{

 "cells": [

  {

   "attachments": {},

   "cell_type": "markdown",

   "metadata": {},

   "source": [

    "### Precision Medicine using Machine Learning "

   ]

  },

  {

   "attachments": {},

   "cell_type": "markdown",

   "metadata": {},

   "source": [

    "Once sequenced, a cancer tumor can have thousands of genetic mutations. The time-consuming challenging part is to classify those mutations as tumor growth(drivers) or neutral mutations(passengers). This is a time-consuming and challenging task currently done manually by clinicans based on text-based clinical literature. \n",

    "\n",

    "This project uses an expert-annotated knowledge base released by Memorial Sloan Kettering Cancer Center (MSKCC) containing thousands of manually annotated mutations by onclogists. \n",

    "\n",

    "Objective: We need your help to develop a Machine Learning algorithm that, using this knowledge base as a baseline, automatically classifies genetic variations."

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 1,

   "metadata": {},

   "outputs": [],

   "source": [

    "import numpy as np \n",

    "import pandas as pd\n",

    "import seaborn as sns\n",

    "import matplotlib.pyplot as plt\n",

    "import seaborn as sns\n",

    "import re\n",

    "#import warnings \n",

    "import math\n",

    "import nltk\n",

    "from nltk.corpus import stopwords"

   ]

  },

  {

   "cell_type": "markdown",

   "metadata": {},

   "source": [

    "Step 1: Loading Data "

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 2,

   "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>ID</th>\n",

       "      <th>Gene</th>\n",

       "      <th>Variation</th>\n",

       "      <th>Class</th>\n",

       "    </tr>\n",

       "  </thead>\n",

       "  <tbody>\n",

       "    <tr>\n",

       "      <th>0</th>\n",

       "      <td>0</td>\n",

       "      <td>FAM58A</td>\n",

       "      <td>Truncating Mutations</td>\n",

       "      <td>1</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>1</th>\n",

       "      <td>1</td>\n",

       "      <td>CBL</td>\n",

       "      <td>W802*</td>\n",

       "      <td>2</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>2</th>\n",

       "      <td>2</td>\n",

       "      <td>CBL</td>\n",

       "      <td>Q249E</td>\n",

       "      <td>2</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>3</th>\n",

       "      <td>3</td>\n",

       "      <td>CBL</td>\n",

       "      <td>N454D</td>\n",

       "      <td>3</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>4</th>\n",

       "      <td>4</td>\n",

       "      <td>CBL</td>\n",

       "      <td>L399V</td>\n",

       "      <td>4</td>\n",

       "    </tr>\n",

       "  </tbody>\n",

       "</table>\n",

       "</div>"

      ],

      "text/plain": [

       "   ID    Gene             Variation  Class\n",

       "0   0  FAM58A  Truncating Mutations      1\n",

       "1   1     CBL                 W802*      2\n",

       "2   2     CBL                 Q249E      2\n",

       "3   3     CBL                 N454D      3\n",

       "4   4     CBL                 L399V      4"

      ]

     },

     "execution_count": 2,

     "metadata": {},

     "output_type": "execute_result"

    }

   ],

   "source": [

    "train_model_variants = pd.read_csv(r\"G:\\Shared drives\\Hack(Her)thon 2023 Project\\data\\training_variants\\training_variants\")\n",

    "train_model_variants.head()\n"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 3,

   "metadata": {},

   "outputs": [

    {

     "data": {

      "text/plain": [

       "(3321, 4)"

      ]

     },

     "execution_count": 3,

     "metadata": {},

     "output_type": "execute_result"

    }

   ],

   "source": [

    "train_model_variants.shape"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 4,

   "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>ID</th>\n",

       "      <th>Gene</th>\n",

       "      <th>Variation</th>\n",

       "    </tr>\n",

       "  </thead>\n",

       "  <tbody>\n",

       "    <tr>\n",

       "      <th>0</th>\n",

       "      <td>0</td>\n",

       "      <td>ACSL4</td>\n",

       "      <td>R570S</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>1</th>\n",

