example_notebook.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "#!/usr/bin/env python\n",
    "# coding: utf-8\n",
    "# Imports\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import f1_score, accuracy_score\n",
    "from sklearn.metrics import roc_curve, confusion_matrix\n",
    "import torch\n",
    "import torch.nn as nn  # All neural network modules, nn.Linear, nn.Conv2d, BatchNorm, Loss functions\n",
    "import torch.optim as optim  # For all Optimization algorithms, SGD, Adam, etc.\n",
    "import torch.nn.functional as F  # All functions that don't have any parameters\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training set shape: 4180 420 54\n",
      "val set shape: 1526 420 54\n",
      "Percent positive samples in train: 25.0\n",
      "Percent positive samples in val: 24.901703800786372\n",
      "\n",
      "NOTE:\n",
      "Setting batch-size to 64\n",
      "Using device (CPU/GPU): cpu\n"
     ]
    }
   ],
   "source": [
    "###############################\n",
    "###    Load data            ###\n",
    "###############################\n",
    "\n",
    "data_list = []\n",
    "target_list = []\n",
    "\n",
    "import glob\n",
    "for fp in glob.glob(\"data/train/*input.npz\"):\n",
    "    data = np.load(fp)[\"arr_0\"]\n",
    "    targets = np.load(fp.replace(\"input\", \"labels\"))[\"arr_0\"]\n",
    "    \n",
    "    data_list.append(data)\n",
    "    target_list.append(targets)\n",
    "#print(data_list)\n",
    "\n",
    "# Note:\n",
    "# Choose your own training and val set based on data_list and target_list\n",
    "# Here using the last partition as val set\n",
    "\n",
    "X_train = np.concatenate(data_list[ :-1])\n",
    "y_train = np.concatenate(target_list[:-1])\n",
    "nsamples, nx, ny = X_train.shape\n",
    "print(\"Training set shape:\", nsamples,nx,ny)\n",
    "\n",
    "X_val = np.concatenate(data_list[-1: ])\n",
    "y_val = np.concatenate(target_list[-1: ])\n",
    "nsamples, nx, ny = X_val.shape\n",
    "print(\"val set shape:\", nsamples,nx,ny)\n",
    "\n",
    "p_neg = len(y_train[y_train == 1])/len(y_train)*100\n",
    "print(\"Percent positive samples in train:\", p_neg)\n",
    "\n",
    "p_pos = len(y_val[y_val == 1])/len(y_val)*100\n",
    "print(\"Percent positive samples in val:\", p_pos)\n",
    "\n",
    "# make the data set into one dataset that can go into dataloader\n",
    "train_ds = []\n",
    "for i in range(len(X_train)):\n",
    "    train_ds.append([np.transpose(X_train[i]), y_train[i]])\n",
    "\n",
    "val_ds = []\n",
    "for i in range(len(X_val)):\n",
    "    val_ds.append([np.transpose(X_val[i]), y_val[i]])\n",
    "\n",
    "bat_size = 64\n",
    "print(\"\\nNOTE:\\nSetting batch-size to\", bat_size)\n",
    "train_ldr = torch.utils.data.DataLoader(train_ds,batch_size=bat_size, shuffle=True)\n",
    "val_ldr = torch.utils.data.DataLoader(val_ds,batch_size=bat_size, shuffle=True)\n",
    "\n",
    "\n",
    "# Set device\n",
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "print(\"Using device (CPU/GPU):\", device)\n",
    "#device = torch.device(\"cpu\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Initializing network\n"
     ]
    }
   ],
   "source": [
    "\n",
    "###############################\n",
    "###    Define network       ###\n",
    "###############################\n",
    "\n",
    "print(\"Initializing network\")\n",
    "\n",
    "# Hyperparameters\n",
    "input_size = 420\n",
    "num_classes = 1\n",
    "learning_rate = 0.01\n",
    "\n",
    "class Net(nn.Module):\n",
    "    def __init__(self,  num_classes):\n",
    "        super(Net, self).__init__()   \n",
    "        self.bn0 = nn.BatchNorm1d(54)\n",
    "        self.conv1 = nn.Conv1d(in_channels=54, out_channels=100, kernel_size=3, stride=2, padding=1)\n",
    "        torch.nn.init.kaiming_uniform_(self.conv1.weight)\n",
    "        self.pool = nn.MaxPool1d(kernel_size=2, stride=2)\n",
    "        self.conv1_bn = nn.BatchNorm1d(100)\n",
    "        \n",
    "        self.conv2 = nn.Conv1d(in_channels=100, out_channels=100, kernel_size=3, stride=2, padding=1)\n",
    "        torch.nn.init.kaiming_uniform_(self.conv2.weight)\n",
    "        self.conv2_bn = nn.BatchNorm1d(100)\n",
    "        \n",
    "        self.fc1 = nn.Linear(2600, num_classes)\n",
    "        torch.nn.init.xavier_uniform_(self.fc1.weight)\n",
    "        \n",
    "    def forward(self, x):   \n",
    "        x = self.bn0(x)\n",
    "        x = self.pool(F.relu(self.conv1(x)))\n",
    "        x = self.conv1_bn(x)\n",
    "        \n",
    "        x = self.pool(F.relu(self.conv2(x)))\n",
    "        x = self.conv2_bn(x)\n",
    "        \n",
    "        x = x.view(x.size(0), -1)\n",
    "        x = torch.sigmoid(self.fc1(x))\n",
    "        \n",
    "        return x\n",
    "    \n",
    "# Initialize network\n",
    "net = Net(num_classes=num_classes).to(device)\n",
    "\n",
    "# Loss and optimizer\n",
    "criterion = nn.BCELoss()\n",
    "optimizer = optim.SGD(net.parameters(), lr=learning_rate)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training\n",
      "\n",
      "Epoch: 1\n",
      "Training loss: 0.009340314194560051 Validation loss: 0.01131409127265215\n",
      "MCC Train: 0.19240144331224063 MCC val: 0.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/admin/opt/anaconda3/envs/py37/lib/python3.7/site-packages/sklearn/metrics/classification.py:872: RuntimeWarning: invalid value encountered in double_scalars\n",
      "  mcc = cov_ytyp / np.sqrt(cov_ytyt * cov_ypyp)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Epoch: 2\n",
      "Training loss: 0.008121270686388016 Validation loss: 0.016945263370871544\n",
      "MCC Train: 0.3511904816756956 MCC val: 0.0\n",
      "\n",
      "Epoch: 3\n",
      "Training loss: 0.007594200316816568 Validation loss: 0.008494502864778042\n",
      "MCC Train: 0.4219054819559676 MCC val: 0.3975353268677228\n",
      "\n",
      "Epoch: 4\n",
      "Training loss: 0.007058790884912014 Validation loss: 0.011326366104185581\n",
      "MCC Train: 0.4948193585274027 MCC val: 0.29086902469870035\n",
      "\n",
      "Epoch: 5\n",
      "Training loss: 0.006774592213332653 Validation loss: 0.007119739893823862\n",
      "MCC Train: 0.5021554498324791 MCC val: 0.5489242182510553\n"
     ]
    }
   ],
   "source": [
    "###############################\n",
    "###         TRAIN           ###\n",
    "###############################\n",
    "\n",
    "print(\"Training\")\n",
    "\n",
    "num_epochs = 5\n",
    "\n",
    "train_acc, train_loss = [], []\n",
    "valid_acc, valid_loss = [], []\n",
    "losses = []\n",
    "val_losses = []\n",
    "\n",
