01.add_megadataset_and_split_on_train_test_sets.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "1f9864a5",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from scipy import stats\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "from   pathlib import Path\n",
    "import subprocess\n",
    "import json\n",
    "import networkx as nx\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.metrics import mean_absolute_error"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3f63941f",
   "metadata": {},
   "source": [
    "### Function to calculate average number of sequences within 30% identity clusters "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b83e4282",
   "metadata": {},
   "outputs": [],
   "source": [
    "def count_average_number_of_proteins(df, n_select = None):\n",
    "    used = set()\n",
    "    Path(\"./temp/\").mkdir(exist_ok=True)\n",
    "    with open(f\"./temp/all_sequences_.fasta\",'w') as fo:\n",
    "        for d_ in df.iloc():\n",
    "            if d_['pdb_id'] in used:\n",
    "                continue\n",
    "            fo.write(f\">{d_['pdb_id']}\\n{d_['wt_seq']}\\n\")\n",
    "            used.add(d_['pdb_id'])\n",
    "    used = set()\n",
    "    subprocess.call(f\"makeblastdb -in all_sequences_.fasta -dbtype prot\", shell=True, cwd=\"./temp/\")\n",
    "    subprocess.call(f\"blastp      -db all_sequences_.fasta -query  all_sequences_.fasta -outfmt 6 -out hits_.tsv -num_threads 4\", shell=True, cwd=\"./temp/\")\n",
    "    hit_data    = pd.read_csv(f\"./temp/hits_.tsv\", delimiter='\\t', header=None)\n",
    "    hit_data    = hit_data[(hit_data.iloc[:,2]>30.0) & (hit_data.iloc[:,-2]<0.05)]\n",
    "    G = nx.Graph()\n",
    "    for h in hit_data.iloc():\n",
    "        e = [(h.iloc()[0],h.iloc()[1])]\n",
    "        G.add_edges_from(e)\n",
    "    subcomponents = list(nx.connected_components(G))\n",
    "    n_length = []\n",
    "    output   = []\n",
    "    for s in subcomponents:\n",
    "        df_ = df[df[\"pdb_id\"].isin(s)]\n",
    "        n_length.append(df_.shape[0])\n",
    "    print(f\"Average number of proteins: {np.average(n_length)}\")\n",
    "    \n",
    "    if n_select is None:\n",
    "        return np.average(n_length)\n",
    "    \n",
    "    sub_df = []\n",
    "    for s in subcomponents:\n",
    "        df_ = df[df[\"pdb_id\"].isin(s)]\n",
    "        df_ = df_.sample(n_select)\n",
    "        sub_df.append(df_)\n",
    "    df_concat = pd.concat(sub_df,  ignore_index=True).reset_index(drop=True)\n",
    "    df_concat.drop(columns=['reverse'], inplace=True)\n",
    "    return df_concat"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "693ce28e",
   "metadata": {},
   "source": [
    "### Function to filter out sequences from the training set that share more than 30% identity with the test sequences.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "b92b3ce9",
   "metadata": {},
   "outputs": [],
   "source": [
    "def homology_filter(df_train, \n",
    "                                 df_test):\n",
    "    used = set()\n",
    "    name = \"test_\"\n",
    "    subprocess.call(\"rm -r temp\",shell=True)\n",
    "    Path(\"./temp/\").mkdir(exist_ok=True)\n",
    "    with open(f\"./temp/all_sequences_{name}.fasta\",'w') as fo:\n",
    "        for d_ in df_train.iloc():\n",
    "            if d_['pdb_id'] in used:\n",
    "                continue\n",
    "            fo.write(f\">{d_['pdb_id']}\\n{d_['wt_seq']}\\n\")\n",
    "            used.add(d_['pdb_id'])\n",
    "    used = set()\n",
    "    with open(f\"./temp/test_sequences_{name}.fasta\",'w') as fo_test:\n",
    "        for d_ in df_test.iloc():\n",
    "            if d_['pdb_id'] in used:\n",
    "                continue\n",
    "            fo_test.write(f\">{d_['pdb_id']}\\n{d_['wt_seq']}\\n\")\n",
    "            used.add(d_['pdb_id'])\n",
    "            \n",
    "    subprocess.call(f\"makeblastdb -in all_sequences_{name}.fasta -dbtype prot\", shell=True, cwd=\"./temp/\")\n",
    "    subprocess.call(f\"blastp      -db all_sequences_{name}.fasta -query  test_sequences_{name}.fasta -outfmt 6 -out hits_{name}.tsv -num_threads 4\", shell=True, cwd=\"./temp/\")\n",
    "    \n",
    "    hit_data    = pd.read_csv(f\"./temp/hits_{name}.tsv\", delimiter='\\t', header=None)\n",
