 
    }
   ],
   "source": [
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 
    "# For binary classification, construct y vector for a sigle selected tumor type\n",
    "\n",
    "target_type = \"Biliary-AdenoCA\"\n",
    "\n",
    "y_prf_bin = [1 if tumor_type == target_type else 0 for tumor_type in profile_mut_all[\"tumor_types\"]]\n",
    "\n",
    "# Split the data for fitting and prediction, use simple splitting here\n",
    "X_prf_train, X_prf_test, y_prf_train, y_prf_test = train_test_split(X_prf, y_prf_bin, test_size = 0.3, random_state=1)\n",
    "\n",
    "\n",
    "print(f\"Dimension of the training data {X_prf_train.shape} and test data {X_prf_test.shape}\")\n",
    "\n",
    "# Make a model\n",
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1702 
    "model = RandomForestClassifier(n_estimators=100, random_state=0)\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1703 
    "\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 
    "# Fit the model \n",
    "model.fit(X_prf_train, y_prf_train)\n",
    "\n",
    "# Predict with unused (test) data \n",
    "y_prf_pred = model.predict(X_prf_test)\n",
    "\n",
    "print(f\"Accuracy:\", accuracy_score(y_prf_test, y_prf_pred))\n",
    "\n",
    "# Plot some results\n",
    "plot_confusion_mat(y_prf_pred, y_prf_test)"
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1714 1715 1716 1717 1718 1719 
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1720 1721 1722 
    "### Multiple models experiment, but still binary classification\n",
    "\n",
    "Note: this takes quite a while to compute"
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1723 1724 1725 1726 
   ]
  },
  {
   "cell_type": "code",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1727 
   "execution_count": 49,
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1728 1729 1730 1731 1732 1733 
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 
      "Accuracy score for ALL: 0.9875507063925024\n",
      "Accuracy score for AML: 0.9790180444817457\n",
      "Accuracy score for Adrenal-neoplasm: 0.9889495034270528\n",
      "Accuracy score for Biliary-AdenoCA: 0.9795775632955658\n",
      "Accuracy score for Bladder-TCC: 0.9934256539376136\n",
      "Accuracy score for Blood-CMDI: 0.9896489019443279\n",
      "Accuracy score for Bone-Benign: 0.9994404811861799\n",
      "Accuracy score for Bone-Epith: 0.9995803608896349\n",
      "Accuracy score for Bone-Osteosarc: 0.9980416841516296\n",
      "Accuracy score for Bone-cancer: 0.99972024059309\n",
      "Accuracy score for Breast-AdenoCA: 0.9913274583857882\n",
      "Accuracy score for Breast-DCIS: 0.9998601202965449\n",
      "Accuracy score for Breast-Fibroadenoma: 0.9994404811861799\n",
      "Accuracy score for Breast-LobularCA: 0.9994404811861799\n",
      "Accuracy score for Breast-cancer: 0.9328577423415862\n",
      "Accuracy score for CNS-GBM: 0.9739823751573646\n",
      "Accuracy score for CNS-LGG: 0.9862917890614072\n",
      "Accuracy score for CNS-Medullo: 0.9749615330815499\n",
      "Accuracy score for CNS-Oligo: 0.9991607217792698\n",
      "Accuracy score for CNS-PiloAstro: 0.995244090082529\n",
      "Accuracy score for CNS-glioma-NOS: 0.9938452930479788\n",
      "Accuracy score for Cervix-AdenoCA: 1.0\n",
      "Accuracy score for Cervix-CA: 0.9895090222408729\n",
      "Accuracy score for Cervix-SCC: 0.9990208420758148\n",
      "Accuracy score for ColoRect-AdenoCA: 0.9585956077773116\n",
      "Accuracy score for ColoRect-Adenoma: 0.9984613232619947\n",
      "Accuracy score for DLBC: 0.9785984053713807\n",
      "Accuracy score for Eso-AdenoCA: 0.9811162400335711\n",
      "Accuracy score for Eso-SCC: 0.9783186459644706\n",
      "Accuracy score for Ewings: 0.9847531123234019\n",
      "Accuracy score for Eye-Melanoma: 0.995244090082529\n",
      "Accuracy score for Eye-RB: 0.99972024059309\n",
      "Accuracy score for Head-SCC: 0.9657294726535179\n",
      "Accuracy score for Kidney-ChRCC: 0.9930060148272486\n",
      "Accuracy score for Kidney-NOS: 0.9983214435585397\n",
      "Accuracy score for Kidney-Papillary: 0.9878304657994125\n",
      "Accuracy score for Kidney-RCC: 0.9734228563435445\n",
