05_Selectivity_model_v02.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Machine Learning applied to identify Selective Antimicobrial Peptides\n",
    "\n",
    "In this notebook is described all the procedure used to train and test the Selectivity model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import torch\n",
    "import pickle\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.svm import SVR\n",
    "\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.linear_model import Lasso\n",
    "from sklearn.linear_model import LinearRegression\n",
    "\n",
    "import joblib\n",
    "from sklearn import metrics\n",
    "from sklearn.model_selection import train_test_split, GridSearchCV, cross_val_score, LeaveOneOut\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore', category=UserWarning, module='seaborn.axisgrid', message='The figure layout has changed to tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "100"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# This function wil convert the undefined MIC values to 100.\n",
    "def convert_mic(mic_value):\n",
    "    if type(mic_value) == str and mic_value.startswith('>'):\n",
    "        mic_value = float(mic_value.replace('>', ''))\n",
    "        v = 100\n",
    "        return v\n",
    "    else:\n",
    "        return float(mic_value)\n",
    "\n",
    "convert_mic('>64')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>Sequence</th>\n",
       "      <th>MBC</th>\n",
       "      <th>%Hemolysis</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>PG-1.37</td>\n",
       "      <td>HVRRLCYCRRRFCACVGS</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.600000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PG-1.20d</td>\n",
       "      <td>RGGRLRYCRRRFCVSVGR</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.607281</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>PG-1.62</td>\n",
       "      <td>RGGRLCYCRRRFCTCVGR</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.233927</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>PG-1.12A</td>\n",
       "      <td>RGGRLCYCRRRACVCVGR</td>\n",
       "      <td>2.0</td>\n",
       "      <td>12.103021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>PG-1.16A</td>\n",
       "      <td>RGGRLCYCRRRFCVCAGR</td>\n",
       "      <td>2.0</td>\n",
       "      <td>15.821069</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>88</th>\n",
       "      <td>PG-1.32</td>\n",
       "      <td>HGWRLCYCRRRFCVCVGC</td>\n",
       "      <td>100.0</td>\n",
       "      <td>35.460367</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89</th>\n",
       "      <td>PG-1.60</td>\n",
       "      <td>RGGRLCYCRRRFCVCVGF</td>\n",
       "      <td>100.0</td>\n",
       "      <td>36.664085</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90</th>\n",
       "      <td>PG-1.25</td>\n",
       "      <td>RGWRLCYCRRRFCVCVGH</td>\n",
       "      <td>100.0</td>\n",
       "      <td>46.397292</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>91</th>\n",
       "      <td>PG-1.7</td>\n",
       "      <td>RRWCLCFCRRRFCVCVCR</td>\n",
       "      <td>100.0</td>\n",
       "      <td>51.181276</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>PG-1.24</td>\n",
       "      <td>RGWRLCYCRRRFCVCVGC</td>\n",
       "      <td>100.0</td>\n",
       "      <td>78.280710</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>93 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          ID            Sequence    MBC  %Hemolysis\n",
       "0    PG-1.37  HVRRLCYCRRRFCACVGS    1.0    2.600000\n",
       "1   PG-1.20d  RGGRLRYCRRRFCVSVGR    2.0    1.607281\n",
       "2    PG-1.62  RGGRLCYCRRRFCTCVGR    2.0    3.233927\n",
       "3   PG-1.12A  RGGRLCYCRRRACVCVGR    2.0   12.103021\n",
       "4   PG-1.16A  RGGRLCYCRRRFCVCAGR    2.0   15.821069\n",
       "..       ...                 ...    ...         ...\n",
       "88   PG-1.32  HGWRLCYCRRRFCVCVGC  100.0   35.460367\n",
       "89   PG-1.60  RGGRLCYCRRRFCVCVGF  100.0   36.664085\n",
       "90   PG-1.25  RGWRLCYCRRRFCVCVGH  100.0   46.397292\n",
       "91    PG-1.7  RRWCLCFCRRRFCVCVCR  100.0   51.181276\n",
       "92   PG-1.24  RGWRLCYCRRRFCVCVGC  100.0   78.280710\n",
       "\n",
       "[93 rows x 4 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load meta data\n",
    "meta_data = pd.read_excel(\"data/Cumulative_Peptide_data_PG1_v01.2.xlsx\")\n",
    "meta_data[\"MBC\"] = [convert_mic(x) for x in meta_data[\"MBC\"]]\n",
    "meta_data"
   ]
  },
  {
   "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>Sequence</th>\n",
       "      <th>MBC</th>\n",
       "      <th>%Hemolysis</th>\n",
       "      <th>selectivity</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>PG-1.37</td>\n",
       "      <td>HVRRLCYCRRRFCACVGS</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.600000</td>\n",
       "      <td>0.414973</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PG-1.20d</td>\n",
       "      <td>RGGRLRYCRRRFCVSVGR</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.607281</td>\n",
       "      <td>0.507122</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>PG-1.62</td>\n",
       "      <td>RGGRLCYCRRRFCTCVGR</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.233927</td>\n",
       "      <td>0.810760</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>PG-1.12A</td>\n",
       "      <td>RGGRLCYCRRRACVCVGR</td>\n",
       "      <td>2.0</td>\n",
       "      <td>12.103021</td>\n",
       "      <td>1.383924</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>PG-1.16A</td>\n",
       "      <td>RGGRLCYCRRRFCVCAGR</td>\n",
       "      <td>2.0</td>\n",
       "      <td>15.821069</td>\n",
       "      <td>1.500266</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>88</th>\n",
       "      <td>PG-1.32</td>\n",
       "      <td>HGWRLCYCRRRFCVCVGC</td>\n",
       "      <td>100.0</td>\n",
       "      <td>35.460367</td>\n",
       "      <td>3.549743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89</th>\n",
       "      <td>PG-1.60</td>\n",
       "      <td>RGGRLCYCRRRFCVCVGF</td>\n",
       "      <td>100.0</td>\n",
       "      <td>36.664085</td>\n",
       "      <td>3.564241</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90</th>\n",
       "      <td>PG-1.25</td>\n",
       "      <td>RGWRLCYCRRRFCVCVGH</td>\n",
       "      <td>100.0</td>\n",
       "      <td>46.397292</td>\n",
       "      <td>3.666493</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>91</th>\n",
       "      <td>PG-1.7</td>\n",
       "      <td>RRWCLCFCRRRFCVCVCR</td>\n",
       "      <td>100.0</td>\n",
       "      <td>51.181276</td>\n",
       "      <td>3.709111</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>PG-1.24</td>\n",
       "      <td>RGWRLCYCRRRFCVCVGC</td>\n",
       "      <td>100.0</td>\n",
       "      <td>78.280710</td>\n",
       "      <td>3.893655</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>93 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          ID            Sequence    MBC  %Hemolysis  selectivity\n",
       "0    PG-1.37  HVRRLCYCRRRFCACVGS    1.0    2.600000     0.414973\n",
       "1   PG-1.20d  RGGRLRYCRRRFCVSVGR    2.0    1.607281     0.507122\n",
       "2    PG-1.62  RGGRLCYCRRRFCTCVGR    2.0    3.233927     0.810760\n",
       "3   PG-1.12A  RGGRLCYCRRRACVCVGR    2.0   12.103021     1.383924\n",
       "4   PG-1.16A  RGGRLCYCRRRFCVCAGR    2.0   15.821069     1.500266\n",
       "..       ...                 ...    ...         ...          ...\n",
       "88   PG-1.32  HGWRLCYCRRRFCVCVGC  100.0   35.460367     3.549743\n",
       "89   PG-1.60  RGGRLCYCRRRFCVCVGF  100.0   36.664085     3.564241\n",
       "90   PG-1.25  RGWRLCYCRRRFCVCVGH  100.0   46.397292     3.666493\n",
       "91    PG-1.7  RRWCLCFCRRRFCVCVCR  100.0   51.181276     3.709111\n",
       "92   PG-1.24  RGWRLCYCRRRFCVCVGC  100.0   78.280710     3.893655\n",
       "\n",
       "[93 rows x 5 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create the new target selectivity\n",
    "meta_data['selectivity'] = meta_data['MBC'] * meta_data['%Hemolysis']\n",
    "meta_data['selectivity'] = np.log10(meta_data['selectivity'])\n",
    "meta_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# target distribuition\n",
    "sns.histplot(data=meta_data, x='selectivity', bins=10, kde=True, stat='percent')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Embedding Extraction\n",
    "\n",
    "The embeddings I used for the model were computed using the ESM-2 15B model and I used the mean of the last layer as the embedding.\n",
    "\n",
    "Here's how to exytract them using the pre-trained model from esm2.  \n",
    "\n",
    "Download the python file extract from [esm2-github](https://github.com/facebookresearch/esm), and make sure to have a conda enviromment seted up accordinly.\n",
    "\n",
    "Then run:  \n",
    "python extract.py esm2_t48_15B_UR50D data/Cumulative_Peptide_data_PG1.fasta data/embedding/esm2_15B_PG1 --repr_layers 48 --include mean"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>Sequence</th>\n",
       "      <th>MBC</th>\n",
       "      <th>%Hemolysis</th>\n",
       "      <th>selectivity</th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>...</th>\n",
       "      <th>5110</th>\n",
       "      <th>5111</th>\n",
       "      <th>5112</th>\n",
       "      <th>5113</th>\n",
       "      <th>5114</th>\n",
       "      <th>5115</th>\n",
       "      <th>5116</th>\n",
       "      <th>5117</th>\n",
       "      <th>5118</th>\n",
       "      <th>5119</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>PG-1.37</td>\n",
       "      <td>HVRRLCYCRRRFCACVGS</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.600000</td>\n",
       "      <td>0.414973</td>\n",
       "      <td>0.056671</td>\n",
       "      <td>-0.074099</td>\n",
       "      <td>-0.040746</td>\n",
       "      <td>0.217794</td>\n",
       "      <td>-0.197767</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.130863</td>\n",
       "      <td>-0.107892</td>\n",
       "      <td>0.140624</td>\n",
       "      <td>-0.169443</td>\n",
       "      <td>0.165775</td>\n",
       "      <td>-0.120193</td>\n",
       "      <td>0.094954</td>\n",
       "      <td>-0.000617</td>\n",
       "      <td>-0.199790</td>\n",
       "      <td>0.001056</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PG-1.20d</td>\n",
       "      <td>RGGRLRYCRRRFCVSVGR</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.607281</td>\n",
       "      <td>0.507122</td>\n",
       "      <td>0.133009</td>\n",
       "      <td>-0.124348</td>\n",
       "      <td>-0.109747</td>\n",
       "      <td>0.222604</td>\n",
       "      <td>-0.261453</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.105108</td>\n",
       "      <td>-0.053451</td>\n",
       "      <td>0.092585</td>\n",
       "      <td>-0.201894</td>\n",
       "      <td>0.094672</td>\n",
       "      <td>-0.104331</td>\n",
       "      <td>0.135752</td>\n",
       "      <td>0.036512</td>\n",
       "      <td>-0.230271</td>\n",
       "      <td>0.020399</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>PG-1.62</td>\n",
       "      <td>RGGRLCYCRRRFCTCVGR</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.233927</td>\n",
       "      <td>0.810760</td>\n",
       "      <td>0.155696</td>\n",
       "      <td>-0.016959</td>\n",
       "      <td>-0.179328</td>\n",
       "      <td>0.153639</td>\n",
       "      <td>-0.241203</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.160722</td>\n",
       "      <td>-0.136943</td>\n",
       "      <td>0.049061</td>\n",
       "      <td>-0.175572</td>\n",
       "      <td>0.109034</td>\n",
       "      <td>-0.239965</td>\n",
       "      <td>-0.050614</td>\n",
       "      <td>-0.096664</td>\n",
       "      <td>-0.249927</td>\n",
       "      <td>0.020688</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>PG-1.12A</td>\n",
       "      <td>RGGRLCYCRRRACVCVGR</td>\n",
       "      <td>2.0</td>\n",
       "      <td>12.103021</td>\n",
       "      <td>1.383924</td>\n",
       "      <td>0.056310</td>\n",
       "      <td>-0.137829</td>\n",
       "      <td>-0.098538</td>\n",
       "      <td>0.162526</td>\n",
       "      <td>-0.252675</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.124351</td>\n",
       "      <td>-0.181434</td>\n",
       "      <td>0.037783</td>\n",
       "      <td>-0.201551</td>\n",
       "      <td>0.092573</td>\n",
       "      <td>-0.168867</td>\n",
       "      <td>0.078524</td>\n",
       "      <td>-0.003103</td>\n",
       "      <td>-0.212419</td>\n",