    "    hit_data    = hit_data[(hit_data.iloc[:,2]>30.0) & (hit_data.iloc[:,-2]<0.05)]\n",
    "    \n",
    "    G = nx.Graph()\n",
    "    for h in hit_data.iloc():\n",
    "        e = [(h.iloc()[0],h.iloc()[1])]\n",
    "        G.add_edges_from(e)\n",
    "    subcomponents = list(nx.connected_components(G))    \n",
    "    all_skip_nodes = set()\n",
    "    for s in subcomponents:\n",
    "        all_skip_nodes|=set(s)\n",
    "    n_before = df_train.shape\n",
    "    df_train = df_train[~df_train[\"pdb_id\"].isin(all_skip_nodes)]\n",
    "    n_after  = df_train.shape\n",
    "    \n",
    "    print(\"Size before filtering:\",  n_before)\n",
    "    print(\"Size after filtering:\",   n_after)\n",
    "    \n",
    "    return df_train\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5367a8be",
   "metadata": {},
   "source": [
    "### Make balanced mix with megadataset\n",
    "\n",
    "1. Calculate average number of proteins (N) within 30% identity cluster within PROSTATA dataset without megadataset\n",
    "2. Add proteins N proteins from megadataset for each cluster center to the PROSTATA dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "f2db2850",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:45:59\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_.fasta\n",
      "New DB title:  all_sequences_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 560 sequences in 0.00458407 seconds.\n",
      "\n",
      "\n",
      "Average number of proteins: 70.734375\n",
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:46:04\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 560 sequences in 0.00410914 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (9054, 7)\n",
      "Size after filtering: (8428, 7)\n",
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:46:21\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 388 sequences in 0.00331092 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (735716, 8)\n",
      "Size after filtering: (640716, 8)\n",
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:46:36\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_.fasta\n",
      "New DB title:  all_sequences_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 340 sequences in 0.00258708 seconds.\n",
      "\n",
      "\n",
      "Average number of proteins: 8542.88\n"
     ]
    }
   ],
   "source": [
    "df_old       = pd.read_csv(\"./DATASETS/new_ds_for_s669.csv\")\n",
    "N            = count_average_number_of_proteins(df_old)\n",
    "s669         = pd.read_csv(\"./DATASETS/s669.csv\")\n",
    "df_old       = homology_filter(df_old,  s669)\n",
    "df_mega      = pd.read_csv(\"./DATA/megadataset/train.csv\")\n",
    "df_mega      = homology_filter(df_mega, s669)\n",
    "df_mega      = count_average_number_of_proteins(df_mega, n_select = int(N))\n",
    "mix_dataset  = pd.concat([df_old, df_mega], ignore_index=True).reset_index(drop=True)\n",
    "datasets     = {}\n",
    "df_old.to_csv(     \"./DATASETS/prostata_ds_no_megadataset_for_s669.csv\"    )\n",
    "mix_dataset.to_csv(\"./DATASETS/prostata_ds_for_s669.csv\"                   )\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ec652d01",
   "metadata": {},
   "source": [
    "### Prepare test set of de novo and natural miniproteins from the megadataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "5e9f6508",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:52:01\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 388 sequences in 0.00287604 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (735716, 8)\n",
      "Size after filtering: (493284, 8)\n",
      "(493284, 8)\n"
     ]
    },
    {
     "ename": "NameError",
