6. BioNet-OS.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "da103003-0a84-4cc7-a129-d17c7246df8f",
   "metadata": {},
   "outputs": [],
   "source": [
    "### General analysis package\n",
    "import os\n",
    "import time\n",
    "import random\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import scipy.stats as stats\n",
    "from scipy.stats import norm, pearsonr, spearmanr\n",
    "from scipy.spatial import distance\n",
    "### Drawing package\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib import gridspec\n",
    "import matplotlib.lines as mlines\n",
    "from matplotlib.font_manager import FontProperties\n",
    "#import seaborn as sns\n",
    "import matplotlib as mpl\n",
    "mpl.rcParams['pdf.fonttype'] = 42 # Output editable PDF text\n",
    "### else\n",
    "#import cloudpickle as pickle\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\") # Error alert\n",
    "from IPython.core.interactiveshell import InteractiveShell\n",
    "InteractiveShell.ast_node_interactivity = 'all' #last_expr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "eab2c8ba-7db6-41f4-ae78-e99b43fe03c1",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Global seed set to 42\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "42"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using torch 1.10.1\n"
     ]
    }
   ],
   "source": [
    "from sklearn import model_selection\n",
    "from sklearn.preprocessing import StandardScaler, Normalizer\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.metrics import roc_auc_score, accuracy_score\n",
    "from sklearn.model_selection import train_test_split, StratifiedShuffleSplit\n",
    "from sklearn.metrics import accuracy_score, balanced_accuracy_score, recall_score, precision_score, roc_auc_score, average_precision_score\n",
    "\n",
    "import math\n",
    "import torch\n",
    "from torch import nn\n",
    "import torch.nn.functional as F\n",
    "import torch.utils.data as TorchData\n",
    "from torchmetrics import Accuracy\n",
    "from pytorch_lightning import LightningModule, Trainer, seed_everything\n",
    "from pytorch_lightning.callbacks.early_stopping import EarlyStopping\n",
    "from pytorch_lightning.loggers import TensorBoardLogger\n",
    "\n",
    "from imblearn.over_sampling import RandomOverSampler, SMOTE\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "SEED = 42\n",
    "seed_everything(SEED, workers=True)\n",
    "\n",
    "print(\"Using torch\", torch.__version__)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9c53c835-571c-4c48-aa66-08316fc9fa1a",
   "metadata": {},
   "source": [
    "# 1. Knowledge network information（Mask）"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "5a6cdccd-ebb5-485e-89af-d5f4b01c4d80",
   "metadata": {},
   "outputs": [],
   "source": [
    "microRNA = pd.read_csv('20220610_Results/1. diff_microRNA(FC2_FDR0.05).csv', index_col=0)['ID']"
   ]
  },
  {
   "cell_type": "raw",
   "id": "fb1c57c7-bde4-451a-b7ad-b6901ceadcc5",
   "metadata": {
    "jupyter": {
     "source_hidden": true
    },
    "tags": []
   },
   "source": [
    "path = 'Breast cancer/'\n",
    "### Target Gene\n",
    "tar_data = pd.read_csv(path+'Target/hsa_MTI.csv')\n",
    "mir2gene = {}\n",
    "for i in tar_data.index:\n",
    "    mir = str(tar_data.loc[i, 'miRNA']).strip()\n",
    "    gene = str(tar_data.loc[i, 'Target Gene (Entrez Gene ID)']).strip()\n",
    "    mir2gene[mir] = mir2gene.get(mir, set())\n",
    "    mir2gene[mir].add(gene)\n",
    "gene_set = set()\n",
    "for mir in microRNA:\n",
    "    gene_set.update(mir2gene.get(mir, set()))\n",
    "GENE_LIST = list(gene_set)\n",
    "GENE_MASK = []\n",
    "for mir in microRNA:\n",
    "    genes = mir2gene.get(mir, set())\n",
    "    GENE_MASK.append([1 if g in genes else 0 for g in GENE_LIST])\n",
    "GENE_MASK = torch.tensor(np.array(GENE_MASK).T, dtype=torch.float)\n",
    "### Module\n",
    "module_data = pd.read_csv(path+'KEGG/gene2module.txt', sep='\\t')\n",
    "gene2module = {}\n",
    "for i in module_data.index:\n",
    "    mid = str(module_data.loc[i, 'Module']).split('hsa_')[1].strip()\n",
    "    gene = str(module_data.loc[i, 'gene']).split('hsa:')[1].strip()\n",
    "    gene2module[gene] = gene2module.get(gene, set())\n",
    "    gene2module[gene].add(mid)\n",
    "module_set = set()\n",
    "for gene in GENE_LIST:\n",
    "    module_set.update(gene2module.get(gene, set()))\n",
    "MODULE_LIST = list(module_set)\n",
    "MODULE_MASK = []\n",
    "for gene in GENE_LIST:\n",
    "    modules = gene2module.get(gene, set())\n",
    "    MODULE_MASK.append([1 if m in modules else 0 for m in MODULE_LIST])\n",
    "MODULE_MASK = torch.tensor(np.array(MODULE_MASK).T, dtype=torch.float)\n",
    "### Pathway\n",
    "pathway_data = pd.read_csv(path+'KEGG/module2pathway.txt', sep='\\t')\n",
    "module2pathway = {}\n",
    "for i in pathway_data.index:\n",
    "    pathway = str(pathway_data.loc[i, 'Pathway']).split('path:')[1].strip()\n",
    "    module = str(pathway_data.loc[i, 'Module']).split('md:')[1].strip()\n",
    "    module2pathway[module] = module2pathway.get(module, set())\n",
    "    module2pathway[module].add(pathway)\n",
    "pathway_set = set()\n",
    "for module in MODULE_LIST:\n",
    "    pathway_set.update(module2pathway.get(module, set()))\n",
    "PATHWAY_LIST = list(pathway_set)\n",
    "\n",
    "PATHWAY_MASK = []\n",
    "for module in MODULE_LIST:\n",