    "for epoch in range(num_epochs):\n",
    "    cur_loss = 0\n",
    "    val_loss = 0\n",
    "    \n",
    "    net.train()\n",
    "    train_preds, train_targs = [], [] \n",
    "    for batch_idx, (data, target) in enumerate(train_ldr):\n",
    "        X_batch =  data.float().detach().requires_grad_(True)\n",
    "        target_batch = torch.tensor(np.array(target), dtype = torch.float).unsqueeze(1)\n",
    "        \n",
    "        optimizer.zero_grad()\n",
    "        output = net(X_batch)\n",
    "        \n",
    "        batch_loss = criterion(output, target_batch)\n",
    "        batch_loss.backward()\n",
    "        optimizer.step()\n",
    "        \n",
    "        preds = np.round(output.detach().cpu())\n",
    "        train_targs += list(np.array(target_batch.cpu()))\n",
    "        train_preds += list(preds.data.numpy().flatten())\n",
    "        cur_loss += batch_loss.detach()\n",
    "\n",
    "    losses.append(cur_loss / len(train_ldr.dataset))        \n",
    "    \n",
    "    net.eval()\n",
    "    ### Evaluate validation\n",
    "    val_preds, val_targs = [], []\n",
    "    with torch.no_grad():\n",
    "        for batch_idx, (data, target) in enumerate(val_ldr): ###\n",
    "            x_batch_val = data.float().detach()\n",
    "            y_batch_val = target.float().detach().unsqueeze(1)\n",
    "            \n",
    "            output = net(x_batch_val)\n",
    "            \n",
    "            val_batch_loss = criterion(output, y_batch_val)\n",
    "            \n",
    "            preds = np.round(output.detach())\n",
    "            val_preds += list(preds.data.numpy().flatten()) \n",
    "            val_targs += list(np.array(y_batch_val))\n",
    "            val_loss += val_batch_loss.detach()\n",
    "            \n",
    "        val_losses.append(val_loss / len(val_ldr.dataset))\n",
    "        print(\"\\nEpoch:\", epoch+1)\n",
    "        \n",
    "        train_acc_cur = accuracy_score(train_targs, train_preds)  \n",
    "        valid_acc_cur = accuracy_score(val_targs, val_preds) \n",
    "\n",
    "        train_acc.append(train_acc_cur)\n",
    "        valid_acc.append(valid_acc_cur)\n",
    "        \n",
    "        from sklearn.metrics import matthews_corrcoef\n",
    "        print(\"Training loss:\", losses[-1].item(), \"Validation loss:\", val_losses[-1].item(), end = \"\\n\")\n",
    "        print(\"MCC Train:\", matthews_corrcoef(train_targs, train_preds), \"MCC val:\", matthews_corrcoef(val_targs, val_preds))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## MH"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MCC Train: 0.5021554498324791\n",
      "MCC Test: 0.5489242182510553\n",
      "Confusion matrix train:\n",
      "[[2941  194]\n",
      " [ 524  521]]\n",
      "Confusion matrix test:\n",
      "[[1130   16]\n",
      " [ 221  159]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY4AAAEGCAYAAABy53LJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nO3de5zM9f7A8dfbWtb9slRYWqWbe8txdKeLQ05IOijJLSddJdH11HHUryRJuimp5LgccqlIiS4qhKJI5VZtlCX3S+z6/P54D9aa3Z3ZnZnv7Mz7+XjMY2dnvvOd987uzns+t/dHnHMYY4wxgSrmdQDGGGOKFkscxhhjgmKJwxhjTFAscRhjjAmKJQ5jjDFBKe51AJFQpUoVl5qa6nUYxhhTpCxbtmyrc65qztvjInGkpqaydOlSr8MwxpgiRUR+8ne7dVUZY4wJiiUOY4wxQbHEYYwxJihxMcZhjAm/Q4cOkZ6ezoEDB7wOxQQpKSmJlJQUEhMTAzreEocxJiTS09MpV64cqampiIjX4ZgAOefYtm0b6enp1K5dO6DHWFeVMSYkDhw4QHJysiWNIkZESE5ODqqlaInDGBMyljSKpmB/b5Y4TMgsWQKLF3sdhTEm3CxxmJBYtw4uvRT+/nfYt8/raEw82rZtG40bN6Zx48accsop1KhR4+j3Bw8eDOgcPXv25Pvvvw/4OV955RX69+9f0JCLLBscN4WWlQXdu+vXrVth3Di49VavozLxJjk5ma+//hqARx55hLJlyzJw4MDjjnHO4ZyjWDH/n5nHjRsX9jhjgbU4TKENGwaffw6vvALnnQfDh0NmptdRGaPWrl1L/fr1ufnmm0lLS2Pz5s307duXpk2bUq9ePYYMGXL02AsvvJCvv/6azMxMKlasyL333kujRo0477zz2LJlS8DP+eabb9KgQQPq16/P/fffD0BmZiY33HDD0dtHjRoFwNNPP03dunVp1KgR3bp1C+0PHybW4jCF8tVX8K9/QefOcN11UKYMXH01TJ0KXbp4HZ3xTP/+4Pv0HzKNG8PIkQV66OrVqxk3bhwvvvgiAI8//jiVK1cmMzOTli1b0qlTJ+rWrXvcY3bu3Mkll1zC448/zoABA3j11Ve59957832u9PR0HnzwQZYuXUqFChW4/PLLeeedd6hatSpbt27lm2++AWDHjh0ADBs2jJ9++okSJUocvS3aWYvDFNj+/dCtG5x0Ejz/PIhAu3Zw1lnwxBNg29mbaHH66afzl7/85ej3EydOJC0tjbS0NL777jtWr159wmNKlSpFmzZtAGjSpAkbN24M6LkWL17MpZdeSpUqVUhMTOS6667jk08+oU6dOnz//ffceeedzJ07lwoVKgBQr149unXrxoQJEwJegOc1a3GYArv/fli9Gt5/HypX1tuKFYN77oE+fWDePLjiCm9jNB4pYMsgXMqUKXP0+o8//sgzzzzDkiVLqFixIt26dfO7hqFEiRJHryckJJAZYP+ry+UTU3JyMitXrmTOnDmMGjWKadOmMWbMGObOncvHH3/MzJkzGTp0KN9++y0JCQlB/oSRZS0OUyAffqjvDbfffmJy6NYNqlfXVocx0WbXrl2UK1eO8uXLs3nzZubOnRvS8zdv3pwFCxawbds2MjMzmTRpEpdccgkZGRk457j22mv597//zfLly8nKyiI9PZ1LL72UJ598koyMDPYVgWmJ1uIwQdu+HXr0gLPPhscfP/H+kiW1i3vQIFi2DJo0iXiIxuQqLS2NunXrUr9+fU477TQuuOCCQp1v7NixTJ069ej3S5cuZciQIbRo0QLnHFdddRVt27Zl+fLl9O7dG+ccIsITTzxBZmYm1113Hbt37+bw4cMMHjyYcuXKFfZHDDvJrVkVS5o2bepsI6fQuf56mDIFvvgCmjb1f8zOnVCrFrRuDZMnRzY+443vvvuOc845x+swTAH5+/2JyDLn3An/5dZVZYIyaRL897/w8MO5Jw2AChWgXz+dXbVuXeTiM8aEnyUOE7Bff9Vk0Lw5BDArkTvvhOLF4amnwh+bMSZyLHGYgBw+DD17wsGDMH68JoT8VKumK8rHjYMg1k4ZY6KcJQ4TkOeegw8+gBEjoE6dwB83cCD8+Sf4FskaY2KAJQ6Tr+++0xlSV14JffsG99izzoIOHTTx7NkTnviMMZFlicPk6eBBXZdRpgyMHaurw4M1eDDs2AEvvxz6+IwxkRfWxCEirUXkexFZKyInDKeKSEkRmey7f7GIpPpuTxaRBSKyR0RG53hMCREZIyI/iMgaEbkmnD9DvPvPf2D5chgzBk45pWDn+Otf4eKLtZvr0KHQxmfMES1atDhhMd/IkSO55ZZb8nxc2bJlAdi0aROdOnXK9dz5TekfOXLkcYv3rrzyypDUnnrkkUcYPnx4oc8TSmFLHCKSADwHtAHqAl1FpG6Ow3oD251zdYCngSNrjQ8ADwEDOdEDwBbn3Jm+834chvANuk7jscd0sV/HjoU71+DBkJ4OEyeGJDRjTtC1a1cmTZp03G2TJk2ia9euAT2+evXqxy3kC1bOxDF79mwqVqxY4PNFs3C2OJoBa51z651zB4FJQPscx7QHXvddnwpcJiLinNvrnFuIJpCcegH/B+CcO+yc2xqe8OPbnj1www1QsyY880zhz9emDdSvryXYDx8u/PmMyalTp0688847/PnnnwBs3LiRTZs2ceGFF7Jnzx4uu+wy0tLSaNCgATNnzjzh8Rs3bqR+/foA7N+/ny5dutCwYUM6d+7M/v37jx7Xr1+/oyXZH374YQBGjRrFpk2baNmyJS1btgQgNTWVrVv17WnEiBHUr1+f+vXrM9JXx2vjxo2cc8453HTTTdSrV49WrVod9zz58XfOvXv30rZtWxo1akT9+vWZ7Ft9e++991K3bl0aNmx4wh4lBRHOkiM1gF+yfZ8O/DW3Y5xzmSKyE0gG/CYDETmSvv8jIi2AdcBtzrnfQxi3Ae6+G9avh48+gvLlC38+ER1g794d5syBtm0Lf04Tvbyoqp6cnEyzZs147733aN++PZMmTaJz586ICElJSUyfPp3y5cuzdetWmjdvTrt27XLda/uFF16gdOnSrFy5kpUrV5KWlnb0vkcffZTKlSuTlZXFZZddxsqVK7njjjsYMWIECxYsoEqVKseda9myZYwbN47FixfjnOOvf/0rl1xyCZUqVeLHH39k4sSJvPzyy/zjH/9g2rRpAe3Jkds5169fT/Xq1Xn33XcBLQ3/xx9/MH36dNasWYOIhKT7LJwtDn+/kZz1TQI5JrviQArwmXMuDfgC8Nv5JyJ9RWSpiCzNyMgIJF7j8847OqZxzz06NhEqXbpoC8aKH5pwyd5dlb2byjnH/fffT8OGDbn88sv59ddf+f333D9vfvLJJ0ffwBs2bEjDhg2P3jdlyhTS0tI499xzWbVqld+S7NktXLiQq6++mjJlylC2bFk6duzIp59+CkDt2rVp3LgxEFzp9tzO2aBBA+bNm8fgwYP59NNPqVChAuXLlycpKYk+ffrw1ltvUbp06YCeIy/hbHGkAzWzfZ8CbMrlmHQRKQ5UAP7I45zbgH3AdN/3/0PHSU7gnBsDjAGtVRVs8PEqIwN694aGDSHbxmghkZgIAwbAXXfp+Ml554X2/CZ6eFVVvUOHDgwYMIDly5ezf//+oy2FCRMmkJGRwbJly0hMTCQ1NdVvKfXs/LVGNmzYwPDhw/nyyy+pVKkSPXr0yPc8edUDLFmy5NHrCQkJAXdV5XbOM888k2XLljF79mzuu+8+WrVqxb/+9S+WLFnChx9+yKRJkxg9ejTz588P6HlyE84Wx5fAGSJSW0RKAF2AWTmOmQXc6LveCZjv8niVffe9DbTw3XQZkHe6NwFzTtdp7NgBb76pVW5DrU8fqFRJxzqMCbWyZcvSokULevXqddyg+M6dOznppJNITExkwYIF/PTTT3me5+KLL2bChAkAfPvtt6xcuRLQkuxlypShQoUK/P7778yZM+foY8qVK8fu3bv9nmvGjBns27ePvXv3Mn36dC666KJC/Zy5nXPTpk2ULl2abt26MXDgQJYvX86ePXvYuXMnV155JSNHjjy6L3thhK3F4RuzuA2YCyQArzrnVonIEGCpc24WMBYYLyJr0ZbG0c1GRWQjUB4oISIdgFbOudXAYN9jRgIZQM9w/Qzx5rXXYMYM3TO8QYPwPEfZsnDbbTB0KKxZo6XZjQmlrl270rFjx+NmWF1//fVcddVVNG3alMaNG3N2Pn94/fr1o2fPnjRs2JDGjRvTrFkzABo1asS5555LvXr1TijJ3rdvX9q0aUO1atVYsGDB0dvT0tLo0aPH0XP06dOHc889N+BuKYChQ4ceHQAH3Z7W3znnzp3LPffcQ7FixUhMTOSFF15g9+7dtG/fngMHDuCc4+mnnw74eXNjZdUNABs2aPdU06a6SVOxMLZFt2yBU0/V8uyvvBK+5zGRZWXVizYrq26CkpWls52KFdNWRziTBuge5b16abHETTlHvYwxUc8Sh2H4cFi4EEaP1pZAJNx9N2RmhmaNiDEmsixxxLmvv4aHHoJOnbQmVaScdhpcey28+KLuFmhiQzx0fceiYH9vljji2IEDmiyqVNE38IIUMCyMQYNg1y59blP0JSUlsW3bNkseRYxzjm3btpGUlBTwY8K5jsNEuQcegFWrdCV3cnLknz8tDS6/XOf89+8fnum/JnJSUlJIT0/HFtwWPUlJSaSkpAR8vCWOODV/vlarveUWaN3auzgGD4YrrtCB8j59vIvDFF5iYiK1a9f2OgwTATYdNw7t2KFTb0uV0pLpZcp4F4tz0KQJ7N2rG0aFe0aXMSZwNh3XHHX77ToNdvx4b5MG6LjK4MHwww/gp2CpMSYKWeKIM1OmaDmRhx4C36JTz11zDdSurcUP46ABbEyRZ4kjjvz6K9x8syaM++/3OppjiheHgQNh8WLwFQ01xkQxSxxxwjldrX3ggHZRJSZ6HdHxevaEqlWt5LoxRYEljjjx/PPw/vvw1FNw5pleR3OiUqV07GX2bPjmG6+jMcbkxRJHHFizRjdlat1au6qi1a23QunS8OSTXkdijMmLJY4Yd+iQ7h1eqhS8+mrkV4cHo3JluOkmmDgRfv7Z62iMMbmxxBHjhg6FpUt1K9hq1byOJn8DBujXEGwZYIwJE0scMWzRInj0US2Zfs01XkcTmFq1oGtXePll+COvTYSNMZ6xxBGj9u7VLqoaNWDUKK+jCc4992j8zz/vdSTGGH8sccSogQNh3Tp44w2oUMHraILToAFceaUmvP37vY7GGJOTJY4YNHu2liq/+2645BKvoymYQYMgIwPGjfM6EmNMTlbkMMZs3Qr16+v2rF9+WXRLlTsH552nyeP773V1uTEmsqzIYRxwDvr2he3btR5VUU0acKz44fr1MG2a19EYY7KzxBFD3ngDpk/XKbgNG3odTeG1a6er3IcNs+KHxkQTSxwxYuNGLdlx8cXH1kIUdQkJOsNq+XL48EOvozHGHGGJIwZkZelaDYDXX9c33FjRrRuccooVPzQmmljiiAEjRmg58mefhdRUr6MJraQk3Y983jxteRhjvGeJo4hbsQIeeAA6djzW6og1N98M5crpWIcxxnuWOIqwAwd0dXjlyvDSS9FdwLAwKlTQ5PG//+ksK2OMtyxxFGEPPaR7V7z6KlSp4nU04dW/v47dPPWU15EYYyxxFFEff6xvojffrOU5Yl316tq6evVV2LLF62iMiW+WOIqgnTt1POP002H4cK+jiZx77oE//4TRo72OxJj4FtbEISKtReR7EVkrIvf6ub+kiEz23b9YRFJ9tyeLyAIR2SMift8mRGSWiHwbzvij1Z13wq+/6urwMmW8jiZyzj4b2rfXxLFnj9fRGBO/wpY4RCQBeA5oA9QFuopI3RyH9Qa2O+fqAE8DR2brHwAeAgbmcu6OQFy+dUybpms1HngA/vpXr6OJvEGDtKTK2LFeR2JM/Apni6MZsNY5t945dxCYBLTPcUx74HXf9anAZSIizrm9zrmFaAI5joiUBQYAQ8MXenTavBn++U9o2hQefNDraLxx3nlw0UW6duXQIa+jMSY+hTNx1AB+yfZ9uu82v8c45zKBnUByPuf9D/AUsC+vg0Skr4gsFZGlGRkZwcQdlZyD3r1h3z7tokpM9Doi7wwapHuST5rkdSTGxKdwJg5/qwpylqoL5JhjB4s0Buo456bn9+TOuTHOuabOuaZVq1bN7/Co99JLMGcOPPkknHWW19F468oroV49K35ojFfCmTjSgZrZvk8BNuV2jIgUByoAee00fR7QREQ2AguBM0XkoxDFG7V++EE3Zfrb3+CWW7yOxnvFimmr49tvNZkaYyIrnInjS+AMEaktIiWALsCsHMfMAm70Xe8EzHd57CzlnHvBOVfdOZcKXAj84JxrEfLIo0hmpq5fKFlS1zDE6urwYHXpAikpVobEGC+ELXH4xixuA+YC3wFTnHOrRGSIiLTzHTYWSBaRteiA99Epu75WxQigh4ik+5mRFRceewyWLNGuqurVvY4mepQooeXjP/4YFi/2Ohpj4ottHRvFvvxSZxF17Qrjx3sdTfTZvRtq1YJLL7VdAo0JB9s6tojZt0/3oqheXculmxOVKwe33qq7Hn7/vdfRGBM/LHFEqUGDdFD89dehYkWvo4led9yh3VbxVHrFGK9Z4ohC770Hzz2nffgtW3odTXQ76STo2VP3W9+82etojIkPljiizLZt0KuXrlN49FGvoyka7r5bZ58984zXkRgTHyxxRBHntEz61q26OjwpyeuIioY6deCaa+CFF2DXLq+jMSb2WeKIIhMmwNSp8J//QOPGXkdTtAwerEnjpZe8jsSY2GfTcaPEzz9DgwbQsCF89JHudmeCc/nlsHo1bNigCyaNMYVj03Gj2OHDcOON+vX11y1pFNSgQTpAPmGC15EYE9sscUSBkSO1lfHMM3DaaV5HU3RdcYV28Q0bpknYGBMeljg89u23cN99urNdz55eR1O0iWir4/vv4e23vY7GmNhlicNDf/6pq8MrVoQxY6yAYShcey2kpsITT1jJdWPCxRKHhx5+GFasgFde0YVspvCKF9d1HV98AQsXeh2NMbHJEodHPv1U++JvugmuusrraGJLr15QpYqVXDcmXCxxeGDXLujeHWrX1r2zTWiVLg233w7vvKNjSMaY0LLE4YH+/XXdxvjxULas19HEpltv1QRixQ+NCT1LHBE2fTqMG6czqc4/3+toYldyMvTpo2s6fvnF62iMiS2WOCLot9+gb19IS4N//cvraGLfXXfpzKqnn/Y6EmNiiyWOCHFOPwHv2aNdVCVKeB1R7EtN1b3Jx4yB7du9jsaY2GGJI0JefhnefVfXF9SNy93TvTFoEOzdC88/73UkxsQOSxwRsHatdptcfjncdpvX0cSXhg2hdWsYNQr27/c6GmNigyWOMMvMhBtu0K6pceOgmL3iETd4MGzZogUkjTGFZ29jYfb447BokW4ylJLidTTx6ZJL4C9/0am5WVleR2NM0WeJI4yWLoV//xu6dtVBWuMNEW11rFsHb73ldTTGFH22kVOY7NsHTZrA7t3wzTdQqVJEn97kkJUF55wD5cvDl19aQUljAmEbOUXYvffCmjXw2muWNKJBQgIMHAjLlsGCBV5HY0zRZokjDN5/H559Fu68U2dSmejQvTucfLJOiTbGFJwljhD74w/dkOmcc+D//s/raEx2SUmazN9/H776yutojCm6LHGEkHPQr59O/XzzTShVyuuITE79+kG5cvDkk15HYkzRFVDiEJHTRaSk73oLEblDRCqGN7SiZ+JEmDJFZ1KlpXkdjfGnYkX45z9h8mTYsMHraIwpmgJtcUwDskSkDjAWqA38N2xRFUG//AK33KIVbwcN8joak5f+/XWw3PZCMaZgAk0ch51zmcDVwEjn3F1AtfweJCKtReR7EVkrIvf6ub+kiEz23b9YRFJ9tyeLyAIR2SMio7MdX1pE3hWRNSKySkQeDzD+sDp8GHr00FXib7yh25ea6FWjhu71PnYsZGR4HY0xRU+gieOQiHQFbgTe8d2WmNcDRCQBeA5oA9QFuopIzvJ+vYHtzrk6wNPAkfkuB4CHgIF+Tj3cOXc2cC5wgYi0CfBnCJtRo2D+fBg5Ek4/3etoTCDuuUdrV40enf+xxpjjBZo4egLnAY865zaISG3gzXwe0wxY65xb75w7CEwC2uc4pj1wpILQVOAyERHn3F7n3EI0gRzlnNvnnFvgu34QWA54Wshj1Spds3HVVdC7t5eRmGCccw60a6eJY+9er6MxpmgJKHE451Y75+5wzk0UkUpAOedcft1ENYDse6+l+27ze4yvK2wnkBxITL7B+auAD3O5v6+ILBWRpRlh6o84eFC7PMqX17Lpthq5aBk8WKdPjx3rdSTGFC2Bzqr6SETKi0hlYAUwTkTyG1r09zaas75JIMf4i6c4MBEY5Zxb7+8Y59wY51xT51zTqlWr5nfKAnnkEfj6a00aJ58clqcwYXT++XDBBfDUU3DokNfRGFN0BNpVVcE5twvoCIxzzjUB8lsTnQ7UzPZ9CrApt2N8yaAC8EcA8YwBfnTOjQzg2LBYuFBXIPfuDe1zdsCZImPwYPj5Z51GbYwJTKCJo7iIVAP+wbHB8fx8CZwhIrVFpATQBZiV45hZ6IA7QCdgvsun6qKIDEUTTP8A4wi53bu1fMWpp9p+1kVd27a6I+OwYbqA0xiTv0ATxxBgLrDOOfeliJwG/JjXA3xjFrf5HvcdMMU5t0pEhohIO99hY4FkEVkLDACOTtkVkY3ACKCHiKSLSF0RSQEeQGdpLReRr0WkT6A/bKjcdRf89JPuHV6uXKSf3YRSsWI6w2rlSpg71+tojCkarKx6kGbOhA4d4L774LHHQnJK47GDB+G00+CMM6xyrjHZFaqsuoikiMh0EdkiIr+LyDTfp/+48vvvcNNN0LixDoyb2FCihLYiP/oIlizxOhpjol+gXVXj0PGI6ugU2rd9t8UN5zRp7NqlBQxLlPA6IhNKfftChQo61mGMyVugiaOqc26ccy7Td3kNCM8c1yg1diy8/bbuIV6vntfRmFArV05rjb31Fvzwg9fRGBPdAk0cW0Wkm4gk+C7dgG3hDCyarFunhfEuvRTuuMPraEy43HmntiSfesrrSIyJboEmjl7oVNzfgM3o1Nme4QoqmmRmwg03aOHC117TWTgmNp18sharfP11+O03r6MxJnoFWnLkZ+dcO+dcVefcSc65DuhiwJg3bBh88QU8/zzUrJn/8aZou/tunWU1apTXkRgTvQrz+XlAyKKIUsuXw8MPQ+fO0LWr19GYSDjjDLjmGv2gsGuX19GY7LZvhzVrvI7CQOESR0yX9Dt0SAsYnnSSvolYAcP4MWgQ7NwJY8Z4HYk5Yt48nZTSoAHMmeN1NKYwiSOmVw4mJsKjj+rGTJUrex2NiaS//AVattRyMgcPeh1NfDt4UOuJtWql2/7Wr68twoULvY4svuWZOERkt4js8nPZja7piGlXXw2XXeZ1FMYLgwfDpk0wYYLXkcSvtWu1evGwYbqGaulSeP99qFUL/v53rUxtvJFn4nDOlXPOlfdzKeecsw1STcxq1QoaNYInn9StgU1kjR8P556rU+GnTYOXXoLSpaFqVU0e5cvD3/4GP+ZZMc+Ei00uNcYPER3r+O47eCfQetCm0Hbt0rHF7t0hLQ1WrICOOeZv1qoFH3ygCf3yyyE93ZtY45klDmNy8Y9/aOn8J57wOpL4sHixtjImToQhQ2D+/NynwJ91llYz3rEDrrgCtm6NbKzxzhKHMbkoXlzXdXz+OXz2mdfRxK7Dh7WUz4UX6oLbTz6Bhx6ChIS8H5eWpmWANm6ENm1s+nQkWeIwJg+9ekFysrU6wmXTJm0x3HefdkmtWKED4oG6+GKYOlUHytu3hwMHwherOcYShzF5KFMGbrtNP9muXu11NLHl7behYUNYtEiLiE6apFNug9W2rZaJ+fhjXaxr+8eHnyUOY/Jx221QqpTOsDKFt3+/vqbt2ukYxvLl2rIrzCLb666D0aNh1izo3dtmwoWbJQ5j8lGlir4ZTZhgM3gKa9UqaNYMnntON89atEgHukPhlltg6FCdytu/v+0hH06WOIwJwN1366fYkSO9jqRocg5efBGaNoUtW7RsyIgRULJkaJ/n/vthwAB49lmdmWXCwxKHMQFITdXpuS+9pMX2TOC2bdOB73794JJLYOVKaN06PM8lAsOHQ8+eur3zM8+E53ninSUOYwI0aBDs2aOfnE1gPvpIV+C/+65ukDV7tu57Ek4iWqCyY0ftsnrjjfA+XzyyxGFMgBo31jIXzzxj0z7zc+gQPPig7ppZpoyOZQwYELmN0IoXh//+V2vN9eoFM2dG5nnjhSUOY4IwaBD8/rtO/zT+bdig6ysefVR3VFy2TBfrRVrJkjBjho6rdO4MCxZEPoZYZYnDmCC0bKlvRMOHQ1aW19FEn0mTtGW2erVef/VVKFvWu3jKltVusjp1dPrvl196F0ssscRhTBCOFD9cuxamT/c6muixZ48OSHftqhsuff21fsqPBsnJWlG3alUtTWILOQvPEocxQerYUT/BDhtmawVAF/ClpWn33YMPaq2p2rW9jup41atrRd3ERC2Zv3Gj1xEVbZY4jAlSQgIMHKjdHh995HU03jl8WGdKNW+uq8EXLID//EcHpqPR6adry2PvXq2P9fvvXkdUdFniMKYAunfX/ejjtfjhb79pt8/Agbob34oVukYj2jVooFOCN23SGXI7dngdUdFkicOYAihVCu68U/eEWLHC62gia84cXZvxySe6pmXaNKhc2euoAnfeeTo+tXq1Fkjcu9friIoeSxzGFFC/fjprZ9gwryOJjD//1PpSV16pi/iWLoV//rNwxQm90qqVrvNYtAg6dYKDB72OqGgJa+IQkdYi8r2IrBWRe/3cX1JEJvvuXywiqb7bk0VkgYjsEZHROR7TRES+8T1mlEhR/LM1saBSJejbFyZPjv3B1jVrdCxj5EitbLtkic6eKso6ddIV5sjKA1MAABgCSURBVO+9p12PNr06cGFLHCKSADwHtAHqAl1FpG6Ow3oD251zdYCngSM9xgeAh4CBfk79AtAXOMN3CVPVG2Pyd9dd+ol7xAivIwkP53SvjCZN4JdftGz5s89CUpLXkYVG795aLn/yZK2ua7PkAhPOFkczYK1zbr1z7iAwCWif45j2wJE1uFOBy0REnHN7nXML0QRylIhUA8o7575wzjngDaBDGH8GY/KUkgLXXw+vvBJ7+17v2KFrMfr00XGBlSvhqqu8jir0Bg7UHQjHjNHquiZ/4UwcNYBfsn2f7rvN7zHOuUxgJ5Cczzmz74jg75wAiEhfEVkqIkszMjKCDN2YwN1zj05Hfe45ryMJnc8+0wHw6dN1P/D339e1ELHq0Ufh5pv1Z42XMavCCGfi8Df2kLMhGMgxBTreOTfGOdfUOde0atWqeZzSmMKpV0+npD77bNGfoZOZCf/+t9aaKl5cE8jgwZErTugVEd1BsEsX/XlfftnriKJbOP8c0oGa2b5PATbldoyIFAcqAH/kc86UfM5pTMQNHqz7Towb53UkBffzz1rN9pFHdCvWr77S3friRUKCrn5v00Zni/3vf15HFL3CmTi+BM4QkdoiUgLoAszKccws4Ebf9U7AfN/YhV/Ouc3AbhFp7ptN1R2wgsnGcxdeCOefryupMzO9jiZ406Zp19RXX+n+FePHQ/nyXkcVeSVKwNSp+ru8/npdp2NOFLbE4RuzuA2YC3wHTHHOrRKRISLSznfYWCBZRNYCA4CjU3ZFZCMwAughIunZZmT1A14B1gLrgDnh+hmMCcagQTotd8oUryMJ3L59OqW4Uyc44wxNHDfc4HVU3ipdGt55R7sgO3aEzz/3OqLoI3l8wI8ZTZs2dUuXLvU6DBPjDh/WN5uSJfUNONpXGK1YodVs16zRpDdkiH7iNur33+GiiyAjAz7+GBo29DqiyBORZc65pjlvj/EhL2Mip1gxnWG1YoXOQopWzsGoUTp+sWOHVo19/HFLGjmdfLK+NmXK6ErztWu9jih6WOIwJoSuv16nrUbrlM6MDF2Lceed+ma4YoVur2r8O/VUTR6ZmVpR99dfvY4oOljiMCaESpbU1eTz52stp2jywQfa3TJvnk4dnjVLNzcyeTvnHC1Lsm2bJttt27yOyHuWOIwJsb59oUKF6Cm5fvCgjmG0aqX1tZYs0XpT0T4GE02aNtVEu26dFnncvdvriLxlicOYECtfXivnTpvmfb/42rVwwQVaj+mf/9RWUDwO8oZCixY6Y27ZMujQAQ4cyPchMcsShzFhcMcduk3p8OHePL9zuh7j3HP1U/K0abp3RunS3sQTK9q100We8+frjLSiuGYnFCxx5OWWW7TD+uOP4/cvxBRItWpw443w2muR36J01y7o1k2fPy1NB8A7doxsDLHshht0VtqMGXDTTToNO95Y4siNczoF5YUXtI16yinQo4f+tezb53V0pggYOFDHF0aNitxzLloEjRtrmfAhQ/STcc2a+T/OBOf227Wm12uvwd13x185dkscuRHRYjUZGfq1TRuYOROuvhqqVIH27bXNapV3TS7OPFP/XJ5/PvyDqVlZ8NhjWvrk8GHd1vWhh7T+kgmPhx7Sac0jR8LQoV5HE2HOuZi/NGnSxIXEwYPOzZvn3G23OVezpnPgXLFizl18sXNPPeXcunWheR4TMxYt0j+Tp54K33OkpzvXsqU+T+fOzm3fHr7nMsfLynKue3d97Z991utoQg9Y6vy8p1rJkYJyTutKzJihl2++0dsbNNDWSIcO2sFscx7jXosWOrtp/frQr86eNQt69dL9QJ59Fnr2tD+5SMvM1FpfM2fCm2/qItBYYSVHQk1EE8OQIbo12rp1un9opUraZ9C0qS47vf12XXF16JDXERuPDB6sK44nTgzdOffv17UY7dtDrVqwfLkmEEsakVe8OEyaBC1b6oSEt9/2OqLwsxZHOGzdquU1Z8zQokX790PFitC2rbZE/vY3KFcucvEYTzmnJcuzsrRhWthNkVat0g2Hvv0WBgzQzyklS4YmVlNwu3dr+ZZvvtGV5pdc4nVEhWctjkiqUuXYDKytW/Vrhw7613TttXp/27a6zdhvv3kdrQkzEV25vXo1zJ5d8PM4p5P8mjaFLVtgzhzd/8OSRnQoV05/v7Vraz2wZcu8jih8rMURSZmZWtz/yLjIhg36rtK8uSaW9u3hrLO8jtKEwaFDUKeOdit9+mnwj9+2Dfr00T+b1q11GujJJ4c8TBMC6ek6u23vXv1dn3221xEVnLU4okHx4rqZ84gROiaycqVOBv/zT+0IP/tsrah2332weHF8riyKUYmJ2q20cGHwGwN99JF2db37rv7pvPuuJY1olpKiw5oJCVpR9+efvY4o9KzFES1+/lmnyMyYoe8UWVm6/LhdO22NtGxpfRJF3N692uK46CL9Nefn0CHd//v//k9355s4UedjmKJhxQod5zj5ZG15nHSS1xEFz1oc0a5WLZ0mM2+eLiocP16r0735pi4+rFoVOnfWd4+dO72O1hRAmTJw6606bfO77/I+dsMGbZw+9phOsV22zJJGUXOklfjLLzofJpb+ba3FEe0OHIAPP9R3m5kzdVQ0MVEXB3TooC2SlBSvozQBysjQWdpdu8LYsf6PmTgRbr5Zr48Zo58XTNH13nv6b9q8uV4vSoUmrcVRVCUl6QysMWNg0yb47DPo3x82btSPrzVr6h6gjz6q8zTj4INAUVa1qq63GD/+xN3kdu/WyXjXXad7l69YYUkjFrRurR0HCxfqpMpYWNJliaMoSUiA88/XfUm//17ndz72mM7MevBBqF9fCyQNHKh/pVlZXkds/BgwQH81I0ceu23pUu2KGj9eayB98gmkpnoWogmxf/xDy9rPnq2LBIv6v6Z1VcWKTZuODa7Pn68fa6pW1TZy+/Zw+eVQqpTXURqfrl21//unn7TL6v77dRB1wgQd2zCx6Ykn4N57daOv556L/pX+uXVVWeKIRTt3amfqjBn6EWfXLu1Ybd1ax0XatoXKlb2OMq599ZW2MGrU0C6rjh11Paj9WmLf4MHaafDAA9FfVdcSRzwljuwOHtTpvTNm6OD6pk3a5XXxxccWHZ56qtdRxqUrr9RfzdNP6z7l0f7p04SGc7qN78sv6w6Rd9/tdUS5s8QRr4kju8OHtTN95kxNJKtX6+2NGx9LIo0a2TtYhOzZo3uCFcX5/aZwsrK0u/J//9Ouyl69vI7IP0scljhO9OOPx5LI55/rR6FTT9Uk0qGD1k0oXtzrKI2JSQcPak2refNgyhS45hqvIzqRJQ5LHHn7/XetBz1zJnzwgZZBqVwZ/v53TSKtWukKNmNMyOzdq2VJli3TgtpXXOF1RMezxGGJI3B79sDcudoSefdd2L5d15NccYUmkauu0hlbxphC275d1/OuW6etj+bNvY7oGEscljgK5tAhLbRzZHD95591Q4nzzz82LlKnjtdRGlOk/fab9gz/8Qd8/LFuJBoNLHFY4ig85+Drr48lkRUr9Pb69Y9tl9ukiQ2uG1MAGzZo8jh8WAtEnHaa1xF5lDhEpDXwDJAAvOKcezzH/SWBN4AmwDags3Nuo++++4DeQBZwh3Nuru/2u4A+gAO+AXo65w7kFYcljjDZsOFYDa1PPtG/+LJl9S/+9NP1a/brp54a+k23jYkhq1bpTPkKFbT4Q/Xq3sYT8cQhIgnAD8AVQDrwJdDVObc62zG3AA2dczeLSBfgaudcZxGpC0wEmgHVgXnAmcApwEKgrnNuv4hMAWY7517LKxZLHBGwdauOh3z1lXbWrl+vlwPZcnqxYlpb60giyfm1UiXv4jcmSixZolvQpqZqt5WXi0JzSxzhnGvZDFjrnFvvC2AS0B5Yne2Y9sAjvutTgdEiIr7bJznn/gQ2iMha3/l+9sVcSkQOAaWBTWH8GUygqlTRIjw33njstsOHtfN2/fpjyeTI11mztNJvdpUqndhKOfK1Zk1duGhMjGvWTBvxbdroItF587QhH03CmThqAL9k+z4d+GtuxzjnMkVkJ5Dsu31RjsfWcM59ISLD0QSyH3jfOfe+vycXkb5AX4BatWoV/qcxwStWTNva1atr521Oe/Yca5lkTyxHxlGylxFNTNSuLn+tldq1dcNnY2LEpZfC5Mm6tuPqq3WqbjTt4xbOxOFvhDRnv1hux/i9XUQqoa2R2sAO4H8i0s059+YJBzs3BhgD2lUVTOAmQsqWhYYN9ZJTVpZu3pyzpbJ+vbblt28//viqVXPvAqtWTZOYMUVIhw66qrxnTy21P3ly9KzHDWcY6UDNbN+ncGK30pFj0kWkOFAB+COPx14ObHDOZQCIyFvA+cAJicMUcQkJ2sI49VT9+JXT9u3+Wyuffw6TJh2/X3tSkrZKciaU007T261qsIlSPXrAjh1w111a3+qVV6Jj0mI4E8eXwBkiUhv4FegCXJfjmFnAjcAXQCdgvnPOicgs4L8iMgIdHD8DWAIcBpqLSGm0q+oywEa941GlSjr1t0mTE+87eFDXm+Rsraxbp1UF9+w5/vjq1XNvrVStGh3/qSZu9e+vn5OGDNE/+yef9P5PMmyJwzdmcRswF52O+6pzbpWIDAGWOudmAWOB8b7B7z/Q5ILvuCnoQHomcKtzLgtYLCJTgeW+27/C1x1lzFElSuiiRH8LE53T/Vtzdn8dWbb7+uvHH2/Ti00UeOQRXRz41FM6y+r++72NxxYAGpPd/v26La+/1sqGDTa92Hjm8GHo3l03+3r+ed0MKty8mI5rTNFTqhScc45ecjp8GDZv9j+2YtOLTZgVKwbjxum+bLfeChUraml2L1iLw5hQ2b1bWyX+WisbN0Jm5rFjs08vznk57TTdsdEYP/bv1zUen32ms9bbtg3fc1mtKkscxktZWfDLLycmlCOXnTuPP75aNf9J5fTTITnZ+9FR46ldu6BlS92L7f334aKLwvM8ljgscZho5ZyOfGZPJNkvm3LMYi9fPvekkpJiXWBxIiNDE8bmzTpZ8NxzQ/8cljgscZiiav/+E1soRy4bNx6/wr5ECS1y5C+p2JqVmPPLL1qUYf9+LYp45pmhPb8lDkscJhYd6QLLrbWye/fxx9eokXtrxctqeqbAfvhBk0dSko571KyZ/2MCZYnDEoeJN85p1eLckspvvx1/fMWKuSeVGjWsbEsU++or3UWwWjXddy1UG3Ra4rDEYczx9u7NvQvsp5+OnwVWsuSxsi2nn66LK49cT02Nrgp8cerTT6FVK6hbFxYs0KGwwrLEYYnDmMBlZh4r2+LvsnfvsWNFji2E9HepUMG7nyPOzJ6tm3FecAHMmVP4IS1LHJY4jAkN53SxY25JJedCyOTk3JNKtWo2tTjEJk6E66/X9R1vvaVLhgrKVo4bY0JDBE4+WS/nn3/i/bt3++8CW7RIa4Nnr1xcqtSxFfU5L1YLrEC6dtWKurfcoiXZ33gj9MNTljiMMaFVrhw0aqSXnA4d0vETfy2VDz7QeaVHFCsGtWqdmEySk/VSubJ+LVPGWi059OunFXVHj9ZlQCkpoT2/dVUZY6KDczrT60giWbv2+MSybZv/x5UocSyJ5PbV321JSZH9+SLsyLrS5OSCn8O6qowx0U1ExzyqVfO/1fDOnbor5LZt+o7o7+u2bZpklizR63/+mfvzlSqVe1LJLQlVrly4QYMIEilc0siLJQ5jTNFQoULwM7T27cs/0Ry5vmrVse+zT0XOqXz54Fo2yckadwyVgrHEYYyJXaVL6yWY5dTO6QB/fonmyNeNG/Xr9u36WH9EtMx+oC2bI/eVKxeV4zeWOIwxJjsRbVWUL6+LHgOVlaXdaYEknC1bYM0a/X7XrtzPWbz4sS6yYBJOqVJhTTiWOIwxJhQSEo69yQfj0CFtreTXsvnjD12U+dVXen3fvtzPWbLksWTyxRe6BXIIWeIwxhgvJSbCSSfpJRgHDuTfsvnjj7BsCmaJwxhjiqKkJKheXS8RZuUujTHGBMUShzHGmKBY4jDGGBMUSxzGGGOCYonDGGNMUCxxGGOMCYolDmOMMUGxxGGMMSYocbEfh4hkAD8V8OFVgK0hDCdULK7gWFzBsbiCE6txneqcq5rzxrhIHIUhIkv9bWTiNYsrOBZXcCyu4MRbXNZVZYwxJiiWOIwxxgTFEkf+xngdQC4sruBYXMGxuIITV3HZGIcxxpigWIvDGGNMUCxxGGOMCYolDkBEXhWRLSLybS73i4iMEpG1IrJSRNKiJK4WIrJTRL72Xf4VobhqisgCEflORFaJyJ1+jon4axZgXBF/zUQkSUSWiMgKX1z/9nNMSRGZ7Hu9FotIapTE1UNEMrK9Xn3CHVe2504Qka9E5B0/90X89QowLk9eLxHZKCLf+J5zqZ/7Q/v/6JyL+wtwMZAGfJvL/VcCcwABmgOLoySuFsA7Hrxe1YA03/VywA9AXa9fswDjivhr5nsNyvquJwKLgeY5jrkFeNF3vQswOUri6gGMjvTfmO+5BwD/9ff78uL1CjAuT14vYCNQJY/7Q/r/aC0OwDn3CfBHHoe0B95wahFQUUSqRUFcnnDObXbOLfdd3w18B9TIcVjEX7MA44o432uwx/dtou+Sc1ZKe+B13/WpwGUiIlEQlydEJAVoC7ySyyERf70CjCtahfT/0RJHYGoAv2T7Pp0oeEPyOc/X1TBHROpF+sl9XQTnop9Ws/P0NcsjLvDgNfN1b3wNbAE+cM7l+no55zKBnUByFMQFcI2ve2OqiNQMd0w+I4FBwOFc7vfk9QogLvDm9XLA+yKyTET6+rk/pP+PljgC4++TTDR8MluO1pJpBDwLzIjkk4tIWWAa0N85tyvn3X4eEpHXLJ+4PHnNnHNZzrnGQArQTETq5zjEk9crgLjeBlKdcw2BeRz7lB82IvJ3YItzblleh/m5LayvV4BxRfz18rnAOZcGtAFuFZGLc9wf0tfLEkdg0oHsnxxSgE0exXKUc27Xka4G59xsIFFEqkTiuUUkEX1znuCce8vPIZ68ZvnF5eVr5nvOHcBHQOscdx19vUSkOFCBCHZT5haXc26bc+5P37cvA00iEM4FQDsR2QhMAi4VkTdzHOPF65VvXB69XjjnNvm+bgGmA81yHBLS/0dLHIGZBXT3zUxoDux0zm32OigROeVIv66INEN/n9si8LwCjAW+c86NyOWwiL9mgcTlxWsmIlVFpKLveingcmBNjsNmATf6rncC5jvfqKaXceXoB2+HjhuFlXPuPudcinMuFR34nu+c65bjsIi/XoHE5cXrJSJlRKTcketAKyDnTMyQ/j8WL3C0MUREJqKzbaqISDrwMDpQiHPuRWA2OithLbAP6BklcXUC+olIJrAf6BLufx6fC4AbgG98/eMA9wO1ssXmxWsWSFxevGbVgNdFJAFNVFOcc++IyBBgqXNuFprwxovIWvSTc5cwxxRoXHeISDsg0xdXjwjE5VcUvF6BxOXF63UyMN33eag48F/n3HsicjOE5//RSo4YY4wJinVVGWOMCYolDmOMMUGxxGGMMSYoljiMMcYExRKHMcaYoFjiMKYARCQrWwXUr0Xk3hCeO1VyqYic7ZjrROSgiDwYquc1JlC2jsOYgtnvK9URcSJyKVovqS4wVkTSnXOveRGLiU/W4jAmhET3RXhCdJ+LJSJSx3f7qSLyoa/43YciUst3+8kiMt1XdHGFiJzvO1WCiLwsuk/G+76V3YhIA2Ao8Dfn3Fp0Udd1IvI3D35cE6cscRhTMKVydFV1znbfLudcM2A0Wk0V3/U3fMXvJgCjfLePAj72FV1MA1b5bj8DeM45Vw/YAVwD4Jz7xjl3vnPud9/3e51zrZxzc8P4sxpzHFs5bkwBiMge51xZP7dvBC51zq33FVz8zTmXLCJbgWrOuUO+2zc756qISAaQkq0w3pGS8B84587wfT8YSHTODQ3/T2ZM/qzFYUzouVyu53aMP39mu56FjUeaKGKJw5jQ65zt6xe+659zrBDf9cBC3/UPgX5wdFOl8pEK0piCsk8xxhRMqWwVeAHec84dmZJbUkQWox/MuvpuuwN4VUTuATI4Vp30TmCMiPRGWxb9AM9L9huTFxvjMCaEfGMcTZ1zW72OxZhwsa4qY4wxQbEWhzHGmKBYi8MYY0xQLHEYY4wJiiUOY4wxQbHEYYwxJiiWOIwxxgTl/wGCraqbOLcqsQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "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": [
    "###############################\n",
    "###        PERFORMANCE      ###\n",
    "###############################\n",
    "\n",
    "epoch = np.arange(1,len(train_acc)+1)\n",
    "plt.figure()\n",
    "plt.plot(epoch, losses, 'r', epoch, val_losses, 'b')\n",
    "plt.legend(['Train Loss','Validation Loss'])\n",
    "plt.xlabel('Epoch^'), plt.ylabel('Loss')\n",
    "\n",
    "epoch = np.arange(1,len(train_acc)+1)\n",
    "plt.figure()\n",
    "plt.plot(epoch, train_acc, 'r', epoch, valid_acc, 'b')\n",
    "plt.legend(['Train Accuracy','Validation Accuracy'])\n",
    "plt.xlabel('Epoch'), plt.ylabel('Acc')\n",
    "\n",
    "#print(\"Train accuracy:\", train_acc, sep = \"\\n\")\n",
    "#print(\"Validation accuracy:\", valid_acc, sep = \"\\n\")\n",
    "\n",
    "from sklearn.metrics import matthews_corrcoef\n",
    "print(\"MCC Train:\", matthews_corrcoef(train_targs, train_preds))\n",
    "print(\"MCC Test:\", matthews_corrcoef(val_targs, val_preds))\n",
    "\n",
    "print(\"Confusion matrix train:\", confusion_matrix(train_targs, train_preds), sep = \"\\n\")\n",
    "print(\"Confusion matrix test:\", confusion_matrix(val_targs, val_preds), sep = \"\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_roc(targets, predictions):\n",
    "    # ROC\n",
    "    fpr, tpr, threshold = metrics.roc_curve(targets, predictions)\n",
    "    roc_auc = metrics.auc(fpr, tpr)\n",
    "\n",
    "    # plot ROC\n",
    "    plt.figure()\n",
    "    plt.title('Receiver Operating Characteristic')\n",
    "    plt.plot(fpr, tpr, 'b', label = 'AUC = %0.2f' % roc_auc)\n",
    "    plt.legend(loc = 'lower right')\n",
    "    plt.plot([0, 1], [0, 1],'r--')\n",
    "    plt.xlim([0, 1])\n",
    "    plt.ylim([0, 1])\n",
    "    plt.ylabel('True Positive Rate')\n",
    "    plt.xlabel('False Positive Rate')\n",
    "    #plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Validation AUC')"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "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": [
    "plot_roc(train_targs, train_preds)\n",
    "plt.title(\"Training AUC\")\n",
    "plot_roc(val_targs, val_preds)\n",
    "plt.title(\"Validation AUC\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Helpful scripts"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Show dataset as copied dataframes with named features\n",
    "The dataset is a 3D numpy array, of dimensions n_complexes x features x positions. This makes viewing the features for individual complexes or samples challenging. Below is a function which copies the entire dataset, and converts it into a list of DataFrames with named indices and columns, in order to make understanding the data easier.\n",
    "\n",
    "NB: This list of dataframes are only copies, and will not be passable into the neural network architecture."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "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>A</th>\n",
       "      <th>C</th>\n",
       "      <th>D</th>\n",
       "      <th>E</th>\n",
       "      <th>F</th>\n",
       "      <th>G</th>\n",
       "      <th>H</th>\n",
       "      <th>I</th>\n",
       "      <th>K</th>\n",
       "      <th>L</th>\n",
       "      <th>...</th>\n",
       "      <th>global_tcr_fa_rep</th>\n",
       "      <th>global_tcr_fa_sol</th>\n",
       "      <th>global_tcr_p_aa_pp</th>\n",
       "      <th>global_pmhc_total_score</th>\n",
       "      <th>global_pmhc_fa_atr</th>\n",
       "      <th>global_pmhc_fa_dun</th>\n",
       "      <th>global_pmhc_fa_elec</th>\n",
       "      <th>global_pmhc_fa_rep</th>\n",
       "      <th>global_pmhc_fa_sol</th>\n",
       "      <th>global_pmhc_p_aa_pp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>681.926</td>\n",
       "      <td>-40.471</td>\n",
       "      <td>2.401</td>\n",
       "      <td>-555.661</td>\n",
       "      <td>246.945</td>\n",
       "      <td>-359.301</td>\n",
       "      <td>2.25</td>\n",
       "      <td>733.179</td>\n",
       "      <td>-39.727</td>\n",
       "      <td>0.481</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>681.926</td>\n",
       "      <td>-40.471</td>\n",
       "      <td>2.401</td>\n",
       "      <td>-555.661</td>\n",
       "      <td>246.945</td>\n",
       "      <td>-359.301</td>\n",
       "      <td>2.25</td>\n",
       "      <td>733.179</td>\n",
       "      <td>-39.727</td>\n",
       "      <td>0.481</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>681.926</td>\n",
       "      <td>-40.471</td>\n",
       "      <td>2.401</td>\n",
       "      <td>-555.661</td>\n",