       "      <td>0.049512</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>PG-1.16A</td>\n",
       "      <td>RGGRLCYCRRRFCVCAGR</td>\n",
       "      <td>2.0</td>\n",
       "      <td>15.821069</td>\n",
       "      <td>1.500266</td>\n",
       "      <td>0.122564</td>\n",
       "      <td>-0.024471</td>\n",
       "      <td>-0.127546</td>\n",
       "      <td>0.165324</td>\n",
       "      <td>-0.258667</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.155189</td>\n",
       "      <td>-0.155856</td>\n",
       "      <td>0.061196</td>\n",
       "      <td>-0.220575</td>\n",
       "      <td>0.092058</td>\n",
       "      <td>-0.220014</td>\n",
       "      <td>0.025014</td>\n",
       "      <td>-0.053914</td>\n",
       "      <td>-0.198191</td>\n",
       "      <td>0.036471</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>88</th>\n",
       "      <td>PG-1.32</td>\n",
       "      <td>HGWRLCYCRRRFCVCVGC</td>\n",
       "      <td>100.0</td>\n",
       "      <td>35.460367</td>\n",
       "      <td>3.549743</td>\n",
       "      <td>0.007245</td>\n",
       "      <td>-0.115889</td>\n",
       "      <td>-0.040796</td>\n",
       "      <td>0.235570</td>\n",
       "      <td>-0.216340</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.166121</td>\n",
       "      <td>-0.146094</td>\n",
       "      <td>0.080064</td>\n",
       "      <td>-0.143502</td>\n",
       "      <td>0.155125</td>\n",
       "      <td>-0.067742</td>\n",
       "      <td>0.087349</td>\n",
       "      <td>-0.027154</td>\n",
       "      <td>-0.214921</td>\n",
       "      <td>0.023557</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89</th>\n",
       "      <td>PG-1.60</td>\n",
       "      <td>RGGRLCYCRRRFCVCVGF</td>\n",
       "      <td>100.0</td>\n",
       "      <td>36.664085</td>\n",
       "      <td>3.564241</td>\n",
       "      <td>0.143025</td>\n",
       "      <td>-0.030791</td>\n",
       "      <td>-0.110694</td>\n",
       "      <td>0.109217</td>\n",
       "      <td>-0.223111</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.119124</td>\n",
       "      <td>-0.189840</td>\n",
       "      <td>-0.030339</td>\n",
       "      <td>-0.218860</td>\n",
       "      <td>0.098898</td>\n",
       "      <td>-0.174355</td>\n",
       "      <td>0.023488</td>\n",
       "      <td>-0.059298</td>\n",
       "      <td>-0.252023</td>\n",
       "      <td>0.029996</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90</th>\n",
       "      <td>PG-1.25</td>\n",
       "      <td>RGWRLCYCRRRFCVCVGH</td>\n",
       "      <td>100.0</td>\n",
       "      <td>46.397292</td>\n",
       "      <td>3.666493</td>\n",
       "      <td>0.068697</td>\n",
       "      <td>-0.038422</td>\n",
       "      <td>-0.142241</td>\n",
       "      <td>0.184863</td>\n",
       "      <td>-0.232976</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.180917</td>\n",
       "      <td>-0.172996</td>\n",
       "      <td>0.046710</td>\n",
       "      <td>-0.179037</td>\n",
       "      <td>0.106190</td>\n",
       "      <td>-0.103997</td>\n",
       "      <td>-0.029128</td>\n",
       "      <td>0.000693</td>\n",
       "      <td>-0.243295</td>\n",
       "      <td>0.019351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>91</th>\n",
       "      <td>PG-1.7</td>\n",
       "      <td>RRWCLCFCRRRFCVCVCR</td>\n",
       "      <td>100.0</td>\n",
       "      <td>51.181276</td>\n",
       "      <td>3.709111</td>\n",
       "      <td>0.066839</td>\n",
       "      <td>-0.037917</td>\n",
       "      <td>-0.087912</td>\n",
       "      <td>0.137855</td>\n",
       "      <td>-0.218455</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.167717</td>\n",
       "      <td>-0.251613</td>\n",
       "      <td>-0.086687</td>\n",
       "      <td>-0.231941</td>\n",
       "      <td>0.015133</td>\n",
       "      <td>-0.017062</td>\n",
       "      <td>0.048716</td>\n",
       "      <td>-0.028251</td>\n",
       "      <td>-0.203544</td>\n",
       "      <td>-0.114390</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>PG-1.24</td>\n",
       "      <td>RGWRLCYCRRRFCVCVGC</td>\n",
       "      <td>100.0</td>\n",
       "      <td>78.280710</td>\n",
       "      <td>3.893655</td>\n",
       "      <td>0.069902</td>\n",
       "      <td>-0.039020</td>\n",
       "      <td>-0.122685</td>\n",
       "      <td>0.183626</td>\n",
       "      <td>-0.228307</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.200931</td>\n",
       "      <td>-0.165114</td>\n",
       "      <td>0.059687</td>\n",
       "      <td>-0.133821</td>\n",
       "      <td>0.111872</td>\n",
       "      <td>-0.142763</td>\n",
       "      <td>-0.034693</td>\n",
       "      <td>-0.022357</td>\n",
       "      <td>-0.249541</td>\n",
       "      <td>0.047750</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>93 rows × 5125 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          ID            Sequence    MBC  %Hemolysis  selectivity         0  \\\n",
       "0    PG-1.37  HVRRLCYCRRRFCACVGS    1.0    2.600000     0.414973  0.056671   \n",
       "1   PG-1.20d  RGGRLRYCRRRFCVSVGR    2.0    1.607281     0.507122  0.133009   \n",
       "2    PG-1.62  RGGRLCYCRRRFCTCVGR    2.0    3.233927     0.810760  0.155696   \n",
       "3   PG-1.12A  RGGRLCYCRRRACVCVGR    2.0   12.103021     1.383924  0.056310   \n",
       "4   PG-1.16A  RGGRLCYCRRRFCVCAGR    2.0   15.821069     1.500266  0.122564   \n",
       "..       ...                 ...    ...         ...          ...       ...   \n",
       "88   PG-1.32  HGWRLCYCRRRFCVCVGC  100.0   35.460367     3.549743  0.007245   \n",
       "89   PG-1.60  RGGRLCYCRRRFCVCVGF  100.0   36.664085     3.564241  0.143025   \n",
       "90   PG-1.25  RGWRLCYCRRRFCVCVGH  100.0   46.397292     3.666493  0.068697   \n",
       "91    PG-1.7  RRWCLCFCRRRFCVCVCR  100.0   51.181276     3.709111  0.066839   \n",
       "92   PG-1.24  RGWRLCYCRRRFCVCVGC  100.0   78.280710     3.893655  0.069902   \n",
       "\n",
       "           1         2         3         4  ...      5110      5111      5112  \\\n",
       "0  -0.074099 -0.040746  0.217794 -0.197767  ... -0.130863 -0.107892  0.140624   \n",
       "1  -0.124348 -0.109747  0.222604 -0.261453  ... -0.105108 -0.053451  0.092585   \n",
       "2  -0.016959 -0.179328  0.153639 -0.241203  ... -0.160722 -0.136943  0.049061   \n",
       "3  -0.137829 -0.098538  0.162526 -0.252675  ... -0.124351 -0.181434  0.037783   \n",
       "4  -0.024471 -0.127546  0.165324 -0.258667  ... -0.155189 -0.155856  0.061196   \n",
       "..       ...       ...       ...       ...  ...       ...       ...       ...   \n",
       "88 -0.115889 -0.040796  0.235570 -0.216340  ... -0.166121 -0.146094  0.080064   \n",
       "89 -0.030791 -0.110694  0.109217 -0.223111  ... -0.119124 -0.189840 -0.030339   \n",
       "90 -0.038422 -0.142241  0.184863 -0.232976  ... -0.180917 -0.172996  0.046710   \n",
       "91 -0.037917 -0.087912  0.137855 -0.218455  ... -0.167717 -0.251613 -0.086687   \n",
       "92 -0.039020 -0.122685  0.183626 -0.228307  ... -0.200931 -0.165114  0.059687   \n",
       "\n",
       "        5113      5114      5115      5116      5117      5118      5119  \n",
       "0  -0.169443  0.165775 -0.120193  0.094954 -0.000617 -0.199790  0.001056  \n",
       "1  -0.201894  0.094672 -0.104331  0.135752  0.036512 -0.230271  0.020399  \n",
       "2  -0.175572  0.109034 -0.239965 -0.050614 -0.096664 -0.249927  0.020688  \n",
       "3  -0.201551  0.092573 -0.168867  0.078524 -0.003103 -0.212419  0.049512  \n",
       "4  -0.220575  0.092058 -0.220014  0.025014 -0.053914 -0.198191  0.036471  \n",
       "..       ...       ...       ...       ...       ...       ...       ...  \n",
       "88 -0.143502  0.155125 -0.067742  0.087349 -0.027154 -0.214921  0.023557  \n",
       "89 -0.218860  0.098898 -0.174355  0.023488 -0.059298 -0.252023  0.029996  \n",
       "90 -0.179037  0.106190 -0.103997 -0.029128  0.000693 -0.243295  0.019351  \n",
       "91 -0.231941  0.015133 -0.017062  0.048716 -0.028251 -0.203544 -0.114390  \n",
       "92 -0.133821  0.111872 -0.142763 -0.034693 -0.022357 -0.249541  0.047750  \n",
       "\n",
       "[93 rows x 5125 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# # creates a empty dictionary, and updates it with the embedds\n",
    "dir_path = 'data/embedding/esm2_15B_PG1/'\n",
    "\n",
    "embeddings = {}\n",
    "for file in os.listdir(dir_path):\n",
    "    file_path = str(dir_path) + file \n",
    "    label = file.split('.pt')[0]\n",
    "    embeddings[label] = np.array(torch.load(file_path)['mean_representations'][48])\n",
    "\n",
    "# here I creating a dictionary with ID as key and semantic value as value \n",
    "df = meta_data.merge(pd.DataFrame.from_dict(embeddings, orient='index'), how='inner', left_on='ID', right_index=True)\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>...</th>\n",
       "      <th>5110</th>\n",
       "      <th>5111</th>\n",
       "      <th>5112</th>\n",
       "      <th>5113</th>\n",
       "      <th>5114</th>\n",
       "      <th>5115</th>\n",
       "      <th>5116</th>\n",
       "      <th>5117</th>\n",
       "      <th>5118</th>\n",
       "      <th>5119</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.056671</td>\n",
       "      <td>-0.074099</td>\n",
       "      <td>-0.040746</td>\n",
       "      <td>0.217794</td>\n",
       "      <td>-0.197767</td>\n",
       "      <td>0.223628</td>\n",
       "      <td>-0.119524</td>\n",
       "      <td>0.080779</td>\n",
       "      <td>-0.021853</td>\n",
       "      <td>0.285255</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.130863</td>\n",
       "      <td>-0.107892</td>\n",
       "      <td>0.140624</td>\n",
       "      <td>-0.169443</td>\n",
       "      <td>0.165775</td>\n",
       "      <td>-0.120193</td>\n",
       "      <td>0.094954</td>\n",
       "      <td>-0.000617</td>\n",
       "      <td>-0.199790</td>\n",
       "      <td>0.001056</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.133009</td>\n",
       "      <td>-0.124348</td>\n",
       "      <td>-0.109747</td>\n",
       "      <td>0.222604</td>\n",
       "      <td>-0.261453</td>\n",
       "      <td>0.138052</td>\n",
       "      <td>-0.132696</td>\n",
       "      <td>0.162016</td>\n",
       "      <td>-0.073308</td>\n",
       "      <td>0.382890</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.105108</td>\n",
       "      <td>-0.053451</td>\n",
       "      <td>0.092585</td>\n",
       "      <td>-0.201894</td>\n",
       "      <td>0.094672</td>\n",
       "      <td>-0.104331</td>\n",
       "      <td>0.135752</td>\n",
       "      <td>0.036512</td>\n",
       "      <td>-0.230271</td>\n",
       "      <td>0.020399</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.155696</td>\n",
       "      <td>-0.016959</td>\n",
       "      <td>-0.179328</td>\n",
       "      <td>0.153639</td>\n",
       "      <td>-0.241203</td>\n",
       "      <td>0.130882</td>\n",
       "      <td>-0.101294</td>\n",
       "      <td>0.178330</td>\n",
       "      <td>-0.043959</td>\n",
       "      <td>0.144877</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.160722</td>\n",
       "      <td>-0.136943</td>\n",
       "      <td>0.049061</td>\n",
       "      <td>-0.175572</td>\n",
       "      <td>0.109034</td>\n",
       "      <td>-0.239965</td>\n",
       "      <td>-0.050614</td>\n",
       "      <td>-0.096664</td>\n",
       "      <td>-0.249927</td>\n",
       "      <td>0.020688</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.056310</td>\n",
       "      <td>-0.137829</td>\n",
       "      <td>-0.098538</td>\n",
       "      <td>0.162526</td>\n",
       "      <td>-0.252675</td>\n",
       "      <td>0.163847</td>\n",
       "      <td>-0.120923</td>\n",
       "      <td>0.205509</td>\n",
       "      <td>-0.078092</td>\n",
       "      <td>0.267088</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.124351</td>\n",
       "      <td>-0.181434</td>\n",
       "      <td>0.037783</td>\n",
       "      <td>-0.201551</td>\n",
       "      <td>0.092573</td>\n",
       "      <td>-0.168867</td>\n",
       "      <td>0.078524</td>\n",
       "      <td>-0.003103</td>\n",
       "      <td>-0.212419</td>\n",
       "      <td>0.049512</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.122564</td>\n",
       "      <td>-0.024471</td>\n",
       "      <td>-0.127546</td>\n",
       "      <td>0.165324</td>\n",
       "      <td>-0.258667</td>\n",
       "      <td>0.142153</td>\n",
       "      <td>-0.093198</td>\n",
       "      <td>0.185513</td>\n",
       "      <td>-0.060469</td>\n",