       "      <td>1</td>\n",

       "      <td>NAGLU</td>\n",

       "      <td>P521L</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>2</th>\n",

       "      <td>2</td>\n",

       "      <td>PAH</td>\n",

       "      <td>L333F</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>3</th>\n",

       "      <td>3</td>\n",

       "      <td>ING1</td>\n",

       "      <td>A148D</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>4</th>\n",

       "      <td>4</td>\n",

       "      <td>TMEM216</td>\n",

       "      <td>G77A</td>\n",

       "    </tr>\n",

       "  </tbody>\n",

       "</table>\n",

       "</div>"

      ],

      "text/plain": [

       "   ID     Gene Variation\n",

       "0   0    ACSL4     R570S\n",

       "1   1    NAGLU     P521L\n",

       "2   2      PAH     L333F\n",

       "3   3     ING1     A148D\n",

       "4   4  TMEM216      G77A"

      ]

     },

     "execution_count": 4,

     "metadata": {},

     "output_type": "execute_result"

    }

   ],

   "source": [

    "test_variants = pd.read_csv(r\"G:\\Shared drives\\Hack(Her)thon 2023 Project\\data\\test_variants\\test_variants\")\n",

    "test_variants.head()"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 5,

   "metadata": {},

   "outputs": [

    {

     "data": {

      "text/plain": [

       "(5668, 3)"

      ]

     },

     "execution_count": 5,

     "metadata": {},

     "output_type": "execute_result"

    }

   ],

   "source": [

    "test_variants.shape"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 6,

   "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>ID</th>\n",

       "      <th>Text</th>\n",

       "    </tr>\n",

       "  </thead>\n",

       "  <tbody>\n",

       "    <tr>\n",

       "      <th>0</th>\n",

       "      <td>0</td>\n",

       "      <td>Cyclin-dependent kinases (CDKs) regulate a var...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>1</th>\n",

       "      <td>1</td>\n",

       "      <td>Abstract Background  Non-small cell lung canc...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>2</th>\n",

       "      <td>2</td>\n",

       "      <td>Abstract Background  Non-small cell lung canc...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>3</th>\n",

       "      <td>3</td>\n",

       "      <td>Recent evidence has demonstrated that acquired...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>4</th>\n",

       "      <td>4</td>\n",

       "      <td>Oncogenic mutations in the monomeric Casitas B...</td>\n",

       "    </tr>\n",

       "  </tbody>\n",

       "</table>\n",

       "</div>"

      ],

      "text/plain": [

       "   ID                                               Text\n",

       "0   0  Cyclin-dependent kinases (CDKs) regulate a var...\n",

       "1   1   Abstract Background  Non-small cell lung canc...\n",

       "2   2   Abstract Background  Non-small cell lung canc...\n",

       "3   3  Recent evidence has demonstrated that acquired...\n",

       "4   4  Oncogenic mutations in the monomeric Casitas B..."

      ]

     },

     "execution_count": 6,

     "metadata": {},

     "output_type": "execute_result"

    }

   ],

   "source": [

    "train_model_text = pd.read_csv(r\"G:\\Shared drives\\Hack(Her)thon 2023 Project\\data\\training_text\\training_text\", sep=\"\\|\\|\", names=[\"ID\", \"Text\"], skiprows=1, engine=\"python\", encoding=\"latin-1\")\n",

    "train_model_text.head()\n"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 7,

   "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>ID</th>\n",

       "      <th>Text</th>\n",

       "    </tr>\n",

       "  </thead>\n",

       "  <tbody>\n",

       "    <tr>\n",

       "      <th>0</th>\n",

       "      <td>0</td>\n",

       "      <td>2. This mutation resulted in a myeloproliferat...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>1</th>\n",

       "      <td>1</td>\n",

       "      <td>Abstract The Large Tumor Suppressor 1 (LATS1)...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>2</th>\n",

       "      <td>2</td>\n",

       "      <td>Vascular endothelial growth factor receptor (V...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>3</th>\n",

       "      <td>3</td>\n",

       "      <td>Inflammatory myofibroblastic tumor (IMT) is a ...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>4</th>\n",

       "      <td>4</td>\n",

       "      <td>Abstract Retinoblastoma is a pediatric retina...</td>\n",

       "    </tr>\n",

       "  </tbody>\n",

       "</table>\n",

       "</div>"

      ],

      "text/plain": [

       "   ID                                               Text\n",

       "0   0  2. This mutation resulted in a myeloproliferat...\n",

       "1   1   Abstract The Large Tumor Suppressor 1 (LATS1)...\n",

       "2   2  Vascular endothelial growth factor receptor (V...\n",

       "3   3  Inflammatory myofibroblastic tumor (IMT) is a ...\n",

       "4   4   Abstract Retinoblastoma is a pediatric retina..."

      ]

     },

     "execution_count": 7,

     "metadata": {},

     "output_type": "execute_result"

    }

   ],

   "source": [

    "test_text = pd.read_csv(r\"G:\\Shared drives\\Hack(Her)thon 2023 Project\\data\\test_text\\test_text\", sep=\"\\|\\|\", names=[\"ID\", \"Text\"], skiprows=1, engine=\"python\", encoding=\"latin-1\")\n",

    "test_text.head()"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 8,

   "metadata": {},

   "outputs": [

    {

     "name": "stderr",

     "output_type": "stream",

     "text": [

      "[nltk_data] Downloading package stopwords to\n",

      "[nltk_data]     C:\\Users\\chawl\\AppData\\Roaming\\nltk_data...\n",

      "[nltk_data]   Package stopwords is already up-to-date!\n"

     ]

    }

   ],

   "source": [

    "#right not the training_text's 'TEXT' column needs cleaning so that it can be processed to look for data that indicates cancerous class\n",

    "\n",

    "#loading stop words from nltk library \n",

    "nltk.download('stopwords')\n",

    "stop_words = set(stopwords.words('english'))\n",

    "\n",

    "def cleaning_text(full_text, index, column):\n",

    "    if type(full_text) is not str:\n",

    "        full_text = str(full_text)\n",

    "    \n",

    "    new_text = \"\"\n",

    "    empty_string = ' '\n",

    "    full_text = re.sub('[^a-zA-Z0-9\\n]', empty_string, full_text)\n",

    "    full_text = re.sub('\\s+', empty_string, full_text)\n",

    "    full_text = full_text.lower()\n",

    "\n",

    "    for word in full_text.split():\n",

    "        if not word in stop_words:\n",

    "            new_text += word + \" \"\n",

    "    \n",

    "    train_model_text[column][index] = new_text"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 9,

   "metadata": {},

   "outputs": [

    {

     "name": "stderr",

     "output_type": "stream",

     "text": [

      "C:\\Users\\chawl\\AppData\\Local\\Temp\\ipykernel_21704\\3679951068.py:21: SettingWithCopyWarning: \n",

      "A value is trying to be set on a copy of a slice from a DataFrame\n",

      "\n",

      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",

      "  train_model_text[column][index] = new_text\n"

     ]

    },

    {

     "data": {

      "text/plain": [

       "str"

      ]

     },

     "execution_count": 9,

     "metadata": {},

     "output_type": "execute_result"

    }

   ],

   "source": [

    "#column 1109 in 'Text' created issues because it was null. And initial type float\n",

    "cleaning_text(train_model_text['Text'][1109], 1109, 'Text')\n",

    "type(train_model_text['Text'][1109])"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 10,

   "metadata": {},

   "outputs": [

    {

     "name": "stderr",

     "output_type": "stream",

     "text": [

      "C:\\Users\\chawl\\AppData\\Local\\Temp\\ipykernel_21704\\3679951068.py:21: SettingWithCopyWarning: \n",

      "A value is trying to be set on a copy of a slice from a DataFrame\n",

      "\n",

      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",

      "  train_model_text[column][index] = new_text\n"

     ]

    }

   ],

   "source": [

    "#processing all the text in the training set\n",

    "\n",

    "for index, row in train_model_text.iterrows():\n",

    "    cleaning_text(row['Text'], index, 'Text')"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 11,