    "    pathways = module2pathway.get(module, set())\n",
    "    PATHWAY_MASK.append([1 if p in pathways else 0 for p in PATHWAY_LIST])\n",
    "PATHWAY_MASK = torch.tensor(np.array(PATHWAY_MASK).T, dtype=torch.float)"
   ]
  },
  {
   "cell_type": "raw",
   "id": "7f645622-5fd8-4791-9980-75a0dd29eddf",
   "metadata": {
    "jupyter": {
     "source_hidden": true
    },
    "tags": []
   },
   "source": [
    "import cloudpickle as pickle\n",
    "pickle.dump([mir2gene, GENE_LIST, GENE_MASK, gene2module, MODULE_LIST, MODULE_MASK,\n",
    "             module2pathway, PATHWAY_LIST, PATHWAY_MASK], open('20220610_Results/6. KEGG_MappingData.pkl', 'wb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "8671f0d5-e6c1-4df5-92fe-ee1964dacf0a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import cloudpickle as pickle\n",
    "[mir2gene, GENE_LIST, GENE_MASK, gene2module, MODULE_LIST, MODULE_MASK,\n",
    " module2pathway, PATHWAY_LIST, PATHWAY_MASK] = pickle.load(open('20220610_Results/6. KEGG_MappingData.pkl', 'rb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "218eae86-e3ae-455d-83cd-30d675865eed",
   "metadata": {},
   "source": [
    "# 2. Constructing neural network model based on knowledge information"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "cef0c244-533e-42ea-a3ae-6595a998fa1a",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "class MaskedModel(LightningModule):\n",
    "    def __init__(self, mirN, geneN, gene_mask, moduleN, module_mask, pathwayN, pathway_mask, learning_rate):\n",
    "        super(MaskedModel, self).__init__()\n",
    "        self.save_hyperparameters()\n",
    "        self.learning_rate = learning_rate\n",
    "        # microRNA 2 Gene\n",
    "        self.mir2gene = nn.Linear(mirN, geneN)\n",
    "        self.mir2gene.weight.data = self.mir2gene.weight * gene_mask\n",
    "        self.gene_mask = gene_mask\n",
    "        self.gene_active = nn.ReLU() # ReLU Tanh\n",
    "        self.gene_bn = nn.BatchNorm1d(geneN, eps=1e-05, momentum=0.5, affine=True)\n",
    "        self.gene_dropout = nn.Dropout(p=0.8)\n",
    "        \n",
    "        ## Gene 2 Module\n",
    "        self.gene2module = nn.Linear(geneN, moduleN)\n",
    "        self.gene2module.weight.data = self.gene2module.weight * module_mask\n",
    "        self.module_mask = module_mask\n",
    "        self.module_active = nn.ReLU()\n",
    "        self.module_bn = nn.BatchNorm1d(moduleN, eps=1e-05, momentum=0.5, affine=True)\n",
    "        self.module_dropout = nn.Dropout(p=0.5)\n",
    "        \n",
    "        ## Module 2 Pathway\n",
    "        self.module2pathway = nn.Linear(moduleN, pathwayN)\n",
    "        self.module2pathway.weight.data = self.module2pathway.weight * pathway_mask\n",
    "        self.pathway_mask = pathway_mask\n",
    "        self.pathway_active = nn.ReLU()\n",
    "        self.pathway_bn = nn.BatchNorm1d(pathwayN, eps=1e-05, momentum=0.5, affine=True)\n",
    "        self.pathway_dropout = nn.Dropout(p=0.2)\n",
    "        \n",
    "        ## Output\n",
    "        self.pathway2out = nn.Linear(pathwayN, 2)\n",
    "        self.out_bn = nn.BatchNorm1d(2, eps=1e-05, momentum=0.5, affine=True)\n",
    "        #self.sigmoid = nn.Sigmoid()\n",
    "        \n",
    "        # Result save\n",
    "        self.gene_result = None\n",
    "        self.module_result = None\n",
    "        self.pathway_result = None\n",
    "    \n",
    "    def forward(self, x):\n",
    "        ## microRNA 2 Gene\n",
    "        #print(self.mir2gene.weight.data.shape, self.mir2gene.weight.shape, self.gene_mask.shape)\n",
    "        self.mir2gene.weight.data = self.mir2gene.weight * self.gene_mask\n",
    "        #print(self.mir2gene.weight.data.shape)\n",
    "        gene_x = self.mir2gene(x)\n",
    "        gene_x = self.gene_active(gene_x)\n",
    "        gene_x = self.gene_bn(gene_x)\n",
    "        gene_x = self.gene_dropout(gene_x)\n",
    "        self.gene_result = gene_x # save result\n",
    "        \n",
    "        ## Gene 2 Module\n",
    "        self.gene2module.weight.data = self.gene2module.weight * self.module_mask\n",
    "        module_x = self.gene2module(gene_x)\n",
    "        module_x = self.module_active(module_x)\n",
    "        module_x = self.module_bn(module_x)\n",
    "        module_x = self.module_dropout(module_x)\n",
    "        self.module_result = module_x # save result\n",
    "        \n",
    "        ## Module 2 Pathway\n",
    "        self.module2pathway.weight.data = self.module2pathway.weight * self.pathway_mask\n",
    "        pathway_x = self.module2pathway(module_x)\n",
    "        pathway_x = self.pathway_active(pathway_x)\n",
    "        pathway_x = self.pathway_bn(pathway_x)\n",
    "        pathway_x = self.pathway_dropout(pathway_x)\n",
    "        self.pathway_result = pathway_x # save result\n",
    "        \n",
    "        ## Output\n",
    "        out_x = self.pathway2out(pathway_x)\n",
    "        out_x = self.out_bn(out_x)\n",
    "        #out_x = self.sigmoid(out_x)\n",
    "        return out_x\n",
    "    \n",
    "    def training_step(self, batch, batch_nb):\n",
    "        x, y = batch\n",
    "        loss = F.cross_entropy(self.forward(x), y)\n",
    "        self.log(\"train_loss\", loss, prog_bar=True)\n",
    "        return loss\n",
    "    \n",
    "    def configure_optimizers(self):\n",