     "evalue": "name 'natural_pdb_ids' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "Cell \u001b[0;32mIn[6], line 16\u001b[0m\n\u001b[1;32m     13\u001b[0m natural_pdb_list \u001b[38;5;241m=\u001b[39m [\u001b[38;5;124m'\u001b[39m\u001b[38;5;124m6YSE.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m5KPH.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1F0M.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2LGW.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2MA4.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m6M3N.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m6EWS.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m7BPM.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2L9R.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m3L1X.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2JT1.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1AOY.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2KFV.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2JWT.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m5JRT.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2L7F.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2LC2.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m5UP5.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2LQK.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2M8U.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2N88.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2RRU.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2K5P.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2MI6.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2JZ2.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2GP8.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2KVT.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2L7M.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2L6Q.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m3DKM.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2MC5.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2WNM.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1K1V.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1R69.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2KWH.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2M2L.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2K1B.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2B88.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2LHR.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m5LXJ.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m6ACV.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2LO1.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2JVD.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m5Z2S.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2JY8.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2N7Y.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1MHN.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2L09.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1GJS.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2OCH.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1QP2.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1V1C.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m6EWT.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2D1U.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1YG0.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2L8D.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2RU9.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2MYX.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1WCL.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2M2J.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m3ZGK.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1QKH.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2MCK.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2M0C.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2LYR.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1LP1.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1PV0.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m6FVC.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2MPK.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2LVN.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1TG0.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m6EWU.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2LYQ.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2MXD.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1IFY.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2QFF.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2LT1.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m4G3O.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2KRU.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2JRR.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2MKY.