       "      <td>0.173319</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.155189</td>\n",
       "      <td>-0.155856</td>\n",
       "      <td>0.061196</td>\n",
       "      <td>-0.220575</td>\n",
       "      <td>0.092058</td>\n",
       "      <td>-0.220014</td>\n",
       "      <td>0.025014</td>\n",
       "      <td>-0.053914</td>\n",
       "      <td>-0.198191</td>\n",
       "      <td>0.036471</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>88</th>\n",
       "      <td>0.007245</td>\n",
       "      <td>-0.115889</td>\n",
       "      <td>-0.040796</td>\n",
       "      <td>0.235570</td>\n",
       "      <td>-0.216340</td>\n",
       "      <td>0.226666</td>\n",
       "      <td>-0.117476</td>\n",
       "      <td>0.136838</td>\n",
       "      <td>-0.047453</td>\n",
       "      <td>0.286242</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.166121</td>\n",
       "      <td>-0.146094</td>\n",
       "      <td>0.080064</td>\n",
       "      <td>-0.143502</td>\n",
       "      <td>0.155125</td>\n",
       "      <td>-0.067742</td>\n",
       "      <td>0.087349</td>\n",
       "      <td>-0.027154</td>\n",
       "      <td>-0.214921</td>\n",
       "      <td>0.023557</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89</th>\n",
       "      <td>0.143025</td>\n",
       "      <td>-0.030791</td>\n",
       "      <td>-0.110694</td>\n",
       "      <td>0.109217</td>\n",
       "      <td>-0.223111</td>\n",
       "      <td>0.105739</td>\n",
       "      <td>-0.079077</td>\n",
       "      <td>0.181002</td>\n",
       "      <td>-0.031332</td>\n",
       "      <td>0.150899</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.119124</td>\n",
       "      <td>-0.189840</td>\n",
       "      <td>-0.030339</td>\n",
       "      <td>-0.218860</td>\n",
       "      <td>0.098898</td>\n",
       "      <td>-0.174355</td>\n",
       "      <td>0.023488</td>\n",
       "      <td>-0.059298</td>\n",
       "      <td>-0.252023</td>\n",
       "      <td>0.029996</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90</th>\n",
       "      <td>0.068697</td>\n",
       "      <td>-0.038422</td>\n",
       "      <td>-0.142241</td>\n",
       "      <td>0.184863</td>\n",
       "      <td>-0.232976</td>\n",
       "      <td>0.162749</td>\n",
       "      <td>-0.080665</td>\n",
       "      <td>0.120402</td>\n",
       "      <td>-0.020260</td>\n",
       "      <td>0.204635</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.180917</td>\n",
       "      <td>-0.172996</td>\n",
       "      <td>0.046710</td>\n",
       "      <td>-0.179037</td>\n",
       "      <td>0.106190</td>\n",
       "      <td>-0.103997</td>\n",
       "      <td>-0.029128</td>\n",
       "      <td>0.000693</td>\n",
       "      <td>-0.243295</td>\n",
       "      <td>0.019351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>91</th>\n",
       "      <td>0.066839</td>\n",
       "      <td>-0.037917</td>\n",
       "      <td>-0.087912</td>\n",
       "      <td>0.137855</td>\n",
       "      <td>-0.218455</td>\n",
       "      <td>0.279858</td>\n",
       "      <td>-0.059348</td>\n",
       "      <td>0.068789</td>\n",
       "      <td>0.069967</td>\n",
       "      <td>0.299982</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.167717</td>\n",
       "      <td>-0.251613</td>\n",
       "      <td>-0.086687</td>\n",
       "      <td>-0.231941</td>\n",
       "      <td>0.015133</td>\n",
       "      <td>-0.017062</td>\n",
       "      <td>0.048716</td>\n",
       "      <td>-0.028251</td>\n",
       "      <td>-0.203544</td>\n",
       "      <td>-0.114390</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>0.069902</td>\n",
       "      <td>-0.039020</td>\n",
       "      <td>-0.122685</td>\n",
       "      <td>0.183626</td>\n",
       "      <td>-0.228307</td>\n",
       "      <td>0.150014</td>\n",
       "      <td>-0.113110</td>\n",
       "      <td>0.142315</td>\n",
       "      <td>-0.034828</td>\n",
       "      <td>0.243144</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.200931</td>\n",
       "      <td>-0.165114</td>\n",
       "      <td>0.059687</td>\n",
       "      <td>-0.133821</td>\n",
       "      <td>0.111872</td>\n",
       "      <td>-0.142763</td>\n",
       "      <td>-0.034693</td>\n",
       "      <td>-0.022357</td>\n",
       "      <td>-0.249541</td>\n",
       "      <td>0.047750</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>93 rows × 5120 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        0         1         2         3         4         5         6     \\\n",
       "0   0.056671 -0.074099 -0.040746  0.217794 -0.197767  0.223628 -0.119524   \n",
       "1   0.133009 -0.124348 -0.109747  0.222604 -0.261453  0.138052 -0.132696   \n",
       "2   0.155696 -0.016959 -0.179328  0.153639 -0.241203  0.130882 -0.101294   \n",
       "3   0.056310 -0.137829 -0.098538  0.162526 -0.252675  0.163847 -0.120923   \n",
       "4   0.122564 -0.024471 -0.127546  0.165324 -0.258667  0.142153 -0.093198   \n",
       "..       ...       ...       ...       ...       ...       ...       ...   \n",
       "88  0.007245 -0.115889 -0.040796  0.235570 -0.216340  0.226666 -0.117476   \n",
       "89  0.143025 -0.030791 -0.110694  0.109217 -0.223111  0.105739 -0.079077   \n",
       "90  0.068697 -0.038422 -0.142241  0.184863 -0.232976  0.162749 -0.080665   \n",
       "91  0.066839 -0.037917 -0.087912  0.137855 -0.218455  0.279858 -0.059348   \n",
       "92  0.069902 -0.039020 -0.122685  0.183626 -0.228307  0.150014 -0.113110   \n",
       "\n",
       "        7         8         9     ...      5110      5111      5112      5113  \\\n",
       "0   0.080779 -0.021853  0.285255  ... -0.130863 -0.107892  0.140624 -0.169443   \n",
       "1   0.162016 -0.073308  0.382890  ... -0.105108 -0.053451  0.092585 -0.201894   \n",
       "2   0.178330 -0.043959  0.144877  ... -0.160722 -0.136943  0.049061 -0.175572   \n",
       "3   0.205509 -0.078092  0.267088  ... -0.124351 -0.181434  0.037783 -0.201551   \n",
       "4   0.185513 -0.060469  0.173319  ... -0.155189 -0.155856  0.061196 -0.220575   \n",
       "..       ...       ...       ...  ...       ...       ...       ...       ...   \n",
       "88  0.136838 -0.047453  0.286242  ... -0.166121 -0.146094  0.080064 -0.143502   \n",
       "89  0.181002 -0.031332  0.150899  ... -0.119124 -0.189840 -0.030339 -0.218860   \n",
       "90  0.120402 -0.020260  0.204635  ... -0.180917 -0.172996  0.046710 -0.179037   \n",
       "91  0.068789  0.069967  0.299982  ... -0.167717 -0.251613 -0.086687 -0.231941   \n",
       "92  0.142315 -0.034828  0.243144  ... -0.200931 -0.165114  0.059687 -0.133821   \n",
       "\n",
       "        5114      5115      5116      5117      5118      5119  \n",
       "0   0.165775 -0.120193  0.094954 -0.000617 -0.199790  0.001056  \n",
       "1   0.094672 -0.104331  0.135752  0.036512 -0.230271  0.020399  \n",
       "2   0.109034 -0.239965 -0.050614 -0.096664 -0.249927  0.020688  \n",
       "3   0.092573 -0.168867  0.078524 -0.003103 -0.212419  0.049512  \n",
       "4   0.092058 -0.220014  0.025014 -0.053914 -0.198191  0.036471  \n",
       "..       ...       ...       ...       ...       ...       ...  \n",
       "88  0.155125 -0.067742  0.087349 -0.027154 -0.214921  0.023557  \n",
       "89  0.098898 -0.174355  0.023488 -0.059298 -0.252023  0.029996  \n",
       "90  0.106190 -0.103997 -0.029128  0.000693 -0.243295  0.019351  \n",
       "91  0.015133 -0.017062  0.048716 -0.028251 -0.203544 -0.114390  \n",
       "92  0.111872 -0.142763 -0.034693 -0.022357 -0.249541  0.047750  \n",
       "\n",
       "[93 rows x 5120 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Define target and features\n",
    "target = df[\"selectivity\"]\n",
    "\n",
    "# define features\n",
    "features = df.iloc[:,meta_data.shape[1]:]\n",
    "features"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Split data into train and test sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Split the data into training and testing sets\n",
    "X_train, X_test, y_train, y_test = train_test_split(features, target, test_size=0.2, random_state=42)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Running Regression Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: Lasso, PCA: 10, Train R2:0.70, Test R2:0.39, best params {'alpha': 0.03, 'max_iter': 100000}\n",
      "R2 CV Scores: [0.59486436 0.59747756 0.59173228 0.70847221 0.45296787 0.24423711\n",
      " 0.08173444 0.66038204 0.40777863 0.40015047]\n",
      "\n",
      "Model: SVR, PCA: 10, Train R2:0.72, Test R2:0.38, best params {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}\n",
      "R2 CV Scores: [0.67100482 0.59897804 0.45713505 0.79065261 0.56874652 0.4205358\n",
      " 0.33937803 0.73327618 0.35931135 0.44385726]\n",
      "\n",
      "Model: Lasso, PCA: 15, Train R2:0.79, Test R2:0.61, best params {'alpha': 0.02, 'max_iter': 100000}\n",
      "R2 CV Scores: [0.76017796 0.69234783 0.51873825 0.71319848 0.52171245 0.50028444\n",
      " 0.64131079 0.83207616 0.462041   0.24371672]\n",
      "\n",
      "Model: SVR, PCA: 15, Train R2:0.80, Test R2:0.64, best params {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}\n",
      "R2 CV Scores: [0.79083061 0.69741266 0.53324881 0.77018843 0.65842081 0.66468546\n",
      " 0.48634512 0.8735169  0.46067425 0.22685178]\n",
      "\n",
      "Model: Lasso, PCA: 20, Train R2:0.84, Test R2:0.64, best params {'alpha': 0.01, 'max_iter': 100000}\n",
      "R2 CV Scores: [0.77573985 0.71377012 0.55808966 0.7153109  0.67589526 0.55372059\n",
      " 0.1959536  0.87652631 0.48119777 0.34781265]\n",
      "\n",
      "Model: SVR, PCA: 20, Train R2:0.83, Test R2:0.62, best params {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}\n",
      "R2 CV Scores: [ 0.7680427   0.73571088  0.65379564  0.7205738   0.72194159  0.59605733\n",
      " -0.36648618  0.82221223  0.53588058  0.4665507 ]\n",
      "\n",
      "Model: Lasso, PCA: 25, Train R2:0.80, Test R2:0.58, best params {'alpha': 0.02, 'max_iter': 100000}\n",
      "R2 CV Scores: [0.77574817 0.71390393 0.55860289 0.71556148 0.63332091 0.5532274\n",
      " 0.19996458 0.84476723 0.48130411 0.34934764]\n",
      "\n",
      "Model: SVR, PCA: 25, Train R2:0.84, Test R2:0.61, best params {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}\n",
      "R2 CV Scores: [0.76211754 0.75575021 0.68024056 0.73230986 0.66237623 0.49794007\n",
      " 0.51642081 0.83048145 0.50044974 0.4884489 ]\n",
      "\n",
      "Model: Lasso, PCA: 30, Train R2:0.80, Test R2:0.58, best params {'alpha': 0.02, 'max_iter': 100000}\n",
      "R2 CV Scores: [ 0.78746958  0.71387568  0.5585328   0.71553185  0.58001906  0.55330931\n",
      "  0.23959179  0.84471533  0.48129242 -0.06312282]\n",
      "\n",
      "Model: SVR, PCA: 30, Train R2:0.85, Test R2:0.58, best params {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}\n",
      "R2 CV Scores: [0.80406457 0.79157279 0.75070362 0.72447619 0.5739767  0.52049332\n",
      " 0.6119747  0.79245647 0.50806453 0.48771747]\n",
      "\n",
      "Model: Lasso, PCA: 35, Train R2:0.80, Test R2:0.58, best params {'alpha': 0.02, 'max_iter': 100000}\n",
      "R2 CV Scores: [ 0.78698874  0.71389237  0.5585526   0.71552597  0.55531685  0.55328409\n",
      "  0.24289307  0.8447159   0.48131445 -0.0519841 ]\n",
      "\n",
      "Model: SVR, PCA: 35, Train R2:0.86, Test R2:0.61, best params {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}\n",
      "R2 CV Scores: [0.83549527 0.79305176 0.79212875 0.74079999 0.61434954 0.50974677\n",
      " 0.68483388 0.6763097  0.53342415 0.49791791]\n",
      "\n",
      "Model: Lasso, PCA: 40, Train R2:0.80, Test R2:0.58, best params {'alpha': 0.02, 'max_iter': 100000}\n",
      "R2 CV Scores: [0.81331303 0.77622909 0.55853608 0.71553439 0.63505907 0.5532641\n",
      " 0.35229112 0.84471077 0.55231561 0.09329793]\n",
      "\n",
      "Model: SVR, PCA: 40, Train R2:0.88, Test R2:0.61, best params {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}\n",
      "R2 CV Scores: [0.86029549 0.81588288 0.79570066 0.80222225 0.64916446 0.65712926\n",
      " 0.75990255 0.7939104  0.53389465 0.58183331]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "pcas = [10, 15, 20, 25, 30, 35, 40]\n",
    "\n",
    "results_final = pd.DataFrame()\n",
    "results_cv_final = pd.DataFrame()\n",
    "\n",
    "# Instantiate LeavePOut cross-validator \n",
    "loo = LeaveOneOut()  \n",
    "\n",
    "for num_pca in pcas:\n",
    "\n",
    "    pca = PCA(num_pca)\n",
    "    pca.fit(X_train)\n",
    "\n",
    "    X_train_pca = pca.transform(X_train)\n",
    "    X_test_pca = pca.transform(X_test)\n",