    "        return torch.optim.Adam(self.parameters(), lr=self.learning_rate)\n",
    "    \n",
    "    def validation_step(self, batch, batch_idx):\n",
    "        x, y = batch\n",
    "        logits = F.softmax(self(x))\n",
    "        loss = F.cross_entropy(logits, y)\n",
    "        preds = torch.argmax(logits, dim=1)\n",
    "        probs = logits[:, 1]\n",
    "        auc = roc_auc_score(y, probs)\n",
    "        bac = balanced_accuracy_score(y, preds)\n",
    "        aps = average_precision_score(y, probs)\n",
    "        recall = recall_score(y, preds)\n",
    "        precision = precision_score(y, preds)\n",
    "        #self.auc(probs, y)\n",
    "\n",
    "        # Calling self.log will surface up scalars for you in TensorBoard\n",
    "        self.log(\"val_loss\", loss, prog_bar=True)\n",
    "        self.log(\"val_bac\", bac, prog_bar=True)\n",
    "        self.log(\"val_auc\", auc, prog_bar=True)\n",
    "        self.log(\"val_aps\", aps, prog_bar=True)\n",
    "        self.log(\"val_rec\", recall, prog_bar=True)\n",
    "        self.log(\"val_pre\", precision, prog_bar=True)\n",
    "        return loss\n",
    "    \n",
    "    def validation_epoch_end(self, validation_step_outputs):\n",
    "        avg_loss = torch.stack(validation_step_outputs).mean()\n",
    "        #avg_loss = torch.stack([x[\"val_loss\"] for x in outputs]).mean() # Outputs are the values of return for validation\n",
    "        self.logger.experiment.add_scalar('loss', avg_loss, self.current_epoch)\n",
    "    \n",
    "    def test_step(self, batch, batch_idx):\n",
    "        # Here we just reuse the validation_step for testing\n",
    "        return self.validation_step(batch, batch_idx)\n",
    "    \n",
    "    def predict_step(self, batch, batch_idx):\n",
    "        x, y = batch\n",
    "        logits = F.softmax(self(x))\n",
    "        loss = F.cross_entropy(logits, y)\n",
    "        preds = torch.argmax(logits, dim=1)\n",
    "        probs = logits[:, 1]\n",
    "        return preds, probs\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8cdd7c5f-c72f-43f7-98cd-2b5751bc005f",
   "metadata": {},
   "source": [
    "# 3. Model Initialization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "479d1b3b-4f91-4ecb-8179-69394e1b8aa0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "710 11418 116 70\n"
     ]
    }
   ],
   "source": [
    "mirN = len(microRNA)\n",
    "geneN = len(GENE_LIST)\n",
    "moduleN = len(MODULE_LIST)\n",
    "pathwayN = len(PATHWAY_LIST)\n",
    "print(mirN, geneN, moduleN, pathwayN)\n",
    "GENE_No_MASK = torch.ones_like(GENE_MASK)\n",
    "MODULE_No_MASK = torch.ones_like(MODULE_MASK)\n",
    "PATHWAY_No_MASK = torch.ones_like(PATHWAY_MASK)\n",
    "model = MaskedModel(mirN, geneN, GENE_MASK, moduleN, MODULE_MASK, pathwayN, PATHWAY_MASK, 0.01)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "8e90412d-a3b3-4ee9-a6ab-1760780eba14",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MaskedModel(\n",
      "  (mir2gene): Linear(in_features=710, out_features=11418, bias=True)\n",
      "  (gene_active): ReLU()\n",
      "  (gene_bn): BatchNorm1d(11418, eps=1e-05, momentum=0.5, affine=True, track_running_stats=True)\n",
      "  (gene_dropout): Dropout(p=0.8, inplace=False)\n",
      "  (gene2module): Linear(in_features=11418, out_features=116, bias=True)\n",
      "  (module_active): ReLU()\n",
      "  (module_bn): BatchNorm1d(116, eps=1e-05, momentum=0.5, affine=True, track_running_stats=True)\n",
      "  (module_dropout): Dropout(p=0.5, inplace=False)\n",
      "  (module2pathway): Linear(in_features=116, out_features=70, bias=True)\n",
      "  (pathway_active): ReLU()\n",
      "  (pathway_bn): BatchNorm1d(70, eps=1e-05, momentum=0.5, affine=True, track_running_stats=True)\n",
      "  (pathway_dropout): Dropout(p=0.2, inplace=False)\n",
      "  (pathway2out): Linear(in_features=70, out_features=2, bias=True)\n",
      "  (out_bn): BatchNorm1d(2, eps=1e-05, momentum=0.5, affine=True, track_running_stats=True)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "print(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "e44b5f2d-bee5-452a-86d4-b55960ddc47b",
   "metadata": {
    "scrolled": true,
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parameter mir2gene.weight, shape torch.Size([11418, 710])\n",
      "Parameter mir2gene.bias, shape torch.Size([11418])\n",
      "Parameter gene_bn.weight, shape torch.Size([11418])\n",
      "Parameter gene_bn.bias, shape torch.Size([11418])\n",
      "Parameter gene2module.weight, shape torch.Size([116, 11418])\n",
      "Parameter gene2module.bias, shape torch.Size([116])\n",
      "Parameter module_bn.weight, shape torch.Size([116])\n",
      "Parameter module_bn.bias, shape torch.Size([116])\n",
      "Parameter module2pathway.weight, shape torch.Size([70, 116])\n",
      "Parameter module2pathway.bias, shape torch.Size([70])\n",
      "Parameter pathway_bn.weight, shape torch.Size([70])\n",
      "Parameter pathway_bn.bias, shape torch.Size([70])\n",
      "Parameter pathway2out.weight, shape torch.Size([2, 70])\n",
      "Parameter pathway2out.bias, shape torch.Size([2])\n",
      "Parameter out_bn.weight, shape torch.Size([2])\n",
      "Parameter out_bn.bias, shape torch.Size([2])\n"
     ]
    }
   ],
   "source": [
    "for name, param in model.named_parameters():\n",