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2JRO.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2MWA.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m3V1A.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2MCH.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2MRL.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2M8E.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2JN4.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m2K28.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m1ENH.pdb\u001b[39m\u001b[38;5;124m'\u001b[39m]\n\u001b[1;32m     15\u001b[0m df_de_novo \u001b[38;5;241m=\u001b[39m df_mega_new[df_mega_new[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mpdb_id\u001b[39m\u001b[38;5;124m\"\u001b[39m]\u001b[38;5;241m.\u001b[39misin(de_novo_pdb_list)]\n\u001b[0;32m---> 16\u001b[0m df_natural \u001b[38;5;241m=\u001b[39m df_mega_new[df_mega_new[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mpdb_id\u001b[39m\u001b[38;5;124m\"\u001b[39m]\u001b[38;5;241m.\u001b[39misin(\u001b[43mnatural_pdb_ids\u001b[49m)]\n\u001b[1;32m     18\u001b[0m df_de_novo[\u001b[38;5;241m~\u001b[39mdf_de_novo[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mreverse\u001b[39m\u001b[38;5;124m\"\u001b[39m]]\u001b[38;5;241m.\u001b[39mto_csv(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m./DATASETS/miniproteins_denovo_megadataset.csv\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n\u001b[1;32m     19\u001b[0m df_natural[\u001b[38;5;241m~\u001b[39mdf_natural[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mreverse\u001b[39m\u001b[38;5;124m\"\u001b[39m]]\u001b[38;5;241m.\u001b[39mto_csv(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m./DATASETS/miniproteins_natural_megadataset.csv\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n",
      "\u001b[0;31mNameError\u001b[0m: name 'natural_pdb_ids' is not defined"
     ]
    }
   ],
   "source": [
    "df_mega = pd.read_csv(\"DATA/megadataset/train.csv\")\n",
    "df_test = pd.read_csv(\"DATA/megadataset/test.csv\")\n",
    "df_val  = pd.read_csv(\"DATA/megadataset/val.csv\")\n",
    "df_all  = pd.concat([df_mega, df_test, df_val])\n",
    "dfs = []\n",
    "for d in [\"S2648\", \"S3421\", \"S3488\", \"s669\", \"ssym\", \"myoglobin\", \"p53\"]:\n",
    "    dfs.append(pd.read_csv(f\"./DATASETS/{d}.csv\"))\n",
    "dfs = pd.concat(dfs)\n",
    "df_mega_new = homology_filter(df_mega, dfs)\n",
    "print(df_mega_new.shape)\n",
    "\n",
    "de_novo_pdb_list = ['HEEH_KT_rd6_0200.pdb', 'EEHEE_rd2_0770.pdb', 'HEEH_rd4_0097.pdb', 'EEHEE_rd3_0144.pdb', 'EEHEE_rd4_0418.pdb', 'HEEH_rd4_0094.pdb', 'EHEE_rd2_0372.pdb', 'EHEE_rd4_0044.pdb', 'EEHEE_rd3_0657.pdb', 'EHEE_rd2_0191.pdb', 'EHEE_rd2_0751.pdb', 'EEHEE_rd3_1587.pdb', 'HEEH_KT_rd6_0793.pdb', 'HEEH_KT_rd6_0790.pdb', 'EHEE_rd4_0172.pdb', 'EEHEE_rd3_1810.pdb', 'EHEE_rd4_0463.pdb', 'EEHEE_rd3_1702.pdb', 'EEHEE_rd3_1817.pdb', 'EEHEE_rd3_1498.pdb', 'EHEE_rd2_0152.pdb', 'EHEE_rd3_0067.pdb', 'HEEH_KT_rd6_3632.pdb', 'HEEH_rd3_0223.pdb', 'EEHEE_rd3_0146.pdb', 'EHEE_rd4_0086.pdb', 'EHEE_rd3_0035.pdb', 'EHEE_rd4_0394.pdb', 'EHEE_rd4_0098.pdb', 'EHEE_rd4_0325.pdb', 'EHEE_rd3_0053.pdb', 'EHEE_rd2_0487.pdb', 'EHEE_rd1_0101.pdb', 'EHEE_rd4_0300.pdb', 'EEHEE_rd3_1558.pdb', 'EEHEE_rd4_0647.pdb', 'EEHEE_rd3_0094.pdb', 'EEHEE_rd4_0003.pdb', 'EHEE_rd2_0196.pdb', 'EEHEE_rd4_0424.pdb', 'HEEH_rd4_0349.pdb', 'HEEH_KT_rd6_1415.pdb', 'HEEH_KT_rd6_0746.pdb', 'EEHEE_rd3_0602.pdb', 'EHEE_rd4_0195.pdb', 'EEHEE_rd4_0481.pdb', 'EHEE_rd2_0601.pdb', 'EHEE_rd4_0864.pdb', 'EEHEE_rd3_1627.pdb', 'EHEE_rd4_0726.pdb', 'EHEE_rd2_1257.pdb', 'EEHEE_rd4_0256.pdb', 'EHEE_rd4_0840.pdb', 'EHEE_rd2_0628.pdb', 'EEHEE_rd3_1367.pdb', 'EEHEE_rd4_0363.pdb', 'EHEE_rd1_0407.pdb', 'EEHEE_rd3_0019.pdb', 'EEHEE_rd4_0469.pdb', 'HEEH_KT_rd6_0007.pdb', 'HEEH_KT_rd6_0280.pdb', 'EEHEE_rd4_0470.pdb', 'EHEE_rd4_0340.pdb', 'EEHEE_rd4_0794.pdb', 'EEHEE_rd3_1058.pdb']\n",