    "    \n",
    "    # store the results\n",
    "    results = pd.DataFrame()\n",
    "    results_cv = pd.DataFrame()\n",
    "\n",
    "    # Create a list of models to test\n",
    "    models = [Lasso(), SVR()]\n",
    "\n",
    "    # Create a dictionary of hyperparameters for each model\n",
    "    param_grid = [\n",
    "        # Lasso\n",
    "        {'alpha': np.arange(0.01, 1, 0.01),'max_iter': [100000]},\n",
    "        # SVR\n",
    "        {'C': [0.1, 1, 10], 'kernel': ['linear'], 'gamma': ['scale', 0.1, 0.01, 0.001]},\n",
    "        ]\n",
    "    \n",
    "    \n",
    "    # Iterate over the models and fit and predict using grid search\n",
    "    for model, param_grid in zip(models, param_grid):\n",
    "\n",
    "        # search best hyperparameter \n",
    "        grid_search = GridSearchCV(model, param_grid, scoring='r2', cv=10, n_jobs=25)\n",
    "\n",
    "        # train in the whole training set\n",
    "        grid_search.fit(X_train_pca, y_train)\n",
    "\n",
    "        # k-fold cross-validation on the training set\n",
    "        r2_cv_scores = cross_val_score(grid_search, X_train_pca, y_train, cv=10, scoring='r2', n_jobs=25)\n",
    "        mae_cv_scores = cross_val_score(grid_search, X_train_pca, y_train, cv=10, scoring='neg_mean_absolute_error', n_jobs=25)\n",
    "\n",
    "        # make predictions\n",
    "        train_pred = grid_search.predict(X_train_pca)\n",
    "        y_pred = grid_search.predict(X_test_pca)\n",
    "\n",
    "        # Model evaluation on test data\n",
    "        train_r2 = metrics.r2_score(y_train, train_pred)\n",
    "        train_mae = metrics.mean_absolute_error(y_train, train_pred)\n",
    "        train_mse = metrics.mean_squared_error(y_train, train_pred)\n",
    "        train_rmse = np.sqrt(train_mse)\n",
    "\n",
    "        # Model evaluation on test data\n",
    "        test_r2 = metrics.r2_score(y_test, y_pred)\n",
    "        test_mae = metrics.mean_absolute_error(y_test, y_pred)\n",
    "        test_mse = metrics.mean_squared_error(y_test, y_pred)\n",
    "        test_rmse = np.sqrt(test_mse)\n",
    "\n",
    "        model_name = grid_search.best_estimator_.__class__.__name__\n",
    "        embed_name = dir_path.split('/')[-2]\n",
    "        best_params = str(grid_search.best_params_)\n",
    "\n",
    "        # print best scores\n",
    "        print(f\"Model: {model_name}, PCA: {num_pca}, Train R2:{train_r2:.2f}, Test R2:{test_r2:.2f}, best params {best_params}\")\n",
    "        print(\"R2 CV Scores:\", r2_cv_scores)\n",
    "        print()\n",
    "\n",
    "        # Saving results\n",
    "        res_dict = {}\n",
    "        res_dict[\"Embedding\"] = embed_name\n",
    "        res_dict[\"model\"] = model_name\n",
    "        res_dict[\"num_pca\"] = num_pca\n",
    "        res_dict[\"Train_R2_score\"] = train_r2.round(2)\n",
    "        res_dict[\"Train_MAE_score\"] = train_mae.round(2)\n",
    "        res_dict[\"Train_RMSE_score\"] = train_rmse.round(2)\n",
    "\n",
    "        # test scores\n",
    "        res_dict[\"Test_R2_score\"] = test_r2.round(2)\n",
    "        res_dict[\"Test_MAE_score\"] = test_mae.round(2)\n",
    "        res_dict[\"Test_RMSE_score\"] = test_rmse.round(2)\n",
    "        res_dict[\"Test_Best_params\"] = best_params\n",
    "\n",
    "        # updating data frame with results\n",
    "        results = pd.concat([results, pd.DataFrame(res_dict, index=[0])]).reset_index(drop=True)\n",
    "\n",
    "        #save results from cross-validation\n",
    "        res_cv_dict = {\n",
    "            \"model\": [model_name for i in range(len(mae_cv_scores))],\n",
    "            \"num_pca\": [num_pca for i in range(len(mae_cv_scores))],\n",
    "            \"r2_CV_scores\": r2_cv_scores,\n",
    "            \"Mae_CV_scores\": mae_cv_scores,\n",
    "        }\n",
    "        #updating data frame with results\n",
    "        results_cv = pd.concat([results_cv, pd.DataFrame(res_cv_dict)]).reset_index(drop=True)\n",
    "\n",
    "    \n",
    "    results_final = pd.concat([results_final, results]).reset_index(drop=True)\n",
    "    results_cv_final = pd.concat([results_cv_final, results_cv]).reset_index(drop=True)\n",
    "\n",
    "# save results\n",
    "#results_final.to_excel('results/results_regression_selectivity_v02.xlsx')\n",
    "#results_cv_final.to_excel('results/results_cv_regression_selectivity_v02.xlsx')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Results Visualization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### MAE across the validation sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#results_cv_final = pd.read_excel('results/results_cv_regression_selectivity_v02.xlsx', index_col=0)\n",
    "sns.boxplot(data=results_cv_final, x='num_pca', y='Mae_CV_scores', hue='model' , showfliers=False)\n",
    "plt.title('Cross-validation MAE scores')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## R2 across the validation sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.boxplot(data=results_cv_final, x='num_pca', y='r2_CV_scores', hue='model' , showfliers=False)\n",
    "plt.title('Cross-validation R2 scores')\n",
    "plt.ylim(-1, 1)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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>Embedding</th>\n",
       "      <th>model</th>\n",
       "      <th>num_pca</th>\n",
       "      <th>Train_R2_score</th>\n",
       "      <th>Train_MAE_score</th>\n",
       "      <th>Train_RMSE_score</th>\n",
       "      <th>Test_R2_score</th>\n",
       "      <th>Test_MAE_score</th>\n",
       "      <th>Test_RMSE_score</th>\n",
       "      <th>Test_Best_params</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>Lasso</td>\n",
       "      <td>10</td>\n",
       "      <td>0.70</td>\n",
       "      <td>0.34</td>\n",
       "      <td>0.42</td>\n",
       "      <td>0.39</td>\n",
       "      <td>0.45</td>\n",
       "      <td>0.64</td>\n",
       "      <td>{'alpha': 0.03, 'max_iter': 100000}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>SVR</td>\n",
       "      <td>10</td>\n",
       "      <td>0.72</td>\n",
       "      <td>0.31</td>\n",
       "      <td>0.41</td>\n",
       "      <td>0.38</td>\n",
       "      <td>0.46</td>\n",
       "      <td>0.65</td>\n",
       "      <td>{'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>Lasso</td>\n",
       "      <td>15</td>\n",
       "      <td>0.79</td>\n",
       "      <td>0.28</td>\n",
       "      <td>0.36</td>\n",
       "      <td>0.61</td>\n",
       "      <td>0.35</td>\n",
       "      <td>0.52</td>\n",
       "      <td>{'alpha': 0.02, 'max_iter': 100000}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>SVR</td>\n",
       "      <td>15</td>\n",
       "      <td>0.80</td>\n",
       "      <td>0.26</td>\n",
       "      <td>0.35</td>\n",
       "      <td>0.64</td>\n",
       "      <td>0.35</td>\n",
       "      <td>0.50</td>\n",
       "      <td>{'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>Lasso</td>\n",
       "      <td>20</td>\n",
       "      <td>0.84</td>\n",
       "      <td>0.24</td>\n",
       "      <td>0.31</td>\n",
       "      <td>0.64</td>\n",
       "      <td>0.33</td>\n",
       "      <td>0.49</td>\n",
       "      <td>{'alpha': 0.01, 'max_iter': 100000}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>SVR</td>\n",
       "      <td>20</td>\n",
       "      <td>0.83</td>\n",
       "      <td>0.24</td>\n",
       "      <td>0.32</td>\n",
       "      <td>0.62</td>\n",
       "      <td>0.33</td>\n",
       "      <td>0.51</td>\n",
       "      <td>{'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>Lasso</td>\n",
       "      <td>25</td>\n",
       "      <td>0.80</td>\n",
       "      <td>0.27</td>\n",
       "      <td>0.35</td>\n",
       "      <td>0.58</td>\n",
       "      <td>0.36</td>\n",
       "      <td>0.53</td>\n",
       "      <td>{'alpha': 0.02, 'max_iter': 100000}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>SVR</td>\n",
       "      <td>25</td>\n",
       "      <td>0.84</td>\n",
       "      <td>0.23</td>\n",
       "      <td>0.31</td>\n",
       "      <td>0.61</td>\n",
       "      <td>0.35</td>\n",
       "      <td>0.52</td>\n",
       "      <td>{'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>Lasso</td>\n",
       "      <td>30</td>\n",
       "      <td>0.80</td>\n",
       "      <td>0.27</td>\n",
       "      <td>0.35</td>\n",
       "      <td>0.58</td>\n",
       "      <td>0.36</td>\n",
       "      <td>0.53</td>\n",
       "      <td>{'alpha': 0.02, 'max_iter': 100000}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>SVR</td>\n",
       "      <td>30</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.22</td>\n",
       "      <td>0.30</td>\n",
       "      <td>0.58</td>\n",
       "      <td>0.38</td>\n",
       "      <td>0.53</td>\n",
       "      <td>{'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>Lasso</td>\n",
       "      <td>35</td>\n",
       "      <td>0.80</td>\n",
       "      <td>0.27</td>\n",
       "      <td>0.35</td>\n",
       "      <td>0.58</td>\n",
       "      <td>0.36</td>\n",
       "      <td>0.53</td>\n",
       "      <td>{'alpha': 0.02, 'max_iter': 100000}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>SVR</td>\n",
       "      <td>35</td>\n",
       "      <td>0.86</td>\n",
       "      <td>0.21</td>\n",
       "      <td>0.29</td>\n",
       "      <td>0.61</td>\n",
       "      <td>0.37</td>\n",
       "      <td>0.52</td>\n",
       "      <td>{'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>Lasso</td>\n",
       "      <td>40</td>\n",
       "      <td>0.80</td>\n",
       "      <td>0.27</td>\n",
       "      <td>0.35</td>\n",
       "      <td>0.58</td>\n",
       "      <td>0.36</td>\n",
       "      <td>0.53</td>\n",
       "      <td>{'alpha': 0.02, 'max_iter': 100000}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>SVR</td>\n",
       "      <td>40</td>\n",
       "      <td>0.88</td>\n",
       "      <td>0.19</td>\n",
       "      <td>0.27</td>\n",
       "      <td>0.61</td>\n",
       "      <td>0.37</td>\n",
       "      <td>0.51</td>\n",
       "      <td>{'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Embedding  model  num_pca  Train_R2_score  Train_MAE_score  \\\n",
       "0   esm2_15B_PG1  Lasso       10            0.70             0.34   \n",
       "1   esm2_15B_PG1    SVR       10            0.72             0.31   \n",
       "2   esm2_15B_PG1  Lasso       15            0.79             0.28   \n",
       "3   esm2_15B_PG1    SVR       15            0.80             0.26   \n",
       "4   esm2_15B_PG1  Lasso       20            0.84             0.24   \n",
       "5   esm2_15B_PG1    SVR       20            0.83             0.24   \n",
       "6   esm2_15B_PG1  Lasso       25            0.80             0.27   \n",
       "7   esm2_15B_PG1    SVR       25            0.84             0.23   \n",
       "8   esm2_15B_PG1  Lasso       30            0.80             0.27   \n",
       "9   esm2_15B_PG1    SVR       30            0.85             0.22   \n",
       "10  esm2_15B_PG1  Lasso       35            0.80             0.27   \n",
       "11  esm2_15B_PG1    SVR       35            0.86             0.21   \n",
       "12  esm2_15B_PG1  Lasso       40            0.80             0.27   \n",
       "13  esm2_15B_PG1    SVR       40            0.88             0.19   \n",
       "\n",
       "    Train_RMSE_score  Test_R2_score  Test_MAE_score  Test_RMSE_score  \\\n",
       "0               0.42           0.39            0.45             0.64   \n",
       "1               0.41           0.38            0.46             0.65   \n",
       "2               0.36           0.61            0.35             0.52   \n",
       "3               0.35           0.64            0.35             0.50   \n",
       "4               0.31           0.64            0.33             0.49   \n",
       "5               0.32           0.62            0.33             0.51   \n",
       "6               0.35           0.58            0.36             0.53   \n",
       "7               0.31           0.61            0.35             0.52   \n",
       "8               0.35           0.58            0.36             0.53   \n",
       "9               0.30           0.58            0.38             0.53   \n",
       "10              0.35           0.58            0.36             0.53   \n",
       "11              0.29           0.61            0.37             0.52   \n",
       "12              0.35           0.58            0.36             0.53   \n",
       "13              0.27           0.61            0.37             0.51   \n",