    "    print(f\"Parameter {name}, shape {param.shape}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "51a24536-38cb-45c8-a7df-107d758e414e",
   "metadata": {},
   "source": [
    "# 4. Data import"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "ad87cac9-9616-4b17-8014-2a205ca0c635",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = pd.read_csv('20220610_Results/1. X_train_SMOTE.csv', index_col=0)\n",
    "X_test = pd.read_csv('20220610_Results/1. X_test.csv', index_col=0)\n",
    "y_train = pd.read_csv('20220610_Results/1. y_train_SMOTE.csv', index_col=0)\n",
    "y_test = pd.read_csv('20220610_Results/1. y_test.csv', index_col=0)\n",
    "\n",
    "cv = model_selection.StratifiedShuffleSplit(n_splits=2, test_size=.4, random_state=SEED)\n",
    "X = X_train.copy()\n",
    "y = y_train.copy()\n",
    "for train_index, val_index in cv.split(X, y):pass\n",
    "X_train, X_valid = X.iloc[train_index, :], X.iloc[val_index, :]\n",
    "y_train, y_valid = y.iloc[train_index], y.iloc[val_index]\n",
    "\n",
    "X_train = torch.from_numpy(X_train[microRNA].values).type(torch.FloatTensor)\n",
    "y_train = torch.from_numpy(y_train['Label'].values).type(torch.LongTensor)\n",
    "X_valid = torch.from_numpy(X_valid[microRNA].values).type(torch.FloatTensor)\n",
    "y_valid = torch.from_numpy(y_valid['Label'].values).type(torch.LongTensor)\n",
    "X_test = torch.from_numpy(X_test[microRNA].values).type(torch.FloatTensor)\n",
    "y_test = torch.from_numpy(y_test['Label'].values).type(torch.LongTensor)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b3efade8-0b8c-4c6f-bace-4c7048b3f2fd",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data_loader = TorchData.DataLoader(dataset=TorchData.TensorDataset(X_train, y_train), \n",
    "                                         batch_size=64, shuffle=True)\n",
    "\n",
    "valid_data_loader = TorchData.DataLoader(dataset=TorchData.TensorDataset(X_valid, y_valid), \n",
    "                                         batch_size=64, shuffle=True)\n",
    "\n",
    "test_data_loader = TorchData.DataLoader(dataset=TorchData.TensorDataset(X_test, y_test), \n",
    "                                        batch_size=X_test.shape[0], shuffle=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c66f126f-56de-401f-b50c-c87416b86fbd",
   "metadata": {},
   "source": [
    "# 5. Train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "5d3532df-b8e4-438e-b6c2-097847e38dc4",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "GPU available: False, used: False\n",
      "TPU available: False, using: 0 TPU cores\n",
      "IPU available: False, using: 0 IPUs\n",
      "HPU available: False, using: 0 HPUs\n"
     ]
    }
   ],
   "source": [
    "logger = TensorBoardLogger(\"tb_logs\", name=\"BioNet_SMOTE_val_loss\")\n",
    "# Initialize a trainer\n",
    "trainer = Trainer(\n",
    "    accelerator=\"cpu\",\n",
    "    #gpus=AVAIL_GPUS,\n",
    "    num_processes=1,\n",
    "    min_epochs=100,\n",
    "    max_epochs=500,\n",
    "    #progress_bar_refresh_rate=20,\n",
    "    callbacks=[EarlyStopping(monitor=\"val_loss\", mode='min')], # validate Stops when the AUC in the data is highest\n",
    "    auto_lr_find=True,\n",
    "    stochastic_weight_avg=False,\n",
    "    #deterministic=False, # GPU repeatable\n",
    "    #val_check_interval=100,\n",
    "    logger=logger,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "55f98644-6955-48f8-9260-c0bf777ba1cf",
   "metadata": {},
   "outputs": [],
   "source": [
    "# tune the model (batch_size, lr_find)\n",
    "#trainer.tune(model, train_data_loader, valid_data_loader)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "408e769d-ccee-461b-a05f-c40f5b126ff1",
   "metadata": {
    "scrolled": true,
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "   | Name            | Type        | Params\n",
      "-------------------------------------------------\n",
      "0  | mir2gene        | Linear      | 8.1 M \n",
      "1  | gene_active     | ReLU        | 0     \n",
      "2  | gene_bn         | BatchNorm1d | 22.8 K\n",
      "3  | gene_dropout    | Dropout     | 0     \n",
      "4  | gene2module     | Linear      | 1.3 M \n",
      "5  | module_active   | ReLU        | 0     \n",
      "6  | module_bn       | BatchNorm1d | 232   \n",
      "7  | module_dropout  | Dropout     | 0     \n",
      "8  | module2pathway  | Linear      | 8.2 K \n",
      "9  | pathway_active  | ReLU        | 0     \n",
      "10 | pathway_bn      | BatchNorm1d | 140   \n",
      "11 | pathway_dropout | Dropout     | 0     \n",
      "12 | pathway2out     | Linear      | 142   \n",
      "13 | out_bn          | BatchNorm1d | 4     \n",
      "-------------------------------------------------\n",
      "9.5 M     Trainable params\n",
      "0         Non-trainable params\n",
      "9.5 M     Total params\n",
      "37.897    Total estimated model params size (MB)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Sanity Checking: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f9d4097332f64b3db884fc3c9f86b585",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Training: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Trainer was signaled to stop but required minimum epochs (100) or minimum steps (None) has not been met. Training will continue...\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Validation: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Train the model\n",
    "trainer.fit(model, train_data_loader, valid_data_loader)\n",