    "natural_pdb_list = ['6YSE.pdb', '5KPH.pdb', '1F0M.pdb', '2LGW.pdb', '2MA4.pdb', '6M3N.pdb', '6EWS.pdb', '7BPM.pdb', '2L9R.pdb', '3L1X.pdb', '2JT1.pdb', '1AOY.pdb', '2KFV.pdb', '2JWT.pdb', '5JRT.pdb', '2L7F.pdb', '2LC2.pdb', '5UP5.pdb', '2LQK.pdb', '2M8U.pdb', '2N88.pdb', '2RRU.pdb', '2K5P.pdb', '2MI6.pdb', '2JZ2.pdb', '2GP8.pdb', '2KVT.pdb', '2L7M.pdb', '2L6Q.pdb', '3DKM.pdb', '2MC5.pdb', '2WNM.pdb', '1K1V.pdb', '1R69.pdb', '2KWH.pdb', '2M2L.pdb', '2K1B.pdb', '2B88.pdb', '2LHR.pdb', '5LXJ.pdb', '6ACV.pdb', '2LO1.pdb', '2JVD.pdb', '5Z2S.pdb', '2JY8.pdb', '2N7Y.pdb', '1MHN.pdb', '2L09.pdb', '1GJS.pdb', '2OCH.pdb', '1QP2.pdb', '1V1C.pdb', '6EWT.pdb', '2D1U.pdb', '1YG0.pdb', '2L8D.pdb', '2RU9.pdb', '2MYX.pdb', '1WCL.pdb', '2M2J.pdb', '3ZGK.pdb', '1QKH.pdb', '2MCK.pdb', '2M0C.pdb', '2LYR.pdb', '1LP1.pdb', '1PV0.pdb', '6FVC.pdb', '2MPK.pdb', '2LVN.pdb', '1TG0.pdb', '6EWU.pdb', '2LYQ.pdb', '2MXD.pdb', '1IFY.pdb', '2QFF.pdb', '2LT1.pdb', '4G3O.pdb', '2KRU.pdb', '2JRR.pdb', '2MKY.pdb', '2JRO.pdb', '2MWA.pdb', '3V1A.pdb', '2MCH.pdb', '2MRL.pdb', '2M8E.pdb', '2JN4.pdb', '2K28.pdb', '1ENH.pdb']\n",
    "\n",
    "df_de_novo = df_mega_new[df_mega_new[\"pdb_id\"].isin(de_novo_pdb_list)]\n",
    "df_natural = df_mega_new[df_mega_new[\"pdb_id\"].isin(natural_pdb_ids)]\n",
    "\n",
    "df_de_novo[~df_de_novo[\"reverse\"]].to_csv(\"./DATASETS/miniproteins_denovo_megadataset.csv\")\n",
    "df_natural[~df_natural[\"reverse\"]].to_csv(\"./DATASETS/miniproteins_natural_megadataset.csv\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fdf8c595",
   "metadata": {},
   "source": [
    "###\n",
    "### Prepare train and test splits for experiments in the paper\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "49706cbf",
   "metadata": {},
   "outputs": [],
   "source": [
    "def reverse_dataset(df):\n",
    "    ### function to add inverse mutations\n",
    "    df = df.copy()\n",
    "    w,m = df[ \"wt_seq\"].copy(),df[\"mut_seq\"].copy()\n",
    "    df[\"mut_seq\"] =  w\n",
    "    df[\"wt_seq\"]  =  m\n",
    "    df[\"ddg\"]     *= -1\n",
    "    mut_ids = []\n",
    "    for d_ in df.iloc():\n",
    "        mut     = d_[\"mut_info\"]\n",
    "        mut_rev = mut[-1]+mut[1:-1]+mut[0]\n",
    "        mut_ids.append(mut_rev)\n",
    "    df[\"mut_info\"]= mut_ids\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "7310ff86",
   "metadata": {},
   "outputs": [],
   "source": [
    "datasets                     = {}\n",
    "datasets[\"prostata\"]         = pd.read_csv(\"DATASETS/prostata_ds_for_s669.csv\")\n",
    "datasets[\"hem_test\"]         = pd.read_csv(\"DATASETS/case_study_dimer_test_with_labels.csv\"      )\n",
    "datasets[\"oligo_test\"]       = pd.read_csv(\"DATASETS/case_study_gem_test_with_labels.csv\"        )\n",
    "datasets[\"s669_with_dssp\"]   = pd.read_csv(\"DATASETS/s669_with_dssp.csv\"                         )\n",
    "datasets[\"S2648\"]            = pd.read_csv(\"DATASETS/S2648.csv\"         )\n",
    "datasets[\"S2648_r\"]          = reverse_dataset(pd.read_csv(\"DATASETS/S2648.csv\"                 ))\n",
    "datasets[\"Q3421\"]            = pd.read_csv(\"DATASETS/S3421.csv\"         )\n",
    "datasets[\"Q3488\"]            = pd.read_csv(\"DATASETS/S3488.csv\"         )\n",
    "datasets[\"Vb\"]               = pd.read_csv(\"DATASETS/ACDC_varibench.csv\")\n",
    "datasets[\"Vb_r\"]             = reverse_dataset(datasets[\"Vb\"]               )\n",
    "datasets[\"ssym\"]             = pd.read_csv(\"DATASETS/ssym.csv\"          )\n",
    "datasets[\"ssym_r\"]           = pd.read_csv(\"DATASETS/ssym_r.csv\"        ) \n",
    "datasets[\"s669\"]             = pd.read_csv(\"DATASETS/s669.csv\"          )\n",
    "datasets[\"s669_r\"]           = pd.read_csv(\"DATASETS/s669_r.csv\"        )\n",
    "datasets[\"s669_dssp\"]        = pd.read_csv(\"DATASETS/s669_with_dssp.csv\"     )\n",
    "datasets[\"myoglobin\"]        = pd.read_csv(\"DATASETS/myoglobin.csv\"     )\n",
    "datasets[\"myoglobin_r\"]      = reverse_dataset(datasets[\"myoglobin\"])\n",
    "datasets[\"p53\"]              = pd.read_csv(\"DATASETS/p53.csv\"           )\n",
    "datasets[\"p53_r\"]            = reverse_dataset(datasets[\"p53\"])\n",
    "datasets[\"mini_nat\"]         = pd.read_csv(\"./DATASETS/miniproteins_natural_megadataset.csv\")           \n",
    "datasets[\"mini_nat_r\"]       = reverse_dataset(datasets[\"mini_nat\"])\n",
    "datasets[\"mini_denovo\"]     = pd.read_csv(\"./DATASETS/miniproteins_denovo_megadataset.csv\")    \n",
    "datasets[\"mini_denovo_r\"]   = reverse_dataset(datasets[\"mini_denovo\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "854c2941",
   "metadata": {},
   "source": [