       "\n",
       "                                    Test_Best_params  \n",
       "0                {'alpha': 0.03, 'max_iter': 100000}  \n",
       "1   {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}  \n",
       "2                {'alpha': 0.02, 'max_iter': 100000}  \n",
       "3   {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}  \n",
       "4                {'alpha': 0.01, 'max_iter': 100000}  \n",
       "5   {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}  \n",
       "6                {'alpha': 0.02, 'max_iter': 100000}  \n",
       "7   {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}  \n",
       "8                {'alpha': 0.02, 'max_iter': 100000}  \n",
       "9   {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}  \n",
       "10               {'alpha': 0.02, 'max_iter': 100000}  \n",
       "11  {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}  \n",
       "12               {'alpha': 0.02, 'max_iter': 100000}  \n",
       "13  {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#results_final = pd.read_excel('results/results_regression_selectivity_v02.xlsx', index_col=0)\n",
    "results_final"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train x Test Model performances"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Reshape the dataframe to long format\n",
    "melted_df = pd.melt(results_final, \n",
    "                    id_vars=['num_pca', 'model'], \n",
    "                    value_vars=['Test_R2_score', 'Train_R2_score'], \n",
    "                    var_name='Score Type', value_name='R2 Score')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1289.12x500 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Example using FacetGrid for a categorical breakdown\n",
    "g = sns.FacetGrid(melted_df, col=\"Score Type\", sharey=False, height=5, aspect=1.2, ylim=(0,1), col_order=['Train_R2_score', 'Test_R2_score'])\n",
    "g.map_dataframe(sns.barplot, x=\"num_pca\", y=\"R2 Score\", hue='model', palette='muted')\n",
    "g.set_axis_labels(\"Number of PCAs\", \"R2 Score\")\n",
    "g.set_titles(col_template=\"{col_name}\", row_template=\"{row_name}\")\n",
    "g.add_legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To determine if the model is overfitting, it's important to evaluate its performance on a validation set or using cross-validation.  \n",
    "If the model's performance on the validation set is significantly worse than its performance on the training set, then it may be overfitting.  \n",
    "In such cases, it may be necessary to adjust the hyperparameters of the model or to use a more complex kernel function."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Fine Tunning Model with best hyperparameter found with Gridsearch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "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>Embedding</th>\n",
       "      <th>model</th>\n",
       "      <th>num_pca</th>\n",
       "      <th>Train_R2_score</th>\n",
       "      <th>Train_MAE_score</th>\n",
       "      <th>Train_RMSE_score</th>\n",
       "      <th>Test_R2_score</th>\n",
       "      <th>Test_MAE_score</th>\n",
       "      <th>Test_RMSE_score</th>\n",
       "      <th>Test_Best_params</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>Lasso</td>\n",
       "      <td>20</td>\n",
       "      <td>0.84</td>\n",
       "      <td>0.24</td>\n",
       "      <td>0.31</td>\n",
       "      <td>0.64</td>\n",
       "      <td>0.33</td>\n",
       "      <td>0.49</td>\n",
       "      <td>{'alpha': 0.01, 'max_iter': 100000}</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>esm2_15B_PG1</td>\n",
       "      <td>SVR</td>\n",
       "      <td>20</td>\n",
       "      <td>0.83</td>\n",
       "      <td>0.24</td>\n",
       "      <td>0.32</td>\n",
       "      <td>0.62</td>\n",
       "      <td>0.33</td>\n",
       "      <td>0.51</td>\n",
       "      <td>{'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      Embedding  model  num_pca  Train_R2_score  Train_MAE_score  \\\n",
       "4  esm2_15B_PG1  Lasso       20            0.84             0.24   \n",
       "5  esm2_15B_PG1    SVR       20            0.83             0.24   \n",
       "\n",
       "   Train_RMSE_score  Test_R2_score  Test_MAE_score  Test_RMSE_score  \\\n",
       "4              0.31           0.64            0.33             0.49   \n",
       "5              0.32           0.62            0.33             0.51   \n",
       "\n",
       "                                   Test_Best_params  \n",
       "4               {'alpha': 0.01, 'max_iter': 100000}  \n",
       "5  {'C': 0.1, 'gamma': 'scale', 'kernel': 'linear'}  "
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results_final.query('num_pca == 20')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Split the data into training and testing sets\n",
    "X_train, X_test, y_train, y_test = train_test_split(features, target, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_pca_components = 20\n",
    "\n",
    "pca = PCA(num_pca_components)\n",
    "pca.fit(X_train)\n",
    "\n",
    "X_train_pca = pca.transform(X_train)\n",
    "X_test_pca = pca.transform(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Lasso"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train R2 score: 0.84\n",
      "Test R2 score: 0.64\n"
     ]
    }
   ],
   "source": [
    "lasso = Lasso(\n",
    "    alpha=0.01,\n",
    "    max_iter=10000    \n",
    ")\n",
    "\n",
    "# fined tuned SVR performance\n",
    "lasso.fit(X_train_pca, y_train)\n",
    "\n",
    "# print R2 score\n",
    "print(f'Train R2 score: {lasso.score(X_train_pca, y_train):.2f}')\n",
    "print(f'Test R2 score: {lasso.score(X_test_pca, y_test):.2f}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_pred = lasso.predict(X_train_pca)\n",
    "y_pred = lasso.predict(X_test_pca)\n",
    "\n",
    "plt.scatter(train_pred, y_train, label=\"Train samples\", c='#d95f02')\n",
    "plt.scatter(y_pred, y_test, label=\"Test samples\", c='#7570b3')\n",
    "plt.title(f\"Model: Linear Regressor\")\n",
    "plt.xlabel(\"Predicted Selectivity score\")\n",
    "plt.ylabel(\"True Selectivity score\")\n",
    "plt.plot([0, 6], [0, 6], c='k', zorder=0) \n",
    "plt.text(0, 5, f'Train R$^2$: {lasso.score(X_train_pca, y_train).round(2)}', fontsize = 9)\n",
    "plt.text(0, 4.5, f'Test R$^2$: {lasso.score(X_test_pca, y_test).round(2)}', fontsize = 9)\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SVR"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train R2 score: 0.83\n",
      "Test R2 score: 0.62\n"
     ]
    }
   ],
   "source": [
    "# train support vector machine regressor\n",
    "svmR = SVR(\n",
    "    C=0.1, \n",
    "    gamma='scale',\n",
    "    kernel='linear',\n",
    "    )\n",
    "\n",
    "# fined tuned SVR performance\n",
    "svmR.fit(X_train_pca, y_train)\n",
    "\n",
    "# print R2 score\n",
    "print(f'Train R2 score: {svmR.score(X_train_pca, y_train):.2f}')\n",
    "print(f'Test R2 score: {svmR.score(X_test_pca, y_test):.2f}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_pred = svmR.predict(X_train_pca)\n",
    "y_pred = svmR.predict(X_test_pca)\n",
    "\n",
    "plt.scatter(train_pred, y_train, label=\"Train samples\", c='#d95f02')\n",
    "plt.scatter(y_pred, y_test, label=\"Test samples\", c='#7570b3')\n",
    "plt.title(f\"Model: Linear Regressor\")\n",
    "plt.xlabel(\"Predicted Selectivity score\")\n",
    "plt.ylabel(\"True Selectivity score\")\n",
    "plt.plot([0, 6], [0, 6], c='k', zorder=0) \n",
    "plt.text(0, 5, f'Train R$^2$: {svmR.score(X_train_pca, y_train).round(2)}', fontsize = 9)\n",
    "plt.text(0, 4.5, f'Test R$^2$: {svmR.score(X_test_pca, y_test).round(2)}', fontsize = 9)\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Saving model and PCA weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['models/weights_pca20_esm2_15B_selectivity.pkl']"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "## save the model\n",
    "joblib.dump(lasso, 'models/model_v02_svmR20_esm2_15B_selectivity_regression.pkl')\n",
    "joblib.dump(pca, 'models/weights_pca20_esm2_15B_selectivity.pkl')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import torch\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import pickle\n",
    "import joblib"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Loading the model\n",
    "model = joblib.load('models/model_v02_svmR20_esm2_15B_selectivity_regression.pkl')\n",
    "\n",
    "# Loading PCA weights\n",
    "pca = joblib.load('models/weights_pca20_esm2_15B_selectivity.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "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>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>...</th>\n",
       "      <th>5110</th>\n",
       "      <th>5111</th>\n",
       "      <th>5112</th>\n",
       "      <th>5113</th>\n",
       "      <th>5114</th>\n",
       "      <th>5115</th>\n",
       "      <th>5116</th>\n",
       "      <th>5117</th>\n",
       "      <th>5118</th>\n",
       "      <th>5119</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.109975</td>\n",
       "      <td>-0.134590</td>\n",
       "      <td>-0.111443</td>\n",
       "      <td>0.218644</td>\n",
       "      <td>-0.237748</td>\n",
       "      <td>0.131493</td>\n",
       "      <td>-0.119502</td>\n",
       "      <td>0.234007</td>\n",
       "      <td>-0.042554</td>\n",
       "      <td>0.338132</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.136047</td>\n",
       "      <td>-0.087660</td>\n",
       "      <td>0.107472</td>\n",
       "      <td>-0.270478</td>\n",
       "      <td>0.122341</td>\n",
       "      <td>-0.089307</td>\n",
       "      <td>0.137297</td>\n",
       "      <td>0.075365</td>\n",
       "      <td>-0.198461</td>\n",
       "      <td>-0.035045</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.129701</td>\n",
       "      <td>-0.184889</td>\n",
       "      <td>-0.115536</td>\n",
       "      <td>0.161651</td>\n",
       "      <td>-0.224050</td>\n",
       "      <td>0.112427</td>\n",
       "      <td>-0.157890</td>\n",
       "      <td>0.269912</td>\n",
       "      <td>-0.049523</td>\n",
       "      <td>0.307800</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.106790</td>\n",
       "      <td>-0.129787</td>\n",
       "      <td>0.076798</td>\n",
       "      <td>-0.208119</td>\n",
       "      <td>0.098149</td>\n",
       "      <td>-0.133004</td>\n",
       "      <td>0.089372</td>\n",
       "      <td>0.093026</td>\n",
       "      <td>-0.170869</td>\n",
       "      <td>-0.013929</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.099468</td>\n",
       "      <td>-0.114953</td>\n",
       "      <td>-0.117187</td>\n",
       "      <td>0.183187</td>\n",
       "      <td>-0.209792</td>\n",
       "      <td>0.091260</td>\n",
       "      <td>-0.130392</td>\n",
       "      <td>0.256235</td>\n",
       "      <td>-0.053187</td>\n",
       "      <td>0.229403</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.195330</td>\n",
       "      <td>-0.116773</td>\n",
       "      <td>0.137988</td>\n",
       "      <td>-0.164647</td>\n",
       "      <td>0.119364</td>\n",
       "      <td>-0.178338</td>\n",
       "      <td>0.031593</td>\n",
       "      <td>0.029154</td>\n",
       "      <td>-0.215111</td>\n",
       "      <td>-0.017543</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.156776</td>\n",
       "      <td>-0.105166</td>\n",
       "      <td>-0.063266</td>\n",
       "      <td>0.204592</td>\n",
       "      <td>-0.202521</td>\n",
       "      <td>0.076685</td>\n",
       "      <td>-0.140390</td>\n",
       "      <td>0.251750</td>\n",
       "      <td>-0.082083</td>\n",
       "      <td>0.299491</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.100395</td>\n",
       "      <td>-0.043842</td>\n",
       "      <td>0.144827</td>\n",
       "      <td>-0.125664</td>\n",
       "      <td>0.134358</td>\n",
       "      <td>-0.157067</td>\n",
       "      <td>0.040382</td>\n",
       "      <td>0.030294</td>\n",
       "      <td>-0.248444</td>\n",
       "      <td>0.020361</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.148965</td>\n",
       "      <td>-0.141826</td>\n",
       "      <td>-0.064849</td>\n",
       "      <td>0.198645</td>\n",
       "      <td>-0.237333</td>\n",
       "      <td>0.096351</td>\n",
       "      <td>-0.192772</td>\n",
       "      <td>0.201036</td>\n",
       "      <td>-0.055550</td>\n",
       "      <td>0.323271</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.108473</td>\n",
       "      <td>-0.066790</td>\n",