    "trainer.save_checkpoint(\"Temp/BioNet_SMOTE_val_loss.ckpt\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3d0979f3-7ace-47f5-b50e-3c9f1c3f5d75",
   "metadata": {},
   "source": [
    "# 6. Model evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "fe384690-d5a2-4ca2-b855-533b3103b7f3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0d5833d014da47a58615f1453bc3fef0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Testing: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────\n",
      "       Test metric             DataLoader 0\n",
      "────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────\n",
      "         val_aps            0.9849232890146371\n",
      "         val_auc            0.9883495314554251\n",
      "         val_bac            0.9618513924481993\n",
      "        val_loss            0.3526972532272339\n",
      "         val_pre            0.9723635719704449\n",
      "         val_rec            0.9496828596244183\n",
      "────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[{'val_loss': 0.3526972532272339,\n",
       "  'val_bac': 0.9618513924481993,\n",
       "  'val_auc': 0.9883495314554251,\n",
       "  'val_aps': 0.9849232890146371,\n",
       "  'val_rec': 0.9496828596244183,\n",
       "  'val_pre': 0.9723635719704449}]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trainer.test(model, valid_data_loader)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "d4f9e3a0-d2ab-4f7a-bf9e-5d29fe0dfa76",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e3d38baaba4e405bb5fe7f33cd581f1c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Testing: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────\n",
      "       Test metric             DataLoader 0\n",
      "────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────\n",
      "         val_aps            0.9745004852432015\n",
      "         val_auc            0.9894352127425043\n",
      "         val_bac            0.9602158427028218\n",
      "        val_loss            0.3495907485485077\n",
      "         val_pre            0.9489194499017681\n",
      "         val_rec                0.943359375\n",
      "────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[{'val_loss': 0.3495907485485077,\n",
       "  'val_bac': 0.9602158427028218,\n",
       "  'val_auc': 0.9894352127425043,\n",
       "  'val_aps': 0.9745004852432015,\n",
       "  'val_rec': 0.943359375,\n",
       "  'val_pre': 0.9489194499017681}]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trainer.test(model, test_data_loader)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d6d88f24-14f7-4166-a0ad-70fbf7383ff4",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "b9b7fb29-674c-4f36-980b-3b50d4128269",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dfb8cca0b9694ce9b0811033949389a9",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Predicting: 32it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2400x1800 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 2400x1800 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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": [
    "from sklearn import metrics\n",
    "pred, probs = trainer.predict(model, test_data_loader)[0]\n",
    "#auc = roc_auc_score(y_test, probs.numpy())\n",
    "\n",
    "fpr, tpr, thresholds = metrics.roc_curve(y_test, probs.numpy(), pos_label=1)\n",
    "auc = metrics.auc(fpr, tpr)\n",
    "fig = plt.figure(figsize=(8, 6), dpi=300)\n",
    "_ = plt.plot(fpr, tpr, color=\"darkorange\", lw=2, label=\"ROC curve (area = %0.3f)\" % auc)\n",
    "_ = plt.plot([0, 1], [0, 1], color=\"navy\", lw=2, linestyle=\"--\")\n",
    "_ = plt.xlim([-0.01, 1.01])\n",
    "_ = plt.ylim([-0.01, 1.01])\n",
    "_ = plt.xlabel(\"False Positive Rate\")\n",
    "_ = plt.ylabel(\"True Positive Rate\")\n",
    "_ = plt.legend(loc=\"lower right\")\n",
    "_ = plt.show()\n",
    "fig.savefig('20220610_Results/6. ROC_BioNet_SMOTE_val_loss.pdf', dpi=300)\n",
    "\n",
    "precision, recall, _ = metrics.precision_recall_curve(y_test, pred.numpy())\n",
    "disp = metrics.PrecisionRecallDisplay(precision=precision, recall=recall)\n",
    "fig = plt.figure(figsize=(8, 6), dpi=300)\n",
    "_ = disp.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "efbafd71-4dac-41b8-9a4f-7c4262bf24b7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 480x480 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay\n",
    "fig = plt.figure(1, (6, 6), dpi=80)\n",
    "ax = fig.add_subplot(1,1,1)\n",
    "cm = confusion_matrix(y_test, pred.numpy(), labels=[0, 1])\n",
    "disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=[0, 1])\n",
    "_ = disp.plot(ax=ax, cmap='Blues') #, ax=ax, cmap='Blues'\n",
    "#_ = plot_confusion_matrix(model, X_test[select_microRNA], y_test, ax=ax, cmap='Blues')\n",
    "fig.savefig('20220610_Results/6. plot_confusion_matrix_BioNet_SMOTE_val_loss.pdf', dpi=300)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ffef46f2-9fa2-4f1c-bad4-662b6e21d7e7",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "36d9c848-d96e-48c6-b293-e1c527a62956",
   "metadata": {},
   "source": [
    "# 7. Model predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "d010d585-4d2e-443c-bdbb-a2dba64d3a7e",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "GPU available: False, used: False\n",
      "TPU available: False, using: 0 TPU cores\n",
      "IPU available: False, using: 0 IPUs\n",