      "Accuracy score for Kidney-Wilms: 0.9988809623723598\n",
      "Accuracy score for Liver-Benign: 0.9984613232619947\n",
      "Accuracy score for Liver-HCC: 0.9650300741362429\n",
      "Accuracy score for Lung-AdenoCA: 0.970625262274444\n",
      "Accuracy score for Lung-CArcinoid: 0.9979018044481746\n",
      "Accuracy score for Lung-NOS: 0.9984613232619947\n",
      "Accuracy score for Lung-SCC: 0.975660931598825\n",
      "Accuracy score for Lung-Small: 0.9965030074136243\n",
      "Accuracy score for Lymph-BNHL: 0.9690865855364387\n",
      "Accuracy score for Lymph-CLL: 0.990068541054693\n",
      "Accuracy score for Lymph-NOS: 0.9983214435585397\n",
      "Accuracy score for Lymph-TNHL: 0.9995803608896349\n",
      "Accuracy score for Lymph-cHL: 0.9993006014827248\n",
      "Accuracy score for Meninges-Meningioma: 0.9969226465239893\n",
      "Accuracy score for Mesothelium-Mesothelioma: 0.9934256539376136\n",
      "Accuracy score for Myeloid-AML: 0.99972024059309\n",
      "Accuracy score for Myeloid-MDS: 1.0\n",
      "Accuracy score for Myeloid-MPN: 0.9981815638550846\n",
      "Accuracy score for Neuroblastoma: 0.9840537138061267\n",
      "Accuracy score for Oral-SCC: 0.9941250524548888\n",
      "Accuracy score for Ovary-AdenoCA: 0.9783186459644706\n",
      "Accuracy score for Panc-AdenoCA: 0.960973562736047\n",
      "Accuracy score for Panc-Endocrine: 0.9953839697859841\n",
      "Accuracy score for Panc-Other: 0.9984613232619947\n",
      "Accuracy score for Para-AdenoCA: 0.9998601202965449\n",
      "Accuracy score for Para-Adenoma: 0.9998601202965449\n",
      "Accuracy score for Pheochromocytoma: 0.9930060148272486\n",
      "Accuracy score for Pit-All: 0.9988809623723598\n",
      "Accuracy score for Prost-AdenoCA: 0.960833683032592\n",
      "Accuracy score for Prost-Adenoma: 0.9969226465239893\n",
      "Accuracy score for Sarcoma: 0.9879703455028676\n",
      "Accuracy score for Sarcoma-bone: 0.9885298643166877\n",
      "Accuracy score for Skin-BCC: 0.9974821653378095\n",
      "Accuracy score for Skin-Melanoma: 0.9832144355853966\n",
      "Accuracy score for Skin-SCC: 0.9986012029654497\n",
      "Accuracy score for Small-Intestine-carcinoid: 0.9980416841516296\n",
      "Accuracy score for SoftTissue-Leiomyo: 0.9991607217792698\n",
      "Accuracy score for SoftTissue-Liposarc: 0.9993006014827248\n",
      "Accuracy score for Stomach-AdenoCA: 0.9732829766400896\n",
      "Accuracy score for Testis-CA: 0.9918869771996084\n",
      "Accuracy score for Thy-AdenoCA: 0.9778990068541055\n",
      "Accuracy score for Thymoma: 0.9953839697859841\n",
      "Accuracy score for Transitional-cell-carcinoma: 0.9841935935095818\n",
      "Accuracy score for UCS: 0.9974821653378095\n",
      "Accuracy score for Uterus-AdenoCA: 0.9801370821093859\n"
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1816 1817 1818 1819 
     ]
    }
   ],
   "source": [
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1820 1821 1822 
    "for tgt_tumor_type in unique_tumor_types:\n",
    "    y_prf = [1 if tumor_type == tgt_tumor_type else 0 for tumor_type in profile_mut_all[\"tumor_types\"]]\n",
    "    X_prf_train, X_prf_test, y_prf_train, y_prf_test = train_test_split(X_prf, y_prf, test_size = 0.3, random_state=1)\n",
jpronkko's avatar
Some data source combine WGS, WES testing  
jpronkko committed
1823 
    "    model = RandomForestClassifier(n_estimators=100, random_state=0)\n",
jpronkko's avatar
Profile data matrix preparation  
jpronkko committed
1824 1825 1826 
    "    model.fit(X_prf_train, y_prf_train)\n",
    "    y_prf_pred = model.predict(X_prf_test)\n",
    "    print(f\"Accuracy score for {tgt_tumor_type}: {accuracy_score(y_prf_test, y_prf_pred)}\")\n"
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1827 
   ]
jpronkko's avatar
initial python notebook
jpronkko committed
1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
jpronkko's avatar
Beginnings with RandomForest classifier  
jpronkko committed
1846 
   "version": "3.9.5"
jpronkko's avatar
initial python notebook
jpronkko committed
1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}
Show full blame