       "      <td>0.079047</td>\n",
       "      <td>-0.211575</td>\n",
       "      <td>0.114173</td>\n",
       "      <td>-0.089037</td>\n",
       "      <td>0.146931</td>\n",
       "      <td>0.012782</td>\n",
       "      <td>-0.173229</td>\n",
       "      <td>0.016016</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 5120 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       0         1         2         3         4         5         6     \\\n",
       "0  0.109975 -0.134590 -0.111443  0.218644 -0.237748  0.131493 -0.119502   \n",
       "1  0.129701 -0.184889 -0.115536  0.161651 -0.224050  0.112427 -0.157890   \n",
       "2  0.099468 -0.114953 -0.117187  0.183187 -0.209792  0.091260 -0.130392   \n",
       "3  0.156776 -0.105166 -0.063266  0.204592 -0.202521  0.076685 -0.140390   \n",
       "4  0.148965 -0.141826 -0.064849  0.198645 -0.237333  0.096351 -0.192772   \n",
       "\n",
       "       7         8         9     ...      5110      5111      5112      5113  \\\n",
       "0  0.234007 -0.042554  0.338132  ... -0.136047 -0.087660  0.107472 -0.270478   \n",
       "1  0.269912 -0.049523  0.307800  ... -0.106790 -0.129787  0.076798 -0.208119   \n",
       "2  0.256235 -0.053187  0.229403  ... -0.195330 -0.116773  0.137988 -0.164647   \n",
       "3  0.251750 -0.082083  0.299491  ... -0.100395 -0.043842  0.144827 -0.125664   \n",
       "4  0.201036 -0.055550  0.323271  ... -0.108473 -0.066790  0.079047 -0.211575   \n",
       "\n",
       "       5114      5115      5116      5117      5118      5119  \n",
       "0  0.122341 -0.089307  0.137297  0.075365 -0.198461 -0.035045  \n",
       "1  0.098149 -0.133004  0.089372  0.093026 -0.170869 -0.013929  \n",
       "2  0.119364 -0.178338  0.031593  0.029154 -0.215111 -0.017543  \n",
       "3  0.134358 -0.157067  0.040382  0.030294 -0.248444  0.020361  \n",
       "4  0.114173 -0.089037  0.146931  0.012782 -0.173229  0.016016  \n",
       "\n",
       "[5 rows x 5120 columns]"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# load meta_data with mutations on PG1\n",
    "meta_data = pd.read_csv('results/v01.2/predictions_validation_set_v01.2.csv')\n",
    "\n",
    "####################### Load the embedding ######################\n",
    "dir_path = 'data/embedding/esm2_15B_PG1_validation/'\n",
    "embeddings = {}\n",
    "for file in os.listdir(dir_path):\n",
    "        file_path = str(dir_path) + file \n",
    "        label = file.split('.pt')[0]\n",
    "        embeddings[label] = np.array(torch.load(file_path)['mean_representations'][48])\n",
    "\n",
    "\n",
    "# here I creating a dictionary with ID as key and semantic value as value \n",
    "data = meta_data.merge(pd.DataFrame.from_dict(embeddings, orient='index'), how='inner', left_on='ID2', right_index=True)\n",
    "\n",
    "\n",
    "features = data.iloc[:,meta_data.shape[1]:]\n",
    "features.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# calculates the probability of being active\n",
    "pred = model.predict(pca.transform(features))\n",
    "\n",
    "sum(pred < 0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "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>ID2</th>\n",
       "      <th>Sequence</th>\n",
       "      <th>RGGRLCYCRRRFCVCVGR</th>\n",
       "      <th>Select</th>\n",
       "      <th>Charge</th>\n",
       "      <th>group</th>\n",
       "      <th>MIC MH</th>\n",
       "      <th>%hemo</th>\n",
       "      <th>L10Select</th>\n",
       "      <th>hemolysis_pred_v01.2</th>\n",
       "      <th>selectivity_pred_v01.2</th>\n",
       "      <th>activity_pred_v01.2</th>\n",
       "      <th>hemolysis_pred_v02</th>\n",
       "      <th>selectivity_pred_v02</th>\n",
       "      <th>activity_pred_v02</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>PG1_3mut_1739000</td>\n",
       "      <td>bsPG-1.1</td>\n",
       "      <td>RGGRLQYCRRRFCVARGR</td>\n",
       "      <td>-----Q--------AR--</td>\n",
       "      <td>-0.82</td>\n",
       "      <td>6.742145</td>\n",
       "      <td>&lt;-0.25</td>\n",
       "      <td>16</td>\n",
       "      <td>0.428850</td>\n",
       "      <td>0.836425</td>\n",
       "      <td>-7.675692</td>\n",
       "      <td>-0.987362</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-4.054558</td>\n",
       "      <td>0.448899</td>\n",
       "      <td>92.34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PG1_3mut_1218602</td>\n",
       "      <td>bsPG-1.2</td>\n",
       "      <td>RGGRLQYCRRRGCVTVGR</td>\n",
       "      <td>-----Q-----G--T---</td>\n",
       "      <td>-0.76</td>\n",
       "      <td>5.742155</td>\n",
       "      <td>&lt;-0.25</td>\n",
       "      <td>4</td>\n",
       "      <td>0.341131</td>\n",
       "      <td>0.134981</td>\n",
       "      <td>-14.298704</td>\n",
       "      <td>-1.215473</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-6.763184</td>\n",
       "      <td>0.368799</td>\n",
       "      <td>88.34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>PG1_3mut_1896424</td>\n",
       "      <td>bsPG-1.3</td>\n",
       "      <td>RGGRLAYCRRRDCTCVGR</td>\n",
       "      <td>-----A-----D-T----</td>\n",
       "      <td>-0.65</td>\n",
       "      <td>4.651497</td>\n",
       "      <td>CP</td>\n",
       "      <td>&gt;64</td>\n",
       "      <td>-0.165692</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-14.385390</td>\n",
       "      <td>-1.305665</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-8.310135</td>\n",
       "      <td>0.390410</td>\n",
       "      <td>95.37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>PG1_3mut_1275003</td>\n",
       "      <td>bsPG-1.4</td>\n",
       "      <td>RGGRLDYCRRRFCVTTGR</td>\n",
       "      <td>-----D--------TT--</td>\n",
       "      <td>-0.56</td>\n",
       "      <td>4.742406</td>\n",
       "      <td>&lt;-0.25</td>\n",
       "      <td>64</td>\n",
       "      <td>0.165692</td>\n",
       "      <td>1.025482</td>\n",
       "      <td>-19.078636</td>\n",
       "      <td>-0.686002</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-2.867607</td>\n",
       "      <td>0.552553</td>\n",
       "      <td>90.87</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>PG1_3mut_257802</td>\n",
       "      <td>bsPG-1.5</td>\n",
       "      <td>RGGRLQACRRRFCVTVGR</td>\n",
       "      <td>-----QA-------T---</td>\n",
       "      <td>-0.44</td>\n",
       "      <td>5.744660</td>\n",
       "      <td>&lt;-0.25</td>\n",
       "      <td>8</td>\n",
       "      <td>0.721248</td>\n",
       "      <td>0.761174</td>\n",
       "      <td>-3.053714</td>\n",
       "      <td>-0.247956</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-3.139629</td>\n",
       "      <td>0.636513</td>\n",
       "      <td>94.94</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>PG1_3mut_1210489</td>\n",
       "      <td>bsPG-1.6</td>\n",
       "      <td>RGGRLRYCRRRFDVSVGR</td>\n",
       "      <td>-----R------D-S---</td>\n",
       "      <td>-0.43</td>\n",
       "      <td>5.833305</td>\n",
       "      <td>&lt;-0.25</td>\n",
       "      <td>16</td>\n",
       "      <td>0.735867</td>\n",
       "      <td>1.070920</td>\n",
       "      <td>-13.542688</td>\n",
       "      <td>-0.695068</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-2.812324</td>\n",
       "      <td>0.449201</td>\n",
       "      <td>90.10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>PG1_3mut_1536372</td>\n",
       "      <td>bsPG-1.7</td>\n",
       "      <td>RGGRLCTARRRFCVRVGR</td>\n",
       "      <td>------TA------R---</td>\n",
       "      <td>-0.43</td>\n",
       "      <td>6.744650</td>\n",
       "      <td>&lt;-0.25</td>\n",
       "      <td>2</td>\n",
       "      <td>0.906433</td>\n",
       "      <td>0.258366</td>\n",
       "      <td>-3.345207</td>\n",
       "      <td>-0.416047</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-3.347273</td>\n",
       "      <td>0.609653</td>\n",
       "      <td>94.73</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>PG1_3mut_1147726</td>\n",
       "      <td>bsPG-1.8</td>\n",
       "      <td>RGGRLTYCRRRFCTAVGR</td>\n",
       "      <td>-----T-------TA---</td>\n",
       "      <td>-0.40</td>\n",
       "      <td>5.742155</td>\n",
       "      <td>&lt;-0.25</td>\n",
       "      <td>16</td>\n",
       "      <td>0.043860</td>\n",
       "      <td>-0.153815</td>\n",
       "      <td>-12.337255</td>\n",
       "      <td>-0.839417</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-1.988224</td>\n",
       "      <td>0.644135</td>\n",
       "      <td>98.64</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>PG1_3mut_1980555</td>\n",
       "      <td>bsPG-1.9</td>\n",
       "      <td>RGGRLTYCRRRDCVAVGR</td>\n",
       "      <td>-----T-----D--A---</td>\n",
       "      <td>-0.29</td>\n",
       "      <td>4.742406</td>\n",
       "      <td>&lt;-0.25</td>\n",
       "      <td>32</td>\n",
       "      <td>0.116959</td>\n",
       "      <td>0.573184</td>\n",
       "      <td>-21.809423</td>\n",
       "      <td>-0.699560</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-13.069695</td>\n",
       "      <td>0.206361</td>\n",
       "      <td>93.60</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>PG1_3mut_1658870</td>\n",
       "      <td>bsPG-1.10</td>\n",
       "      <td>RGGRLAYCRRRFCVDTGR</td>\n",
       "      <td>-----A--------DT--</td>\n",
       "      <td>-0.28</td>\n",
       "      <td>4.742406</td>\n",
       "      <td>&lt;-0.25</td>\n",
       "      <td>64</td>\n",
       "      <td>0.092593</td>\n",
       "      <td>0.772756</td>\n",
       "      <td>-9.999792</td>\n",
       "      <td>-0.342884</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-5.106495</td>\n",
       "      <td>0.504043</td>\n",
       "      <td>93.70</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>PG1_3mut_1618529</td>\n",
       "      <td>bsPG-1.11</td>\n",
       "      <td>RGGRLRYCRRRGTVCVGR</td>\n",
       "      <td>-----R-----GT-----</td>\n",
       "      <td>-0.28</td>\n",
       "      <td>6.742145</td>\n",
       "      <td>CP</td>\n",
       "      <td>&gt;64</td>\n",
       "      <td>0.268031</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-5.660681</td>\n",
       "      <td>-0.654460</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-1.498447</td>\n",
       "      <td>0.631221</td>\n",
       "      <td>83.26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>PG1_3mut_720432</td>\n",
       "      <td>bsPG-1.12</td>\n",
       "      <td>RGGRLQYCRRRFCARVGR</td>\n",
       "      <td>-----Q-------AR---</td>\n",
       "      <td>-0.27</td>\n",
       "      <td>6.742145</td>\n",
       "      <td>&lt;-0.25</td>\n",
       "      <td>16</td>\n",
       "      <td>0.141326</td>\n",
       "      <td>0.354341</td>\n",
       "      <td>-2.912143</td>\n",
       "      <td>-0.348571</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-4.230751</td>\n",
       "      <td>0.461692</td>\n",
       "      <td>90.30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>PG1_3mut_1247939</td>\n",
       "      <td>bsPG-1.13</td>\n",
       "      <td>RGGRLCYCRRRNCVTTGR</td>\n",
       "      <td>-----------N--TT--</td>\n",
       "      <td>-0.25</td>\n",
       "      <td>5.651245</td>\n",
       "      <td>-0.25-0</td>\n",
       "      <td>&gt;64</td>\n",
       "      <td>-0.058480</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-14.079775</td>\n",
       "      <td>-0.478759</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-1.894340</td>\n",
       "      <td>0.659782</td>\n",
       "      <td>92.81</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>PG1_3mut_1926344</td>\n",
       "      <td>bsPG-1.14</td>\n",
       "      <td>RGGRLCTARRRFCVHVGR</td>\n",
       "      <td>------TA------H---</td>\n",
       "      <td>-0.22</td>\n",
       "      <td>5.835569</td>\n",
       "      <td>-0.25-0</td>\n",
       "      <td>16</td>\n",
       "      <td>0.638402</td>\n",
       "      <td>1.009214</td>\n",
       "      <td>-1.718951</td>\n",
       "      <td>-0.199801</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-2.905787</td>\n",
       "      <td>0.783185</td>\n",
       "      <td>87.66</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>PG1_3mut_70973</td>\n",
       "      <td>bsPG-1.15</td>\n",
       "      <td>RGGRLCDCRRRFCNMVGR</td>\n",
       "      <td>------D------NM---</td>\n",
       "      <td>-0.15</td>\n",
       "      <td>4.654002</td>\n",
       "      <td>-0.25-0</td>\n",
       "      <td>&gt;64</td>\n",
       "      <td>-0.097466</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-8.957634</td>\n",