      "HPU available: False, using: 0 HPUs\n"
     ]
    }
   ],
   "source": [
    "model = MaskedModel.load_from_checkpoint(checkpoint_path=\"Temp/BioNet_SMOTE_val_loss.ckpt\")\n",
    "trainer = Trainer()\n",
    "#trainer.test(model, dataloaders=test_data_loader)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "8a841710-c016-49a3-890c-6e9cd1dfc7c7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ff4a7ceee7ac4f8b9636286780c07cbf",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Predicting: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "preds = trainer.predict(model, test_data_loader)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "655c5a39-799d-4042-98c5-3ad3a518766e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[array([0, 0, 0, ..., 0, 0, 1]),\n",
       " array([2.9841816e-04, 3.6976105e-04, 8.3620066e-04, ..., 2.1612234e-03,\n",
       "        5.7251956e-03, 9.9773139e-01], dtype=float32)]"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_list = list(preds[0])\n",
    "pred_list[0] = pred_list[0].numpy()\n",
    "pred_list[1] = pred_list[1].numpy()\n",
    "pred_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "2d807452-325b-43e4-9235-a4ee4ed97d8a",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_test_ = pd.read_csv('20220610_Results/1. X_test.csv', index_col=0)\n",
    "pred_df = pd.DataFrame(np.array(pred_list).T, columns=['Pred', 'Prob'], index=X_test_.index)\n",
    "pred_df.to_csv('20220610_Results/6. Pred_result.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "4865933a-bea6-4407-b728-376d11a0b639",
   "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>Pred</th>\n",
       "      <th>Prob</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>GSM1878567</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000298</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GSM1877697</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000370</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GSM1879645</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000836</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GSM1877269</th>\n",
       "      <td>1.0</td>\n",
       "      <td>0.992352</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GSM1879958</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000428</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GSM1876809</th>\n",
       "      <td>1.0</td>\n",
       "      <td>0.997479</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GSM1879670</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000735</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GSM1878229</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002161</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GSM1878868</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005725</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GSM1880433</th>\n",
       "      <td>1.0</td>\n",
       "      <td>0.997731</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1646 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            Pred      Prob\n",
       "GSM1878567   0.0  0.000298\n",
       "GSM1877697   0.0  0.000370\n",
       "GSM1879645   0.0  0.000836\n",
       "GSM1877269   1.0  0.992352\n",
       "GSM1879958   0.0  0.000428\n",
       "...          ...       ...\n",
       "GSM1876809   1.0  0.997479\n",
       "GSM1879670   0.0  0.000735\n",
       "GSM1878229   0.0  0.002161\n",
       "GSM1878868   0.0  0.005725\n",
       "GSM1880433   1.0  0.997731\n",
       "\n",
       "[1646 rows x 2 columns]"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "acf81991-ddb0-43e3-891b-4ff8a0cb5eed",
   "metadata": {},
   "source": [
    "# 8. External data validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "9d9054b7-0e13-4407-8dcc-d200dde7e63e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(710, 523)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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>hsa-miR-9-5p</th>\n",
       "      <th>hsa-miR-126-3p</th>\n",
       "      <th>hsa-miR-140-5p</th>\n",
       "      <th>hsa-miR-105-5p</th>\n",
       "      <th>hsa-miR-297</th>\n",
       "      <th>hsa-let-7i-5p</th>\n",
       "      <th>hsa-miR-130a-3p</th>\n",
       "      <th>hsa-miR-146b-5p</th>\n",
       "      <th>hsa_negative_control_6</th>\n",
       "      <th>hsa_negative_control_7</th>\n",
       "      <th>...</th>\n",
       "      <th>hsa-miR-205-3p</th>\n",
       "      <th>hsa-miR-1260a</th>\n",
       "      <th>hsa-miR-525-5p</th>\n",
       "      <th>hsa-miR-187-5p</th>\n",
       "      <th>hsa-miRPlus-A1086</th>\n",
       "      <th>hsa-miR-551b-5p</th>\n",
       "      <th>hsa-miR-300</th>\n",
       "      <th>hsa-miR-193b-5p</th>\n",
       "      <th>Label</th>\n",
       "      <th>Disease</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6.877643</td>\n",
       "      <td>8.080215</td>\n",
       "      <td>6.245265</td>\n",
       "      <td>6.304782</td>\n",
       "      <td>6.686064</td>\n",
       "      <td>7.805950</td>\n",
       "      <td>6.814781</td>\n",
       "      <td>7.146955</td>\n",
       "      <td>6.170680</td>\n",
       "      <td>6.129586</td>\n",
       "      <td>...</td>\n",
       "      <td>7.176684</td>\n",
       "      <td>7.724408</td>\n",
       "      <td>6.035476</td>\n",
       "      <td>6.255048</td>\n",
       "      <td>6.572019</td>\n",