   "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>ID</th>\n",

       "      <th>Gene</th>\n",

       "      <th>Variation</th>\n",

       "      <th>Class</th>\n",

       "      <th>Text</th>\n",

       "    </tr>\n",

       "  </thead>\n",

       "  <tbody>\n",

       "    <tr>\n",

       "      <th>0</th>\n",

       "      <td>0</td>\n",

       "      <td>FAM58A</td>\n",

       "      <td>Truncating Mutations</td>\n",

       "      <td>1</td>\n",

       "      <td>cyclin dependent kinases cdks regulate variety...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>1</th>\n",

       "      <td>1</td>\n",

       "      <td>CBL</td>\n",

       "      <td>W802*</td>\n",

       "      <td>2</td>\n",

       "      <td>abstract background non small cell lung cancer...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>2</th>\n",

       "      <td>2</td>\n",

       "      <td>CBL</td>\n",

       "      <td>Q249E</td>\n",

       "      <td>2</td>\n",

       "      <td>abstract background non small cell lung cancer...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>3</th>\n",

       "      <td>3</td>\n",

       "      <td>CBL</td>\n",

       "      <td>N454D</td>\n",

       "      <td>3</td>\n",

       "      <td>recent evidence demonstrated acquired uniparen...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>4</th>\n",

       "      <td>4</td>\n",

       "      <td>CBL</td>\n",

       "      <td>L399V</td>\n",

       "      <td>4</td>\n",

       "      <td>oncogenic mutations monomeric casitas b lineag...</td>\n",

       "    </tr>\n",

       "  </tbody>\n",

       "</table>\n",

       "</div>"

      ],

      "text/plain": [

       "   ID    Gene             Variation  Class  \\\n",

       "0   0  FAM58A  Truncating Mutations      1   \n",

       "1   1     CBL                 W802*      2   \n",

       "2   2     CBL                 Q249E      2   \n",

       "3   3     CBL                 N454D      3   \n",

       "4   4     CBL                 L399V      4   \n",

       "\n",

       "                                                Text  \n",

       "0  cyclin dependent kinases cdks regulate variety...  \n",

       "1  abstract background non small cell lung cancer...  \n",

       "2  abstract background non small cell lung cancer...  \n",

       "3  recent evidence demonstrated acquired uniparen...  \n",

       "4  oncogenic mutations monomeric casitas b lineag...  "

      ]

     },

     "execution_count": 11,

     "metadata": {},

     "output_type": "execute_result"

    }

   ],

   "source": [

    "#Merging Training Text and Training Variants\n",

    "training_dataframe = pd.merge(train_model_variants, train_model_text, on = \"ID\")\n",

    "training_dataframe.head()\n"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 12,

   "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>ID</th>\n",

       "      <th>Gene</th>\n",

       "      <th>Variation</th>\n",

       "      <th>Text</th>\n",

       "    </tr>\n",

       "  </thead>\n",

       "  <tbody>\n",

       "    <tr>\n",

       "      <th>0</th>\n",

       "      <td>0</td>\n",

       "      <td>ACSL4</td>\n",

       "      <td>R570S</td>\n",

       "      <td>2. This mutation resulted in a myeloproliferat...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>1</th>\n",

       "      <td>1</td>\n",

       "      <td>NAGLU</td>\n",

       "      <td>P521L</td>\n",

       "      <td>Abstract The Large Tumor Suppressor 1 (LATS1)...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>2</th>\n",

       "      <td>2</td>\n",

       "      <td>PAH</td>\n",

       "      <td>L333F</td>\n",

       "      <td>Vascular endothelial growth factor receptor (V...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>3</th>\n",

       "      <td>3</td>\n",

       "      <td>ING1</td>\n",

       "      <td>A148D</td>\n",

       "      <td>Inflammatory myofibroblastic tumor (IMT) is a ...</td>\n",

       "    </tr>\n",

       "    <tr>\n",

       "      <th>4</th>\n",

       "      <td>4</td>\n",

       "      <td>TMEM216</td>\n",

       "      <td>G77A</td>\n",

       "      <td>Abstract Retinoblastoma is a pediatric retina...</td>\n",

       "    </tr>\n",

       "  </tbody>\n",

       "</table>\n",

       "</div>"