       "      <td>246.945</td>\n",
       "      <td>-359.301</td>\n",
       "      <td>2.25</td>\n",
       "      <td>733.179</td>\n",
       "      <td>-39.727</td>\n",
       "      <td>0.481</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>681.926</td>\n",
       "      <td>-40.471</td>\n",
       "      <td>2.401</td>\n",
       "      <td>-555.661</td>\n",
       "      <td>246.945</td>\n",
       "      <td>-359.301</td>\n",
       "      <td>2.25</td>\n",
       "      <td>733.179</td>\n",
       "      <td>-39.727</td>\n",
       "      <td>0.481</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>681.926</td>\n",
       "      <td>-40.471</td>\n",
       "      <td>2.401</td>\n",
       "      <td>-555.661</td>\n",
       "      <td>246.945</td>\n",
       "      <td>-359.301</td>\n",
       "      <td>2.25</td>\n",
       "      <td>733.179</td>\n",
       "      <td>-39.727</td>\n",
       "      <td>0.481</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>415</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>416</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>417</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>418</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>420 rows × 54 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       A    C    D    E    F    G    H    I    K    L  ...  global_tcr_fa_rep  \\\n",
       "0    0.0  0.0  0.0  0.0  0.0  1.0  0.0  0.0  0.0  0.0  ...            681.926   \n",
       "1    0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...            681.926   \n",
       "2    0.0  0.0  0.0  0.0  0.0  0.0  1.0  0.0  0.0  0.0  ...            681.926   \n",
       "3    0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...            681.926   \n",
       "4    0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...            681.926   \n",
       "..   ...  ...  ...  ...  ...  ...  ...  ...  ...  ...  ...                ...   \n",
       "415  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...              0.000   \n",
       "416  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...              0.000   \n",
       "417  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...              0.000   \n",
       "418  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...              0.000   \n",
       "419  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...              0.000   \n",
       "\n",
       "      global_tcr_fa_sol   global_tcr_p_aa_pp  global_pmhc_total_score  \\\n",
       "0               -40.471                2.401                 -555.661   \n",
       "1               -40.471                2.401                 -555.661   \n",
       "2               -40.471                2.401                 -555.661   \n",
       "3               -40.471                2.401                 -555.661   \n",
       "4               -40.471                2.401                 -555.661   \n",
       "..                  ...                  ...                      ...   \n",
       "415               0.000                0.000                    0.000   \n",
       "416               0.000                0.000                    0.000   \n",
       "417               0.000                0.000                    0.000   \n",
       "418               0.000                0.000                    0.000   \n",
       "419               0.000                0.000                    0.000   \n",
       "\n",
       "      global_pmhc_fa_atr   global_pmhc_fa_dun   global_pmhc_fa_elec  \\\n",
       "0                246.945             -359.301                  2.25   \n",
       "1                246.945             -359.301                  2.25   \n",
       "2                246.945             -359.301                  2.25   \n",
       "3                246.945             -359.301                  2.25   \n",
       "4                246.945             -359.301                  2.25   \n",
       "..                   ...                  ...                   ...   \n",
       "415                0.000                0.000                  0.00   \n",
       "416                0.000                0.000                  0.00   \n",
       "417                0.000                0.000                  0.00   \n",
       "418                0.000                0.000                  0.00   \n",
       "419                0.000                0.000                  0.00   \n",
       "\n",
       "     global_pmhc_fa_rep   global_pmhc_fa_sol   global_pmhc_p_aa_pp  \n",
       "0               733.179              -39.727                 0.481  \n",
       "1               733.179              -39.727                 0.481  \n",
       "2               733.179              -39.727                 0.481  \n",
       "3               733.179              -39.727                 0.481  \n",
       "4               733.179              -39.727                 0.481  \n",
       "..                  ...                  ...                   ...  \n",
       "415               0.000                0.000                 0.000  \n",
       "416               0.000                0.000                 0.000  \n",
       "417               0.000                0.000                 0.000  \n",
       "418               0.000                0.000                 0.000  \n",
       "419               0.000                0.000                 0.000  \n",
       "\n",
       "[420 rows x 54 columns]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.read_csv(\"data/example.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Showing first complex as dataframe. Columns are positions and indices are calculated features\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>A</th>\n",
       "      <th>C</th>\n",
       "      <th>D</th>\n",
       "      <th>E</th>\n",
       "      <th>F</th>\n",
       "      <th>G</th>\n",
       "      <th>H</th>\n",
       "      <th>I</th>\n",
       "      <th>K</th>\n",
       "      <th>L</th>\n",
       "      <th>...</th>\n",
       "      <th>global_tcr_fa_rep</th>\n",
       "      <th>global_tcr_fa_sol</th>\n",
       "      <th>global_tcr_p_aa_pp</th>\n",
       "      <th>global_pmhc_total_score</th>\n",
       "      <th>global_pmhc_fa_atr</th>\n",
       "      <th>global_pmhc_fa_dun</th>\n",
       "      <th>global_pmhc_fa_elec</th>\n",
       "      <th>global_pmhc_fa_rep</th>\n",
       "      <th>global_pmhc_fa_sol</th>\n",
       "      <th>global_pmhc_p_aa_pp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>656.289</td>\n",
       "      <td>-47.814</td>\n",
       "      <td>1.699</td>\n",
       "      <td>-571.57</td>\n",