       "      <td>6.251777</td>\n",
       "      <td>5.873303</td>\n",
       "      <td>6.362097</td>\n",
       "      <td>1</td>\n",
       "      <td>primary breast cancer</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6.544287</td>\n",
       "      <td>6.882722</td>\n",
       "      <td>5.874236</td>\n",
       "      <td>6.324987</td>\n",
       "      <td>6.638048</td>\n",
       "      <td>6.946567</td>\n",
       "      <td>6.488574</td>\n",
       "      <td>6.690773</td>\n",
       "      <td>6.160272</td>\n",
       "      <td>6.104929</td>\n",
       "      <td>...</td>\n",
       "      <td>7.249094</td>\n",
       "      <td>7.308890</td>\n",
       "      <td>6.021796</td>\n",
       "      <td>6.902473</td>\n",
       "      <td>6.347183</td>\n",
       "      <td>6.625569</td>\n",
       "      <td>6.004899</td>\n",
       "      <td>6.560329</td>\n",
       "      <td>1</td>\n",
       "      <td>primary breast cancer</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 1928 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   hsa-miR-9-5p  hsa-miR-126-3p  hsa-miR-140-5p  hsa-miR-105-5p  hsa-miR-297  \\\n",
       "0      6.877643        8.080215        6.245265        6.304782     6.686064   \n",
       "1      6.544287        6.882722        5.874236        6.324987     6.638048   \n",
       "\n",
       "   hsa-let-7i-5p  hsa-miR-130a-3p  hsa-miR-146b-5p  hsa_negative_control_6  \\\n",
       "0       7.805950         6.814781         7.146955                6.170680   \n",
       "1       6.946567         6.488574         6.690773                6.160272   \n",
       "\n",
       "   hsa_negative_control_7  ...  hsa-miR-205-3p  hsa-miR-1260a  hsa-miR-525-5p  \\\n",
       "0                6.129586  ...        7.176684       7.724408        6.035476   \n",
       "1                6.104929  ...        7.249094       7.308890        6.021796   \n",
       "\n",
       "   hsa-miR-187-5p  hsa-miRPlus-A1086  hsa-miR-551b-5p  hsa-miR-300  \\\n",
       "0        6.255048           6.572019         6.251777     5.873303   \n",
       "1        6.902473           6.347183         6.625569     6.004899   \n",
       "\n",
       "   hsa-miR-193b-5p  Label                Disease  \n",
       "0         6.362097      1  primary breast cancer  \n",
       "1         6.560329      1  primary breast cancer  \n",
       "\n",
       "[2 rows x 1928 columns]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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>hsa-miR-5000-3p</th>\n",
       "      <th>hsa-miR-191-5p</th>\n",
       "      <th>hsa-miR-380-5p</th>\n",
       "      <th>hsa-miR-4474-5p</th>\n",
       "      <th>hsa-miR-4311</th>\n",
       "      <th>hsa-miR-518f-5p</th>\n",
       "      <th>hsa-miR-19b-1-5p</th>\n",
       "      <th>hsa-miR-425-3p</th>\n",
       "      <th>hsa-miR-625-5p</th>\n",
       "      <th>hsa-miR-4655-3p</th>\n",
       "      <th>...</th>\n",
       "      <th>hsa-miR-6796-5p</th>\n",
       "      <th>hsa-miR-98-3p</th>\n",
       "      <th>hsa-miR-6720-3p</th>\n",
       "      <th>hsa-miR-6849-3p</th>\n",
       "      <th>hsa-miR-203b-5p</th>\n",
       "      <th>hsa-miR-6769a-5p</th>\n",
       "      <th>hsa-miR-6504-3p</th>\n",
       "      <th>hsa-miR-6861-5p</th>\n",
       "      <th>hsa-miR-1251-5p</th>\n",
       "      <th>hsa-miR-219b-5p</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6.877273</td>\n",
       "      <td>6.930538</td>\n",
       "      <td>6.306224</td>\n",
       "      <td>5.805400</td>\n",
       "      <td>6.670228</td>\n",
       "      <td>6.103235</td>\n",
       "      <td>6.427767</td>\n",
       "      <td>6.634423</td>\n",
       "      <td>6.418833</td>\n",
       "      <td>6.082664</td>\n",
       "      <td>...</td>\n",
       "      <td>3.843195</td>\n",
       "      <td>2.413887</td>\n",
       "      <td>2.352475</td>\n",
       "      <td>3.609278</td>\n",
       "      <td>2.372301</td>\n",
       "      <td>6.129119</td>\n",
       "      <td>2.37833</td>\n",
       "      <td>6.909919</td>\n",
       "      <td>2.434913</td>\n",
       "      <td>3.156666</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7.604156</td>\n",
       "      <td>6.801036</td>\n",
       "      <td>6.278800</td>\n",
       "      <td>5.847458</td>\n",
       "      <td>6.821092</td>\n",
       "      <td>6.155012</td>\n",
       "      <td>6.288384</td>\n",
       "      <td>6.868152</td>\n",
       "      <td>6.500591</td>\n",
       "      <td>6.037955</td>\n",
       "      <td>...</td>\n",
       "      <td>3.843195</td>\n",
       "      <td>2.413887</td>\n",
       "      <td>2.352475</td>\n",
       "      <td>3.609278</td>\n",
       "      <td>2.372301</td>\n",
       "      <td>6.129119</td>\n",
       "      <td>2.37833</td>\n",
       "      <td>6.909919</td>\n",
       "      <td>2.434913</td>\n",
       "      <td>3.156666</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 711 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   hsa-miR-5000-3p  hsa-miR-191-5p  hsa-miR-380-5p  hsa-miR-4474-5p  \\\n",
       "0         6.877273        6.930538        6.306224         5.805400   \n",
       "1         7.604156        6.801036        6.278800         5.847458   \n",
       "\n",
       "   hsa-miR-4311  hsa-miR-518f-5p  hsa-miR-19b-1-5p  hsa-miR-425-3p  \\\n",
       "0      6.670228         6.103235          6.427767        6.634423   \n",