      ],

      "text/plain": [

       "   ID     Gene Variation                                               Text\n",

       "0   0    ACSL4     R570S  2. This mutation resulted in a myeloproliferat...\n",

       "1   1    NAGLU     P521L   Abstract The Large Tumor Suppressor 1 (LATS1)...\n",

       "2   2      PAH     L333F  Vascular endothelial growth factor receptor (V...\n",

       "3   3     ING1     A148D  Inflammatory myofibroblastic tumor (IMT) is a ...\n",

       "4   4  TMEM216      G77A   Abstract Retinoblastoma is a pediatric retina..."

      ]

     },

     "execution_count": 12,

     "metadata": {},

     "output_type": "execute_result"

    }

   ],

   "source": [

    "test_dataframe = pd.merge(test_variants, test_text, on = \"ID\")\n",

    "test_dataframe.head()"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 13,

   "metadata": {},

   "outputs": [

    {

     "name": "stderr",

     "output_type": "stream",

     "text": [

      "C:\\Users\\chawl\\AppData\\Local\\Temp\\ipykernel_21704\\736478266.py:2: FutureWarning: The default value of regex will change from True to False in a future version.\n",

      "  training_dataframe.Variation= training_dataframe.Variation.str.replace('\\s+', '_')\n"

     ]

    }

   ],

   "source": [

    "#fixing the variation column \n",

    "training_dataframe.Variation= training_dataframe.Variation.str.replace('\\s+', '_')"

   ]

  },

  {

   "attachments": {},

   "cell_type": "markdown",

   "metadata": {},

   "source": [

    "Step 2: Splitting the dataset for Training, Validation and Testing"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 14,

   "metadata": {},

   "outputs": [],

   "source": [

    "from sklearn.model_selection import train_test_split"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 15,

   "metadata": {},

   "outputs": [],

   "source": [

    "#split the class column from the dataframe since we are trying to predict it\n",

    "X = training_dataframe.copy()\n",

    "y = training_dataframe['Class'].values\n",

    "\n",

    "X_train, X_rem, y_train, y_rem = train_test_split(X, y, train_size = 0.8)\n",

    "\n",

    "X_vaildation, X_test, y_validation, y_test = train_test_split(X_rem, y_rem, test_size=0.5)"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 16,

   "metadata": {},

   "outputs": [

    {

     "name": "stdout",

     "output_type": "stream",

     "text": [

      "Number of records in training set: (2656, 5)\n",

      "Number of records in Cross validation set: (332, 5)\n",

      "Number of records in testing set: (333, 5)\n"

     ]

    }

   ],

   "source": [

    "print(\"Number of records in training set:\", X_train.shape)\n",

    "print(\"Number of records in Cross validation set:\", X_vaildation.shape)\n",

    "print(\"Number of records in testing set:\", X_test.shape)\n"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 17,

   "metadata": {},

   "outputs": [

    {

     "name": "stdout",

     "output_type": "stream",

     "text": [

      "Counter({7: 768, 4: 549, 1: 443, 2: 361, 6: 219, 5: 196, 3: 77, 9: 27, 8: 16})\n"

     ]

    }

   ],

   "source": [

    "import collections\n",

    "from collections import Counter\n",

    "z_train = y_train.tolist()\n",

    "d_train = Counter(z_train)\n",

    "print(d_train)"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 18,

   "metadata": {},

   "outputs": [

    {

     "name": "stdout",

     "output_type": "stream",

     "text": [

      "Counter({7: 28.91566265060241, 4: 20.670180722891565, 1: 16.67921686746988, 2: 13.591867469879517, 6: 8.245481927710843, 5: 7.379518072289157, 3: 2.8990963855421685, 9: 1.016566265060241, 8: 0.6024096385542169})\n",