       "      <td>229.289</td>\n",
       "      <td>-359.096</td>\n",
       "      <td>2.145</td>\n",
       "      <td>727.371</td>\n",
       "      <td>-40.566</td>\n",
       "      <td>0.554</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>656.289</td>\n",
       "      <td>-47.814</td>\n",
       "      <td>1.699</td>\n",
       "      <td>-571.57</td>\n",
       "      <td>229.289</td>\n",
       "      <td>-359.096</td>\n",
       "      <td>2.145</td>\n",
       "      <td>727.371</td>\n",
       "      <td>-40.566</td>\n",
       "      <td>0.554</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>656.289</td>\n",
       "      <td>-47.814</td>\n",
       "      <td>1.699</td>\n",
       "      <td>-571.57</td>\n",
       "      <td>229.289</td>\n",
       "      <td>-359.096</td>\n",
       "      <td>2.145</td>\n",
       "      <td>727.371</td>\n",
       "      <td>-40.566</td>\n",
       "      <td>0.554</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>656.289</td>\n",
       "      <td>-47.814</td>\n",
       "      <td>1.699</td>\n",
       "      <td>-571.57</td>\n",
       "      <td>229.289</td>\n",
       "      <td>-359.096</td>\n",
       "      <td>2.145</td>\n",
       "      <td>727.371</td>\n",
       "      <td>-40.566</td>\n",
       "      <td>0.554</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>656.289</td>\n",
       "      <td>-47.814</td>\n",
       "      <td>1.699</td>\n",
       "      <td>-571.57</td>\n",
       "      <td>229.289</td>\n",
       "      <td>-359.096</td>\n",
       "      <td>2.145</td>\n",
       "      <td>727.371</td>\n",
       "      <td>-40.566</td>\n",
       "      <td>0.554</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>415</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>416</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>417</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>418</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>420 rows × 54 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       A    C    D    E    F    G    H    I    K    L  ...  global_tcr_fa_rep  \\\n",
       "0    0.0  0.0  0.0  0.0  0.0  1.0  0.0  0.0  0.0  0.0  ...            656.289   \n",
       "1    0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...            656.289   \n",
       "2    0.0  0.0  0.0  0.0  0.0  0.0  1.0  0.0  0.0  0.0  ...            656.289   \n",
       "3    0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...            656.289   \n",
       "4    0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...            656.289   \n",
       "..   ...  ...  ...  ...  ...  ...  ...  ...  ...  ...  ...                ...   \n",
       "415  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...              0.000   \n",
       "416  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...              0.000   \n",
       "417  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...              0.000   \n",
       "418  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...              0.000   \n",
       "419  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  ...              0.000   \n",
       "\n",
       "      global_tcr_fa_sol   global_tcr_p_aa_pp  global_pmhc_total_score  \\\n",
       "0               -47.814                1.699                  -571.57   \n",
       "1               -47.814                1.699                  -571.57   \n",
       "2               -47.814                1.699                  -571.57   \n",
       "3               -47.814                1.699                  -571.57   \n",
       "4               -47.814                1.699                  -571.57   \n",
       "..                  ...                  ...                      ...   \n",
       "415               0.000                0.000                     0.00   \n",
       "416               0.000                0.000                     0.00   \n",
       "417               0.000                0.000                     0.00   \n",
       "418               0.000                0.000                     0.00   \n",
       "419               0.000                0.000                     0.00   \n",
       "\n",
       "      global_pmhc_fa_atr   global_pmhc_fa_dun   global_pmhc_fa_elec  \\\n",
       "0                229.289             -359.096                 2.145   \n",
       "1                229.289             -359.096                 2.145   \n",
       "2                229.289             -359.096                 2.145   \n",
       "3                229.289             -359.096                 2.145   \n",
       "4                229.289             -359.096                 2.145   \n",
       "..                   ...                  ...                   ...   \n",
       "415                0.000                0.000                 0.000   \n",
       "416                0.000                0.000                 0.000   \n",
       "417                0.000                0.000                 0.000   \n",
       "418                0.000                0.000                 0.000   \n",
       "419                0.000                0.000                 0.000   \n",
       "\n",
       "     global_pmhc_fa_rep   global_pmhc_fa_sol   global_pmhc_p_aa_pp  \n",
       "0               727.371              -40.566                 0.554  \n",
       "1               727.371              -40.566                 0.554  \n",
       "2               727.371              -40.566                 0.554  \n",
       "3               727.371              -40.566                 0.554  \n",
       "4               727.371              -40.566                 0.554  \n",
       "..                  ...                  ...                   ...  \n",
       "415               0.000                0.000                 0.000  \n",
       "416               0.000                0.000                 0.000  \n",
       "417               0.000                0.000                 0.000  \n",
       "418               0.000                0.000                 0.000  \n",
       "419               0.000                0.000                 0.000  \n",
       "\n",
       "[420 rows x 54 columns]"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def copy_as_dataframes(dataset_X):\n",
    "    \"\"\"\n",
    "    Returns list of DataFrames with named features from dataset_X,\n",
    "    using example CSV file\n",
    "    \"\"\"\n",
    "    df_raw = pd.read_csv(\"data/example.csv\")\n",
    "    return [pd.DataFrame(arr, columns = df_raw.columns) for arr in dataset_X]\n",
    "\n",
    "named_dataframes = copy_as_dataframes(X_train)\n",
