evaluation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from evaluation import *\n",
    "import torch\n",
    "import utils\n",
    "from esm import pretrained\n",
    "%load_ext autoreload\n",
    "import seaborn as sns\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "eval_esm=pd.read_csv('/Users/guillaumebelissent/Docs/EPFL/MA/ML/model_evaluation_esm.csv')\n",
    "eval_esm['Name']=eval_esm['Name'].apply(lambda x: 'esm_8M' if x==0 else 'esm_35M')\n",
    "eval=pd.concat([eval_esm,pd.read_csv('/Users/guillaumebelissent/Docs/EPFL/MA/ML/ml-project-2-byte-by-byte/model_evaluation_HM.csv')])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def blosum62_matrix():\n",
    "    \"\"\"\n",
    "    Defines and normalizes by column the BLOSUM62 matrix from the blosum62.csv file and returns it as a PyTorch tensor.\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    None \n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    torch.Tensor\n",
    "        BLOSUM62 matrix as a tensor.\n",
    "    \"\"\"\n",
    "    blosum62_df = pd.read_csv('blosum62.csv', delimiter=\";\", header = None)\n",
    "    blosum62_df.columns=['A', 'R', 'N', 'D', 'C', 'E', 'Q', 'G', 'H', 'I', 'L', 'K', 'M', 'F', 'P', 'S', 'T', 'W', 'Y', 'V']\n",
    "    blosum62_df.index=['A', 'R', 'N', 'D', 'C', 'E', 'Q', 'G', 'H', 'I', 'L', 'K', 'M', 'F', 'P', 'S', 'T', 'W', 'Y', 'V']\n",
    "    # Normalization by column\n",
    "    for column in blosum62_df.columns: \n",
    "        blosum62_df[column] = (blosum62_df[column] - blosum62_df[column].min()) / (blosum62_df[column].max() - blosum62_df[column].min())     \n",
    "    return blosum62_df\n",
    "blosum62=blosum62_matrix()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/6_/6g10b43x17ngwmk4l9msb2mc0000gn/T/ipykernel_60370/2076906166.py:2: FutureWarning: \n",
      "\n",
      "Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect.\n",
      "\n",
      "  sns.barplot(eval[eval['Name']!='BERT_depth4_embed512_steps2000_v2'],x='Name',y='Correct',\n",
      "/var/folders/6_/6g10b43x17ngwmk4l9msb2mc0000gn/T/ipykernel_60370/2076906166.py:13: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  filtered_eval['Blosum62'] = filtered_eval.apply(\n",
      "/var/folders/6_/6g10b43x17ngwmk4l9msb2mc0000gn/T/ipykernel_60370/2076906166.py:16: FutureWarning: \n",
      "\n",
      "Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect.\n",
      "\n",
      "  sns.barplot(filtered_eval,x='Name',y='Blosum62', palette='muted',\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig,(ax1,ax2) = plt.subplots(1,2, figsize=(10,5))\n",
    "sns.barplot(eval[eval['Name']!='BERT_depth4_embed512_steps2000_v2'],x='Name',y='Correct',\n",
    "            order=['BERT_depth4_embed256_steps2000','BERT_depth4_embed256_head16_steps1000','BERTMSA_depth4_embed256_head8_steps2000','esm_8M','esm_35M'],\n",
    "            ax=ax1, palette='muted')\n",
    "ax1.set_xticks(np.arange(5))\n",
    "ax1.set_xticklabels(labels=['Baseline','Big Head','MSA','ESM 8M','ESM 35M'], rotation=45, ha='right', size=15)\n",
    "ax1.set_xlabel('')\n",
    "ax1.set_ylabel('Accuracy',size=15)\n",
    "ax1.axhline(y=1/20, color='red', linestyle='--', label='Chance')\n",
    "\n",
    "aa = [\"A\", \"R\", \"N\", \"D\", \"C\", \"Q\", \"E\", \"G\", \"H\", \"I\",\"L\", \"K\", \"M\", \"F\", \"P\", \"S\", \"T\", \"W\", \"Y\", \"V\"]\n",
    "filtered_eval = eval[(eval['Prediction'].isin(aa)) & (eval['Label'].isin(aa))]\n",
    "filtered_eval['Blosum62'] = filtered_eval.apply(\n",
    "    lambda row: blosum62.loc[row['Prediction'], row['Label']], axis=1\n",
    ")\n",
    "sns.barplot(filtered_eval,x='Name',y='Blosum62', palette='muted',\n",
    "            order=['BERT_depth4_embed256_steps2000','BERT_depth4_embed256_head16_steps1000','BERTMSA_depth4_embed256_head8_steps2000','esm_8M','esm_35M'],ax=ax2)\n",
    "ax2.set_xticks(np.arange(5))\n",
    "ax2.set_xticklabels(labels=['Baseline','Big Head','MSA','ESM 8M','ESM 35M'], rotation=45, ha='right', size=15)\n",
    "ax2.set_xlabel('', size=15)\n",
    "ax2.set_ylabel('Blosum62 Score',size=15)\n",
    "ax2.axhline(y=blosum62.mean(axis=None), color='red', linestyle='--', label='Chance')\n",
    "plt.tight_layout()\n",
    "plt.show()\n",
    "\n"
   ]
  }
 ],
 "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": "3.12.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}