      "<class 'collections.Counter'>\n",

      "Counter({7: 29.518072289156628, 1: 20.783132530120483, 4: 18.97590361445783, 2: 13.55421686746988, 5: 7.530120481927711, 6: 6.927710843373494, 3: 1.8072289156626506, 9: 0.6024096385542169, 8: 0.30120481927710846})\n",

      "Counter({7: 26.126126126126128, 4: 22.22222222222222, 1: 16.816816816816818, 2: 13.813813813813814, 6: 9.90990990990991, 5: 6.306306306306307, 9: 2.4024024024024024, 3: 1.8018018018018018, 8: 0.6006006006006006})\n"

     ]

    }

   ],

   "source": [

    "\n",

    "s = sum(d_train.values())\n",

    "for k,v in d_train.items():\n",

    "    pct = v* 100.0/s\n",

    "    d_train[k] = pct\n",

    "print(d_train)\n",

    "print(type(d_train))\n",

    "#print(d.values())\n",

    "\n",

    "z_cv = y_validation.tolist()\n",

    "d_cv = Counter(z_cv)\n",

    "s2 = sum(d_cv.values())\n",

    "for k,v in d_cv.items():\n",

    "    pct = v* 100.0/s2\n",

    "    d_cv[k] = pct\n",

    "print(d_cv)\n",

    "\n",

    "\n",

    "z_testing = y_test.tolist()\n",

    "d_testing = Counter(z_testing)\n",

    "\n",

    "s3 = sum(d_testing.values())\n",

    "for k,v in d_testing.items():\n",

    "    pct = v* 100.0/s3\n",

    "    d_testing[k] = pct\n",

    "print(d_testing)"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 19,

   "metadata": {},

   "outputs": [

    {

     "data": {

      "image/png": "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",

      "text/plain": [

       "<Figure size 300x300 with 1 Axes>"

      ]

     },

     "metadata": {},

     "output_type": "display_data"

    },

    {

     "data": {

      "image/png": "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",

      "text/plain": [

       "<Figure size 300x300 with 1 Axes>"

      ]

     },

     "metadata": {},

     "output_type": "display_data"

    },

    {

     "data": {

      "image/png": "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",

      "text/plain": [

       "<Figure size 300x300 with 1 Axes>"

      ]

     },

     "metadata": {},

     "output_type": "display_data"

    }

   ],

   "source": [

    "#Distribution of classes across all 3 datsets: Training, cross validation and testing should be fair \n",

    "\n",

    "#1. training\n",

    "fig_train = plt.figure(figsize=(3,3))\n",

    "plt.bar(d_train.keys(), d_train.values(), color = 'pink', width = 0.5)\n",

    "plt.xlabel(\"Classes\")\n",

    "plt.ylabel(\"Total cases\")\n",

    "plt.title(\"Distribution of classes in the Training DataSet\")\n",

    "plt.show()\n",

    "\n",

    "#2. cross-validation\n",

    "fig_cv = plt.figure(figsize=(3,3))\n",

    "plt.bar(d_cv.keys(), d_cv.values(), color = 'pink', width = 0.5)\n",

    "plt.xlabel(\"Classes\")\n",

    "plt.ylabel(\"Total cases\")\n",

    "plt.title(\"Distribution of classes in the Cross-Validation DataSet\")\n",

    "plt.show()\n",

    "\n",

    "#3. testing\n",

    "fig = plt.figure(figsize=(3,3))\n",

    "plt.bar(d_testing.keys(), d_testing.values(), color = 'pink', width = 0.5)\n",

    "plt.xlabel(\"Classes\")\n",

    "plt.ylabel(\"Total cases\")\n",

    "plt.title(\"Distribution of classes in the Testing DataSet\")\n",

    "plt.show()\n"

   ]

  },

  {

   "attachments": {},

   "cell_type": "markdown",

   "metadata": {},

   "source": [

    "Setting performance Standards for the model: Log Loss\n",

    "\n",

    "For a multi-classification problem where each class has an equal probability of being classified, \n",