    "print(\"Showing first complex as dataframe. Columns are positions and indices are calculated features\")\n",
    "named_dataframes[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# View complex MHC, peptide and TCR alpha/beta sequences\n",
    "You may want to view the one-hot encoded sequences as sequences in single-letter amino-acid format. The below function will return the TCR, peptide and MHC sequences for the dataset as 3 lists."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "def oneHot(residue):\n",
    "    \"\"\"\n",
    "    Converts string sequence to one-hot encoding\n",
    "    Example usage:\n",
    "    seq = \"GSHSMRY\"\n",
    "    oneHot(seq)\n",
    "    \"\"\"\n",
    "    \n",
    "    mapping = dict(zip(\"ACDEFGHIKLMNPQRSTVWY\", range(20)))\n",
    "    if residue in \"ACDEFGHIKLMNPQRSTVWY\":\n",
    "        return np.eye(20)[mapping[residue]]\n",
    "    else:\n",
    "        return np.zeros(20)\n",
    "def reverseOneHot(encoding):\n",
    "    \"\"\"\n",
    "    Converts one-hot encoded array back to string sequence\n",
    "    \"\"\"\n",
    "    mapping = dict(zip(range(20),\"ACDEFGHIKLMNPQRSTVWY\"))\n",
    "    seq=''\n",
    "    for i in range(len(encoding)):\n",
    "        if np.max(encoding[i])>0:\n",
    "            seq+=mapping[np.argmax(encoding[i])]\n",
    "    return seq\n",
    "\n",
    "def extract_sequences(dataset_X):\n",
    "    \"\"\"\n",
    "    Return DataFrame with MHC, peptide and TCR a/b sequences from\n",
    "    one-hot encoded complex sequences in dataset X\n",
    "    \"\"\"\n",
    "    mhc_sequences = [reverseOneHot(arr[0:179,0:20]) for arr in dataset_X]\n",
    "    pep_sequences = [reverseOneHot(arr[179:190,0:20]) for arr in dataset_X]\n",
    "    tcr_sequences = [reverseOneHot(arr[192:,0:20]) for arr in dataset_X]\n",
    "    df_sequences = pd.DataFrame({\"MHC\":mhc_sequences, \"peptide\":pep_sequences,\n",
    "                                 \"tcr\":tcr_sequences})\n",
    "    return df_sequences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Showing MHC, peptide and TCR alpha/beta sequences for each complex\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>MHC</th>\n",
       "      <th>peptide</th>\n",
       "      <th>tcr</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...</td>\n",
       "      <td>GILGFVFTL</td>\n",
       "      <td>EQSPQFLSIQEGENLTVYCNSSSVFSSLQWYRQEPGEGPVLLVTVV...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...</td>\n",
       "      <td>GILGFVFTL</td>\n",
       "      <td>EQSPQFLSIQEGENLTVYCNSSSVFSSLQWYRQEPGEGPVLLVTVV...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...</td>\n",
       "      <td>GILGFVFTL</td>\n",
       "      <td>QQVKQNSPSLSVQEGRISILNCDYTNSMFDYFLWYKKYPAEGPTFL...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...</td>\n",
       "      <td>GILGFVFTL</td>\n",
       "      <td>EQSPQFLSIQEGENLTVYCNSSSVFSSLQWYRQEPGEGPVLLVTVV...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...</td>\n",
       "      <td>GILGFVFTL</td>\n",
       "      <td>EQSPQFLSIQEGENLTVYCNSSSVFSSLQWYRQEPGEGPVLLVTVV...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1521</th>\n",
       "      <td>GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...</td>\n",
       "      <td>FLYALALLL</td>\n",
       "      <td>QSPQSMFIQEGEDVSMNCTSSSIFNTWLWYKQEPGEGPVLLIALYK...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1522</th>\n",
       "      <td>GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...</td>\n",
       "      <td>LLFGYPVYV</td>\n",
       "      <td>PQALSIQEGENATMNCSYKTSINNLQWYRQNSGRGLVHLILIRSNE...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1523</th>\n",
       "      <td>GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...</td>\n",
       "      <td>GLCTLVAML</td>\n",
       "      <td>EQSPQFLSIQEGENLTVYCNSSSVFSSLQWYRQEPGEGPVLLVTVV...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1524</th>\n",
       "      <td>GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...</td>\n",
       "      <td>FLYALALLL</td>\n",
       "      <td>QQVKQNSPSLSVQEGRISILNCDYTNSMFDYFLWYKKYPAEGPTFL...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1525</th>\n",
       "      <td>GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...</td>\n",
       "      <td>GLCTLVAML</td>\n",
       "      <td>EQSPQSLIVQEGKNLTINCTSSKTLYGLYWYKQKYGEGLIFLMMLQ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1526 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                    MHC    peptide  \\\n",
       "0     GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...  GILGFVFTL   \n",
       "1     GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...  GILGFVFTL   \n",
       "2     GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...  GILGFVFTL   \n",
       "3     GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...  GILGFVFTL   \n",
       "4     GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...  GILGFVFTL   \n",
       "...                                                 ...        ...   \n",
       "1521  GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...  FLYALALLL   \n",
       "1522  GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...  LLFGYPVYV   \n",
       "1523  GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...  GLCTLVAML   \n",
       "1524  GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...  FLYALALLL   \n",
       "1525  GSHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRME...  GLCTLVAML   \n",
       "\n",
       "                                                    tcr  \n",
       "0     EQSPQFLSIQEGENLTVYCNSSSVFSSLQWYRQEPGEGPVLLVTVV...  \n",
       "1     EQSPQFLSIQEGENLTVYCNSSSVFSSLQWYRQEPGEGPVLLVTVV...  \n",
       "2     QQVKQNSPSLSVQEGRISILNCDYTNSMFDYFLWYKKYPAEGPTFL...  \n",
       "3     EQSPQFLSIQEGENLTVYCNSSSVFSSLQWYRQEPGEGPVLLVTVV...  \n",
       "4     EQSPQFLSIQEGENLTVYCNSSSVFSSLQWYRQEPGEGPVLLVTVV...  \n",
       "...                                                 ...  \n",
       "1521  QSPQSMFIQEGEDVSMNCTSSSIFNTWLWYKQEPGEGPVLLIALYK...  \n",
       "1522  PQALSIQEGENATMNCSYKTSINNLQWYRQNSGRGLVHLILIRSNE...  \n",
       "1523  EQSPQFLSIQEGENLTVYCNSSSVFSSLQWYRQEPGEGPVLLVTVV...  \n",
       "1524  QQVKQNSPSLSVQEGRISILNCDYTNSMFDYFLWYKKYPAEGPTFL...  \n",
       "1525  EQSPQSLIVQEGKNLTINCTSSKTLYGLYWYKQKYGEGLIFLMMLQ...  \n",
       "\n",
       "[1526 rows x 3 columns]"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "complex_sequences = extract_sequences(X_val)\n",
    "print(\"Showing MHC, peptide and TCR alpha/beta sequences for each complex\")\n",
    "complex_sequences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}