       "      <td>0.108545</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-7.735706</td>\n",
       "      <td>0.545411</td>\n",
       "      <td>68.43</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>PG1_3mut_969074</td>\n",
       "      <td>bsPG-1.16</td>\n",
       "      <td>RGGRLQYCRRRFDTCVGR</td>\n",
       "      <td>-----Q------DT----</td>\n",
       "      <td>-0.14</td>\n",
       "      <td>4.742406</td>\n",
       "      <td>-0.25-0</td>\n",
       "      <td>&gt;64</td>\n",
       "      <td>0.034113</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-15.366731</td>\n",
       "      <td>-1.027213</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-2.670889</td>\n",
       "      <td>0.618732</td>\n",
       "      <td>91.49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>PG1_3mut_624778</td>\n",
       "      <td>bsPG-1.17</td>\n",
       "      <td>RGGRLCTCRRRFCASVGR</td>\n",
       "      <td>------T------AS---</td>\n",
       "      <td>-0.14</td>\n",
       "      <td>5.653751</td>\n",
       "      <td>-0.25-0</td>\n",
       "      <td>32</td>\n",
       "      <td>0.487329</td>\n",
       "      <td>1.192973</td>\n",
       "      <td>0.506438</td>\n",
       "      <td>-0.073250</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-2.594505</td>\n",
       "      <td>0.872689</td>\n",
       "      <td>98.31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>PG1_3mut_1114717</td>\n",
       "      <td>bsPG-1.18</td>\n",
       "      <td>RGGRLRYCRRRFRVSVGR</td>\n",
       "      <td>-----R------R-S---</td>\n",
       "      <td>-0.13</td>\n",
       "      <td>7.833044</td>\n",
       "      <td>-0.25-0</td>\n",
       "      <td>16</td>\n",
       "      <td>1.525341</td>\n",
       "      <td>1.387487</td>\n",
       "      <td>0.554981</td>\n",
       "      <td>-0.397154</td>\n",
       "      <td>100.0</td>\n",
       "      <td>3.938261</td>\n",
       "      <td>0.569385</td>\n",
       "      <td>95.43</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>PG1_3mut_902003</td>\n",
       "      <td>bsPG-1.19</td>\n",
       "      <td>RGGRLEYCRRRFCVNTGR</td>\n",
       "      <td>-----E--------NT--</td>\n",
       "      <td>-0.13</td>\n",
       "      <td>4.742552</td>\n",
       "      <td>-0.25-0</td>\n",
       "      <td>&gt;64</td>\n",
       "      <td>1.281676</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-10.515178</td>\n",
       "      <td>-0.187527</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-1.731125</td>\n",
       "      <td>0.642734</td>\n",
       "      <td>80.77</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>PG1_3mut_484549</td>\n",
       "      <td>bsPG-1.20</td>\n",
       "      <td>RGGRLTYCRRRFCADVGR</td>\n",
       "      <td>-----T-------AD---</td>\n",
       "      <td>-0.10</td>\n",
       "      <td>4.742406</td>\n",
       "      <td>-0.25-0</td>\n",
       "      <td>64</td>\n",
       "      <td>0.248538</td>\n",
       "      <td>1.201573</td>\n",
       "      <td>-15.554958</td>\n",
       "      <td>-0.281757</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-5.932812</td>\n",
       "      <td>0.547104</td>\n",
       "      <td>95.17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>PG1_3mut_789807</td>\n",
       "      <td>bsPG-1.21</td>\n",
       "      <td>RGGRLCYRRRRTAVCVGR</td>\n",
       "      <td>-------R---TA-----</td>\n",
       "      <td>-0.10</td>\n",
       "      <td>6.742145</td>\n",
       "      <td>CP</td>\n",
       "      <td>8</td>\n",
       "      <td>0.404483</td>\n",
       "      <td>0.509991</td>\n",
       "      <td>-9.616057</td>\n",
       "      <td>-0.224147</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-7.072010</td>\n",
       "      <td>0.671800</td>\n",
       "      <td>77.76</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>PG1_3mut_1838343</td>\n",
       "      <td>bsPG-1.22</td>\n",
       "      <td>RGGRLCYCRRRFCTPVTR</td>\n",
       "      <td>-------------TP-T-</td>\n",
       "      <td>-0.03</td>\n",
       "      <td>5.651245</td>\n",
       "      <td>-0.25-0</td>\n",
       "      <td>64</td>\n",
       "      <td>0.389864</td>\n",
       "      <td>1.397093</td>\n",
       "      <td>-6.641514</td>\n",
       "      <td>-0.214472</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-2.631250</td>\n",
       "      <td>1.082659</td>\n",
       "      <td>84.77</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>PG1_3mut_945905</td>\n",
       "      <td>bsPG-1.23</td>\n",
       "      <td>RTGRLCYCRRRDCTCVGR</td>\n",
       "      <td>-T---------D-T----</td>\n",
       "      <td>-0.02</td>\n",
       "      <td>4.560587</td>\n",
       "      <td>CP</td>\n",
       "      <td>32</td>\n",
       "      <td>0.053606</td>\n",
       "      <td>0.234365</td>\n",
       "      <td>-8.263932</td>\n",
       "      <td>-0.334165</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-2.657715</td>\n",
       "      <td>0.792874</td>\n",
       "      <td>50.85</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>PG1_3mut_26231</td>\n",
       "      <td>bsPG-1.24</td>\n",
       "      <td>RGGRLHYRRRRFCVRVGR</td>\n",
       "      <td>-----H-R------R---</td>\n",
       "      <td>-0.01</td>\n",
       "      <td>7.923953</td>\n",
       "      <td>-0.25-0</td>\n",
       "      <td>8</td>\n",
       "      <td>9.722222</td>\n",
       "      <td>1.890856</td>\n",
       "      <td>-9.540926</td>\n",
       "      <td>0.028367</td>\n",
       "      <td>100.0</td>\n",
       "      <td>2.111896</td>\n",
       "      <td>0.786185</td>\n",
       "      <td>91.49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>PG1_3mut_485919</td>\n",
       "      <td>bsPG-1.25</td>\n",
       "      <td>RGGRPCYCRRRFCGTVGR</td>\n",
       "      <td>----P--------GT---</td>\n",
       "      <td>0.03</td>\n",
       "      <td>5.651245</td>\n",
       "      <td>0-0.5</td>\n",
       "      <td>&gt;64</td>\n",
       "      <td>-0.009747</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.994274</td>\n",
       "      <td>-0.381789</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-0.263622</td>\n",
       "      <td>0.893200</td>\n",
       "      <td>95.69</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>PG1_3mut_437553</td>\n",
       "      <td>bsPG-1.26</td>\n",
       "      <td>RGGRLCDCRRRTCRCVGR</td>\n",
       "      <td>------D----T-R----</td>\n",
       "      <td>0.04</td>\n",
       "      <td>5.563083</td>\n",
       "      <td>CP</td>\n",
       "      <td>&gt;64</td>\n",
       "      <td>0.194932</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-6.507852</td>\n",
       "      <td>-0.163775</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-3.961687</td>\n",
       "      <td>0.685757</td>\n",
       "      <td>0.49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>PG1_3mut_840730</td>\n",
       "      <td>bsPG-1.27</td>\n",
       "      <td>RGGRDIYCRRRFCVTVGR</td>\n",
       "      <td>----DI--------T---</td>\n",
       "      <td>0.10</td>\n",
       "      <td>4.742406</td>\n",
       "      <td>0-0.5</td>\n",
       "      <td>8</td>\n",
       "      <td>0.662768</td>\n",
       "      <td>0.724452</td>\n",
       "      <td>-9.883299</td>\n",
       "      <td>0.118627</td>\n",
       "      <td>100.0</td>\n",
       "      <td>0.154079</td>\n",
       "      <td>0.956514</td>\n",
       "      <td>7.18</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>PG1_3mut_1102178</td>\n",
       "      <td>bsPG-1.28</td>\n",
       "      <td>RGGRLCYCRRRTCVDTGR</td>\n",
       "      <td>-----------T--DT--</td>\n",
       "      <td>0.17</td>\n",
       "      <td>4.651497</td>\n",
       "      <td>0-0.5</td>\n",
       "      <td>32</td>\n",
       "      <td>0.160819</td>\n",
       "      <td>0.711487</td>\n",
       "      <td>-11.357187</td>\n",
       "      <td>-0.072031</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-2.748653</td>\n",
       "      <td>0.853561</td>\n",
       "      <td>88.73</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>PG1_3mut_939583</td>\n",
       "      <td>bsPG-1.29</td>\n",
       "      <td>RGGRLCYSRRRFCKTVGR</td>\n",
       "      <td>-------S-----KT---</td>\n",
       "      <td>0.33</td>\n",
       "      <td>6.741654</td>\n",
       "      <td>0-0.5</td>\n",
       "      <td>64</td>\n",
       "      <td>0.560429</td>\n",
       "      <td>1.554700</td>\n",
       "      <td>1.551211</td>\n",
       "      <td>0.101647</td>\n",
       "      <td>100.0</td>\n",
       "      <td>2.327160</td>\n",
       "      <td>0.850030</td>\n",
       "      <td>97.39</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>PG1_3mut_439660</td>\n",
       "      <td>bsPG-1.30</td>\n",
       "      <td>RGGRLAYCRRRFCVAVAR</td>\n",
       "      <td>-----A--------A-A-</td>\n",
       "      <td>0.36</td>\n",
       "      <td>5.742155</td>\n",
       "      <td>0-0.5</td>\n",
       "      <td>2</td>\n",
       "      <td>1.076998</td>\n",
       "      <td>0.333245</td>\n",
       "      <td>-1.721362</td>\n",
       "      <td>0.664414</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-5.520584</td>\n",
       "      <td>0.812613</td>\n",
       "      <td>96.13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>PG1_3mut_341044</td>\n",
       "      <td>bsPG-1.31</td>\n",
       "      <td>RGGRLCYGRRRFCVNQGR</td>\n",
       "      <td>-------G------NQ--</td>\n",
       "      <td>0.37</td>\n",
       "      <td>5.742155</td>\n",
       "      <td>0-0.5</td>\n",
       "      <td>32</td>\n",
       "      <td>2.061404</td>\n",
       "      <td>1.819313</td>\n",
       "      <td>-1.774004</td>\n",
       "      <td>-0.034093</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-1.552879</td>\n",
       "      <td>0.679024</td>\n",
       "      <td>93.17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>PG1_3mut_1166571</td>\n",
       "      <td>bsPG-1.32</td>\n",
       "      <td>RGGRSCHCRRRFCVIVGR</td>\n",
       "      <td>----S-H-------I---</td>\n",
       "      <td>0.37</td>\n",
       "      <td>5.744660</td>\n",
       "      <td>0-0.5</td>\n",
       "      <td>4</td>\n",
       "      <td>15.696881</td>\n",
       "      <td>1.797873</td>\n",
       "      <td>-2.584238</td>\n",
       "      <td>0.779326</td>\n",
       "      <td>100.0</td>\n",
       "      <td>0.003508</td>\n",
       "      <td>1.037089</td>\n",
       "      <td>67.97</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>PG1_3mut_1080117</td>\n",
       "      <td>bsPG-1.33</td>\n",
       "      <td>RGRRACYCRRRFCVHVGR</td>\n",
       "      <td>--R-A---------H---</td>\n",
       "      <td>0.38</td>\n",
       "      <td>6.742145</td>\n",
       "      <td>0-0.5</td>\n",
       "      <td>4</td>\n",
       "      <td>3.664717</td>\n",
       "      <td>1.166100</td>\n",
       "      <td>-0.897124</td>\n",
       "      <td>0.549621</td>\n",
       "      <td>100.0</td>\n",
       "      <td>2.614080</td>\n",
       "      <td>1.044137</td>\n",
       "      <td>84.35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>PG1_3mut_261082</td>\n",
       "      <td>bsPG-1.34</td>\n",
       "      <td>RGGRLSYCRRRFCVAGGR</td>\n",
       "      <td>-----S--------AG--</td>\n",
       "      <td>0.40</td>\n",
       "      <td>5.742155</td>\n",
       "      <td>0-0.5</td>\n",
       "      <td>16</td>\n",
       "      <td>0.321637</td>\n",
       "      <td>0.711487</td>\n",
       "      <td>0.575238</td>\n",
       "      <td>0.464198</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-0.155737</td>\n",
       "      <td>0.879992</td>\n",
       "      <td>96.83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>PG1_3mut_881885</td>\n",
       "      <td>bsPG-1.35</td>\n",
       "      <td>RGQRLCYCRRRFNVCTGR</td>\n",
       "      <td>--Q---------N--T--</td>\n",
       "      <td>0.41</td>\n",
       "      <td>5.651245</td>\n",
       "      <td>CP</td>\n",
       "      <td>32</td>\n",
       "      <td>0.399610</td>\n",
       "      <td>1.106786</td>\n",
       "      <td>-2.725072</td>\n",
       "      <td>-0.036225</td>\n",
       "      <td>100.0</td>\n",
       "      <td>4.564720</td>\n",
       "      <td>1.079400</td>\n",
       "      <td>64.24</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>PG1_3mut_702701</td>\n",
       "      <td>bsPG-1.36</td>\n",
       "      <td>RGGRLCYCRRRFCAVNGR</td>\n",
       "      <td>-------------AVN--</td>\n",
       "      <td>0.45</td>\n",
       "      <td>5.651245</td>\n",
       "      <td>0-0.5</td>\n",
       "      <td>64</td>\n",
       "      <td>0.116959</td>\n",
       "      <td>0.874214</td>\n",