    "log_loss = -log(1/M)    where M is the number of classes\n",

    "-log(1/9) = 0.954\n",

    "\n",

    "But that is not the case here. As we saw from the graphs above, all 3 dataset are unequally distributed over classes. \n",

    "So, how do we calculate log-loss for imbalanced multi-class problem.\n",

    "\n",

    "By my calculation, the random guessing log-loss for this imbalanced multi class(K=9) problem comes out as 1.829\n",

    "Pictures of calculation on github.\n",

    "\n",

    "Therefore, our model has to do better than this i.e. logloss<1.829"

   ]

  },

  {

   "attachments": {},

   "cell_type": "markdown",

   "metadata": {},

   "source": [

    "### Univariate Analysis"

   ]

  },

  {

   "attachments": {},

   "cell_type": "markdown",

   "metadata": {},

   "source": [

    "##### Gene Feature"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 20,

   "metadata": {},

   "outputs": [

    {

     "name": "stdout",

     "output_type": "stream",

     "text": [

      "There are 244 different categories of genes in the training data\n",

      "Gene\n",

      "BRCA1      220\n",

      "TP53       117\n",

      "EGFR       113\n",

      "PTEN       106\n",

      "BRCA2       89\n",

      "          ... \n",

      "LATS1        1\n",

      "KLF4         1\n",

      "BCL2L11      1\n",

      "KDM6A        1\n",

      "FLT1         1\n",

      "Length: 244, dtype: int64\n"

     ]

    }

   ],

   "source": [

    "#unique, number of occuerences of each gene, cumulative distribution \n",

    "unique_genes =X_train.value_counts('Gene')   \n",

    "print(\"There are\", unique_genes.shape[0], \"different categories of genes in the training data\" )\n",

    "print(unique_genes)"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 21,

   "metadata": {},

   "outputs": [

    {

     "data": {

      "image/png": "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",

      "text/plain": [

       "<Figure size 640x480 with 1 Axes>"

      ]

     },

     "metadata": {},

     "output_type": "display_data"

    }

   ],

   "source": [

    "s= sum(unique_genes.values)\n",

    "h = unique_genes.values/s\n",

    "plt.plot(h, label = \"Histogram of Genes\")\n",

    "plt.ylabel(\"Number of occurences\")\n",

    "plt.xlabel(\"Index of a gene\")\n",

    "plt.legend()\n",

    "plt.grid()\n",

    "plt.show()"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 22,

   "metadata": {},

   "outputs": [

    {

     "data": {

      "image/png": "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",

      "text/plain": [

       "<Figure size 640x480 with 1 Axes>"

      ]

     },

     "metadata": {},

     "output_type": "display_data"

    },

    {

     "name": "stdout",

     "output_type": "stream",

     "text": [

      "Top 50 most frequent genes nearly contribute to 75% of data\n"

     ]

    }

   ],

   "source": [

    "cdf = np.cumsum(h)\n",

    "plt.plot(cdf, color =\"green\", label = \"Cumulative Distribution of Genes\")\n",

    "plt.legend()\n",

    "plt.grid()\n",

    "plt.show()\n",

    "print(\"Top 50 most frequent genes nearly contribute to 75% of data\")\n"

   ]

  },

  {

   "attachments": {},

   "cell_type": "markdown",

   "metadata": {},

   "source": [

    "Converting the genes into vectore by Response Coding and One-Hot Encoding\n"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 23,

   "metadata": {},

   "outputs": [

    {

     "name": "stdout",

     "output_type": "stream",

     "text": [

      "Here's the first 10 items in the Gene array after response coding: \n",

      "\n",

      "('BRCA1', [0.10560344827586207, 0.0021551724137931034, 0.034482758620689655, 0.07543103448275862, 0.17672413793103448, 0.09267241379310345, 0.0021551724137931034, 0.0021551724137931034, 0.0021551724137931034])\n",

      "('TP53', [0.1772853185595568, 0.00554016620498615, 0.00554016620498615, 0.1329639889196676, 0.008310249307479225, 0.0110803324099723, 0.002770083102493075, 0.002770083102493075, 0.002770083102493075])\n",