       "1      6.821092         6.155012          6.288384        6.868152   \n",
       "\n",
       "   hsa-miR-625-5p  hsa-miR-4655-3p  ...  hsa-miR-6796-5p  hsa-miR-98-3p  \\\n",
       "0        6.418833         6.082664  ...         3.843195       2.413887   \n",
       "1        6.500591         6.037955  ...         3.843195       2.413887   \n",
       "\n",
       "   hsa-miR-6720-3p  hsa-miR-6849-3p  hsa-miR-203b-5p  hsa-miR-6769a-5p  \\\n",
       "0         2.352475         3.609278         2.372301          6.129119   \n",
       "1         2.352475         3.609278         2.372301          6.129119   \n",
       "\n",
       "   hsa-miR-6504-3p  hsa-miR-6861-5p  hsa-miR-1251-5p  hsa-miR-219b-5p  \n",
       "0          2.37833         6.909919         2.434913         3.156666  \n",
       "1          2.37833         6.909919         2.434913         3.156666  \n",
       "\n",
       "[2 rows x 711 columns]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "extenal = pd.read_csv('20220610_Results/1. GSE58606.csv', index_col=0)\n",
    "X_train_ = pd.read_csv('20220610_Results/1. X_train.csv', index_col=0)\n",
    "len(microRNA), len(set(microRNA)&set(extenal.columns))\n",
    "extenal.head(2)\n",
    "temp = extenal[list(set(microRNA)&set(extenal.columns))+['Label']]\n",
    "for i in set(microRNA)-set(extenal.columns):\n",
    "    temp[i] = X_train_[i].median()\n",
    "extenal = temp\n",
    "extenal.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "528d060b-5862-4b18-8374-a20203b37694",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_extenal = torch.from_numpy(extenal[microRNA].values).type(torch.FloatTensor)\n",
    "y_extenal = torch.from_numpy(extenal['Label'].values).type(torch.LongTensor)\n",
    "extenal_data_loader = TorchData.DataLoader(dataset=TorchData.TensorDataset(X_extenal, y_extenal), \n",
    "                                           batch_size=X_extenal.shape[0], shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "bf722d9c-e97b-4ac3-b088-44b69588fa3a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3d46f331defc4968b99f24a02743f992",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Testing: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────\n",
      "       Test metric             DataLoader 0\n",
      "────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────\n",
      "         val_aps            0.9617503582370652\n",
      "         val_auc            0.6669150521609538\n",
      "         val_bac                    0.5\n",
      "        val_loss            0.39883047342300415\n",
      "         val_pre            0.9172932330827067\n",
      "         val_rec                    1.0\n",
      "────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[{'val_loss': 0.39883047342300415,\n",
       "  'val_bac': 0.5,\n",
       "  'val_auc': 0.6669150521609538,\n",
       "  'val_aps': 0.9617503582370652,\n",
       "  'val_rec': 1.0,\n",
       "  'val_pre': 0.9172932330827067}]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trainer.test(model, extenal_data_loader)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "0f2de48f-4b49-4d48-8033-adcde3085cbc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ef0af4b7fadd46a3aeda165747b87406",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Predicting: 0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "[(tensor([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "          1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "          1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "          1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "          1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "          1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]),\n",
       "  tensor([0.9939, 0.9933, 0.9931, 0.9927, 0.9917, 0.9885, 0.9942, 0.9912, 0.9917,\n",
       "          0.9886, 0.9933, 0.9887, 0.9964, 0.9932, 0.9939, 0.9907, 0.9890, 0.9912,\n",
       "          0.9929, 0.9939, 0.9919, 0.9909, 0.9929, 0.9969, 0.9933, 0.9943, 0.9916,\n",
       "          0.9937, 0.9912, 0.9914, 0.9921, 0.9927, 0.9947, 0.9968, 0.9933, 0.9895,\n",
       "          0.9930, 0.9913, 0.9916, 0.9898, 0.9916, 0.9921, 0.9947, 0.9928, 0.9922,\n",
       "          0.9915, 0.9939, 0.9931, 0.9939, 0.9944, 0.9928, 0.9927, 0.9926, 0.9927,\n",
       "          0.9929, 0.9935, 0.9965, 0.9921, 0.9931, 0.9920, 0.9929, 0.9937, 0.9956,\n",
       "          0.9938, 0.9925, 0.9921, 0.9936, 0.9926, 0.9927, 0.9934, 0.9929, 0.9961,\n",
       "          0.9951, 0.9915, 0.9906, 0.9925, 0.9924, 0.9919, 0.9928, 0.9926, 0.9936,\n",
       "          0.9934, 0.9924, 0.9910, 0.9898, 0.9909, 0.9910, 0.9932, 0.9920, 0.9910,\n",
       "          0.9934, 0.9913, 0.9917, 0.9921, 0.9923, 0.9908, 0.9953, 0.9876, 0.9900,\n",
       "          0.9904, 0.9849, 0.9865, 0.9893, 0.9882, 0.9909, 0.9878, 0.9940, 0.9954,\n",
       "          0.9927, 0.9932, 0.9953, 0.9918, 0.9833, 0.9893, 0.9886, 0.9899, 0.9960,\n",
       "          0.9919, 0.9929, 0.9898, 0.9932, 0.9949, 0.9929, 0.9827, 0.9873, 0.9942,\n",
       "          0.9902, 0.9937, 0.9919, 0.9907, 0.9950, 0.9913, 0.9914]))]"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trainer.predict(model, extenal_data_loader)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d1083b6b-0600-4fa8-9a5c-09c28e191d4c",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}