       "      <td>-6.823680</td>\n",
       "      <td>0.701898</td>\n",
       "      <td>100.0</td>\n",
       "      <td>-4.328007</td>\n",
       "      <td>0.761783</td>\n",
       "      <td>85.33</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  ID        ID2            Sequence  RGGRLCYCRRRFCVCVGR  \\\n",
       "0   PG1_3mut_1739000   bsPG-1.1  RGGRLQYCRRRFCVARGR  -----Q--------AR--   \n",
       "1   PG1_3mut_1218602   bsPG-1.2  RGGRLQYCRRRGCVTVGR  -----Q-----G--T---   \n",
       "2   PG1_3mut_1896424   bsPG-1.3  RGGRLAYCRRRDCTCVGR  -----A-----D-T----   \n",
       "3   PG1_3mut_1275003   bsPG-1.4  RGGRLDYCRRRFCVTTGR  -----D--------TT--   \n",
       "4    PG1_3mut_257802   bsPG-1.5  RGGRLQACRRRFCVTVGR  -----QA-------T---   \n",
       "5   PG1_3mut_1210489   bsPG-1.6  RGGRLRYCRRRFDVSVGR  -----R------D-S---   \n",
       "6   PG1_3mut_1536372   bsPG-1.7  RGGRLCTARRRFCVRVGR  ------TA------R---   \n",
       "7   PG1_3mut_1147726   bsPG-1.8  RGGRLTYCRRRFCTAVGR  -----T-------TA---   \n",
       "8   PG1_3mut_1980555   bsPG-1.9  RGGRLTYCRRRDCVAVGR  -----T-----D--A---   \n",
       "9   PG1_3mut_1658870  bsPG-1.10  RGGRLAYCRRRFCVDTGR  -----A--------DT--   \n",
       "10  PG1_3mut_1618529  bsPG-1.11  RGGRLRYCRRRGTVCVGR  -----R-----GT-----   \n",
       "11   PG1_3mut_720432  bsPG-1.12  RGGRLQYCRRRFCARVGR  -----Q-------AR---   \n",
       "12  PG1_3mut_1247939  bsPG-1.13  RGGRLCYCRRRNCVTTGR  -----------N--TT--   \n",
       "13  PG1_3mut_1926344  bsPG-1.14  RGGRLCTARRRFCVHVGR  ------TA------H---   \n",
       "14    PG1_3mut_70973  bsPG-1.15  RGGRLCDCRRRFCNMVGR  ------D------NM---   \n",
       "15   PG1_3mut_969074  bsPG-1.16  RGGRLQYCRRRFDTCVGR  -----Q------DT----   \n",
       "16   PG1_3mut_624778  bsPG-1.17  RGGRLCTCRRRFCASVGR  ------T------AS---   \n",
       "17  PG1_3mut_1114717  bsPG-1.18  RGGRLRYCRRRFRVSVGR  -----R------R-S---   \n",
       "18   PG1_3mut_902003  bsPG-1.19  RGGRLEYCRRRFCVNTGR  -----E--------NT--   \n",
       "19   PG1_3mut_484549  bsPG-1.20  RGGRLTYCRRRFCADVGR  -----T-------AD---   \n",
       "20   PG1_3mut_789807  bsPG-1.21  RGGRLCYRRRRTAVCVGR  -------R---TA-----   \n",
       "21  PG1_3mut_1838343  bsPG-1.22  RGGRLCYCRRRFCTPVTR  -------------TP-T-   \n",
       "22   PG1_3mut_945905  bsPG-1.23  RTGRLCYCRRRDCTCVGR  -T---------D-T----   \n",
       "23    PG1_3mut_26231  bsPG-1.24  RGGRLHYRRRRFCVRVGR  -----H-R------R---   \n",
       "24   PG1_3mut_485919  bsPG-1.25  RGGRPCYCRRRFCGTVGR  ----P--------GT---   \n",
       "25   PG1_3mut_437553  bsPG-1.26  RGGRLCDCRRRTCRCVGR  ------D----T-R----   \n",
       "26   PG1_3mut_840730  bsPG-1.27  RGGRDIYCRRRFCVTVGR  ----DI--------T---   \n",
       "27  PG1_3mut_1102178  bsPG-1.28  RGGRLCYCRRRTCVDTGR  -----------T--DT--   \n",
       "28   PG1_3mut_939583  bsPG-1.29  RGGRLCYSRRRFCKTVGR  -------S-----KT---   \n",
       "29   PG1_3mut_439660  bsPG-1.30  RGGRLAYCRRRFCVAVAR  -----A--------A-A-   \n",
       "30   PG1_3mut_341044  bsPG-1.31  RGGRLCYGRRRFCVNQGR  -------G------NQ--   \n",
       "31  PG1_3mut_1166571  bsPG-1.32  RGGRSCHCRRRFCVIVGR  ----S-H-------I---   \n",
       "32  PG1_3mut_1080117  bsPG-1.33  RGRRACYCRRRFCVHVGR  --R-A---------H---   \n",
       "33   PG1_3mut_261082  bsPG-1.34  RGGRLSYCRRRFCVAGGR  -----S--------AG--   \n",
       "34   PG1_3mut_881885  bsPG-1.35  RGQRLCYCRRRFNVCTGR  --Q---------N--T--   \n",
       "35   PG1_3mut_702701  bsPG-1.36  RGGRLCYCRRRFCAVNGR  -------------AVN--   \n",
       "\n",
       "    Select    Charge    group MIC MH      %hemo  L10Select  \\\n",
       "0    -0.82  6.742145   <-0.25     16   0.428850   0.836425   \n",
       "1    -0.76  5.742155   <-0.25      4   0.341131   0.134981   \n",
       "2    -0.65  4.651497       CP    >64  -0.165692        NaN   \n",
       "3    -0.56  4.742406   <-0.25     64   0.165692   1.025482   \n",
       "4    -0.44  5.744660   <-0.25      8   0.721248   0.761174   \n",
       "5    -0.43  5.833305   <-0.25     16   0.735867   1.070920   \n",
       "6    -0.43  6.744650   <-0.25      2   0.906433   0.258366   \n",
       "7    -0.40  5.742155   <-0.25     16   0.043860  -0.153815   \n",
       "8    -0.29  4.742406   <-0.25     32   0.116959   0.573184   \n",
       "9    -0.28  4.742406   <-0.25     64   0.092593   0.772756   \n",
       "10   -0.28  6.742145       CP    >64   0.268031        NaN   \n",
       "11   -0.27  6.742145   <-0.25     16   0.141326   0.354341   \n",
       "12   -0.25  5.651245  -0.25-0    >64  -0.058480        NaN   \n",
       "13   -0.22  5.835569  -0.25-0     16   0.638402   1.009214   \n",
       "14   -0.15  4.654002  -0.25-0    >64  -0.097466        NaN   \n",
       "15   -0.14  4.742406  -0.25-0    >64   0.034113        NaN   \n",
       "16   -0.14  5.653751  -0.25-0     32   0.487329   1.192973   \n",
       "17   -0.13  7.833044  -0.25-0     16   1.525341   1.387487   \n",
       "18   -0.13  4.742552  -0.25-0    >64   1.281676        NaN   \n",
       "19   -0.10  4.742406  -0.25-0     64   0.248538   1.201573   \n",
       "20   -0.10  6.742145       CP      8   0.404483   0.509991   \n",
       "21   -0.03  5.651245  -0.25-0     64   0.389864   1.397093   \n",
       "22   -0.02  4.560587       CP     32   0.053606   0.234365   \n",
       "23   -0.01  7.923953  -0.25-0      8   9.722222   1.890856   \n",
       "24    0.03  5.651245    0-0.5    >64  -0.009747        NaN   \n",
       "25    0.04  5.563083       CP    >64   0.194932        NaN   \n",
       "26    0.10  4.742406    0-0.5      8   0.662768   0.724452   \n",
       "27    0.17  4.651497    0-0.5     32   0.160819   0.711487   \n",
       "28    0.33  6.741654    0-0.5     64   0.560429   1.554700   \n",
       "29    0.36  5.742155    0-0.5      2   1.076998   0.333245   \n",
       "30    0.37  5.742155    0-0.5     32   2.061404   1.819313   \n",
       "31    0.37  5.744660    0-0.5      4  15.696881   1.797873   \n",
       "32    0.38  6.742145    0-0.5      4   3.664717   1.166100   \n",
       "33    0.40  5.742155    0-0.5     16   0.321637   0.711487   \n",
       "34    0.41  5.651245       CP     32   0.399610   1.106786   \n",
       "35    0.45  5.651245    0-0.5     64   0.116959   0.874214   \n",
       "\n",
       "    hemolysis_pred_v01.2  selectivity_pred_v01.2  activity_pred_v01.2  \\\n",
       "0              -7.675692               -0.987362                100.0   \n",
       "1             -14.298704               -1.215473                100.0   \n",
       "2             -14.385390               -1.305665                100.0   \n",
       "3             -19.078636               -0.686002                100.0   \n",
       "4              -3.053714               -0.247956                100.0   \n",
       "5             -13.542688               -0.695068                100.0   \n",
       "6              -3.345207               -0.416047                100.0   \n",
       "7             -12.337255               -0.839417                100.0   \n",
       "8             -21.809423               -0.699560                100.0   \n",
       "9              -9.999792               -0.342884                100.0   \n",
       "10             -5.660681               -0.654460                100.0   \n",
       "11             -2.912143               -0.348571                100.0   \n",
       "12            -14.079775               -0.478759                100.0   \n",
       "13             -1.718951               -0.199801                100.0   \n",
       "14             -8.957634                0.108545                100.0   \n",
       "15            -15.366731               -1.027213                100.0   \n",
       "16              0.506438               -0.073250                100.0   \n",
       "17              0.554981               -0.397154                100.0   \n",
       "18            -10.515178               -0.187527                100.0   \n",
       "19            -15.554958               -0.281757                100.0   \n",
       "20             -9.616057               -0.224147                100.0   \n",
       "21             -6.641514               -0.214472                100.0   \n",
       "22             -8.263932               -0.334165                100.0   \n",
       "23             -9.540926                0.028367                100.0   \n",
       "24             -0.994274               -0.381789                100.0   \n",
       "25             -6.507852               -0.163775                100.0   \n",
       "26             -9.883299                0.118627                100.0   \n",
       "27            -11.357187               -0.072031                100.0   \n",
       "28              1.551211                0.101647                100.0   \n",
       "29             -1.721362                0.664414                100.0   \n",
       "30             -1.774004               -0.034093                100.0   \n",
       "31             -2.584238                0.779326                100.0   \n",
       "32             -0.897124                0.549621                100.0   \n",
       "33              0.575238                0.464198                100.0   \n",
       "34             -2.725072               -0.036225                100.0   \n",
       "35             -6.823680                0.701898                100.0   \n",
       "\n",
       "    hemolysis_pred_v02  selectivity_pred_v02  activity_pred_v02  \n",
       "0            -4.054558              0.448899              92.34  \n",
       "1            -6.763184              0.368799              88.34  \n",
       "2            -8.310135              0.390410              95.37  \n",
       "3            -2.867607              0.552553              90.87  \n",
       "4            -3.139629              0.636513              94.94  \n",
       "5            -2.812324              0.449201              90.10  \n",
       "6            -3.347273              0.609653              94.73  \n",
       "7            -1.988224              0.644135              98.64  \n",
       "8           -13.069695              0.206361              93.60  \n",
       "9            -5.106495              0.504043              93.70  \n",
       "10           -1.498447              0.631221              83.26  \n",
       "11           -4.230751              0.461692              90.30  \n",
       "12           -1.894340              0.659782              92.81  \n",
       "13           -2.905787              0.783185              87.66  \n",
       "14           -7.735706              0.545411              68.43  \n",
       "15           -2.670889              0.618732              91.49  \n",
       "16           -2.594505              0.872689              98.31  \n",
       "17            3.938261              0.569385              95.43  \n",
       "18           -1.731125              0.642734              80.77  \n",
       "19           -5.932812              0.547104              95.17  \n",
       "20           -7.072010              0.671800              77.76  \n",
       "21           -2.631250              1.082659              84.77  \n",
       "22           -2.657715              0.792874              50.85  \n",
       "23            2.111896              0.786185              91.49  \n",
       "24           -0.263622              0.893200              95.69  \n",
       "25           -3.961687              0.685757               0.49  \n",
       "26            0.154079              0.956514               7.18  \n",
       "27           -2.748653              0.853561              88.73  \n",
       "28            2.327160              0.850030              97.39  \n",
       "29           -5.520584              0.812613              96.13  \n",
       "30           -1.552879              0.679024              93.17  \n",
       "31            0.003508              1.037089              67.97  \n",
       "32            2.614080              1.044137              84.35  \n",
       "33           -0.155737              0.879992              96.83  \n",
       "34            4.564720              1.079400              64.24  \n",
       "35           -4.328007              0.761783              85.33  "
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meta_data['selectivity_pred_v02'] = pred\n",
    "meta_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "meta_data.to_csv('results/v01.2/predictions_validation_set_v01.2.csv', index=False)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "ML",
   "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": "3.8.10"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "f1569419c614e8688ff5ac73ded8c36a7b1a231c09910336a299b563fe1d2720"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}