      "('EGFR', [0.0056022408963585435, 0.10084033613445378, 0.0028011204481792717, 0.014005602240896359, 0.011204481792717087, 0.0028011204481792717, 0.19607843137254902, 0.0056022408963585435, 0.0028011204481792717])\n",

      "('PTEN', [0.011428571428571429, 0.002857142857142857, 0.017142857142857144, 0.27714285714285714, 0.005714285714285714, 0.002857142857142857, 0.005714285714285714, 0.002857142857142857, 0.002857142857142857])\n",

      "('BRCA2', [0.02702702702702703, 0.003003003003003003, 0.003003003003003003, 0.009009009009009009, 0.03003003003003003, 0.2132132132132132, 0.003003003003003003, 0.003003003003003003, 0.003003003003003003])\n",

      "('KIT', [0.003076923076923077, 0.11076923076923077, 0.006153846153846154, 0.003076923076923077, 0.003076923076923077, 0.003076923076923077, 0.14153846153846153, 0.003076923076923077, 0.003076923076923077])\n",

      "('BRAF', [0.003205128205128205, 0.07371794871794872, 0.00641025641025641, 0.003205128205128205, 0.019230769230769232, 0.01282051282051282, 0.12179487179487179, 0.003205128205128205, 0.003205128205128205])\n",

      "('ERBB2', [0.006600660066006601, 0.039603960396039604, 0.0033003300330033004, 0.0165016501650165, 0.019801980198019802, 0.039603960396039604, 0.0891089108910891, 0.006600660066006601, 0.0033003300330033004])\n",

      "('ALK', [0.0033112582781456954, 0.026490066225165563, 0.026490066225165563, 0.0033112582781456954, 0.019867549668874173, 0.0033112582781456954, 0.13245033112582782, 0.0033112582781456954, 0.0033112582781456954])\n",

      "('PDGFRA', [0.006802721088435374, 0.013605442176870748, 0.030612244897959183, 0.006802721088435374, 0.017006802721088437, 0.003401360544217687, 0.11564625850340136, 0.003401360544217687, 0.003401360544217687])\n"

     ]

    }

   ],

   "source": [

    "#In Response Coding, each level of a categorical variable is assigned a probability value based \n",

    "#on the propoertion of positive class observations for that level.\n",

    "\n",

    "#response coding for a gene using laplace smoothing \n",

    "alpha = 1\n",

    "\n",

    "#Creating a mehtod for gene one-hot encoding that can be used for all 3 data sets \n",

    "def gene_response_coding(df):\n",

    "    gene_dict = dict()\n",

    "    values = df['Gene'].value_counts()\n",

    "    for gene, freq in values.items():\n",

    "        prob_vec = []\n",

    "        for cancer_class in range(1,10):\n",

    "            class_occurence = df.loc[(df['Class'] == cancer_class) & (df['Gene'] == gene)]\n",

    "            prob_vec.append((class_occurence.shape[0] + alpha)/(freq + (alpha * values.shape[0])))\n",

    "        gene_dict[gene] = prob_vec\n",

    "    \n",

    "    return gene_dict\n",

    "\n",

    "\n",

    "train_gene_responseCoding = gene_response_coding(X_train)\n",

    "cv_gene_responseCoding = gene_response_coding(X_vaildation)\n",

    "test_gene_responseCoding = gene_response_coding(X_test)\n",

    "print(\"Here's the first 10 items in the Gene array after response coding: \\n\")\n",

    "from itertools import islice\n",

    "n_items = list(islice(train_gene_responseCoding.items(), 10))\n",

    "for i in range(0,10):\n",

    "    print(n_items[i]) \n",

    "\n",

    "\n",

    "\n",

    "\n",

    "\n"

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 24,

   "metadata": {},

   "outputs": [],

   "source": [

    "# So if you look at the first entry BRCA 1, its mutatations have the highest probability of being in class 5 \n",

    "#Then, the probability of the mutation falling under class 1, 4, 6 is nearly the same "

   ]

  },

  {

   "cell_type": "code",

   "execution_count": 25,