    "### Remove sequences from the train set that shares similarity with test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "69aef857",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:53:33\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 163 sequences in 0.00183702 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (8036, 7)\n",
      "Size after filtering: (8016, 7)\n",
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:53:35\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 127 sequences in 0.00149798 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (3488, 7)\n",
      "Size after filtering: (3458, 7)\n",
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:53:36\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 127 sequences in 0.00146914 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (3488, 7)\n",
      "Size after filtering: (3488, 7)\n",
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:53:38\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 127 sequences in 0.00150585 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (3488, 7)\n",
      "Size after filtering: (3488, 7)\n",
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:53:39\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 163 sequences in 0.00178289 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (8036, 7)\n",
      "Size after filtering: (7888, 7)\n",
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:53:41\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 149 sequences in 0.0018661 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (3407, 7)\n",
      "Size after filtering: (1812, 7)\n",
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:53:42\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 745 sequences in 0.00491405 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (13678, 8)\n",
      "Size after filtering: (13678, 8)\n",
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:53:44\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 745 sequences in 0.0049541 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (13678, 8)\n",
      "Size after filtering: (13326, 8)\n",
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:53:46\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 745 sequences in 0.005023 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (13678, 8)\n",
      "Size after filtering: (13320, 8)\n",
      "\n",
      "\n",
      "Building a new DB, current time: 09/18/2023 12:53:47\n",
      "New DB name:   /Users/ivanisenko/projects/prostata/git_hub_prostata/PROSTATA/temp/all_sequences_test_.fasta\n",
      "New DB title:  all_sequences_test_.fasta\n",
      "Sequence type: Protein\n",
      "Keep MBits: T\n",
      "Maximum file size: 3000000000B\n",
      "Adding sequences from FASTA; added 149 sequences in 0.00169396 seconds.\n",
      "\n",
      "\n",
      "Size before filtering: (3407, 7)\n",
      "Size after filtering: (1812, 7)\n"
     ]
    }
   ],
   "source": [
    "experiments = [{\"train\":[\"S2648\",\"S2648_r\",\"Vb\",\"Vb_r\"], \"test\":[\"p53\",\"p53_r\", \n",
    "                                                                 \"mini_nat\",\n",
    "                                                                 \"mini_denovo\"]},\n",
    "               {\"train\":[\"Q3488\"],             \"test\":[\"ssym\",           \"ssym_r\"]},\n",
    "               {\"train\":[\"Q3488\"],             \"test\":[\"p53\",             \"p53_r\"]},\n",
    "               {\"train\":[\"Q3488\"],             \"test\":[\"myoglobin\", \"myoglobin_r\"]},\n",
    "               {\"train\":[\"S2648\",\"S2648_r\",\"Vb\",\"Vb_r\"], \"test\":[\"myoglobin\", \"myoglobin_r\"]},\n",
    "               {\"train\":[\"Q3421\"],             \"test\":[\"ssym\",           \"ssym_r\"]},\n",
    "               {\"train\":[\"prostata\"],          \"test\":[\"s669\",\"s669_r\",\"s669_dssp\"]},\n",
    "               {\"train\":[\"prostata\"],          \"test\":[\"hem_test\"]},\n",
    "               {\"train\":[\"prostata\"],          \"test\":[\"oligo_test\"]},\n",
    "               {\"train\":[\"Q3421\"],             \"test\":[\"ssym\", \"ssym_r\"]}]\n",
    "    \n",
    "Path(\"./PROSTATA_EXPERIMENTS/\").mkdir(exist_ok=True)\n",
    "meta_info = []\n",
    "for experiment in experiments:\n",
    "    train_ids, test_ids = experiment[\"train\"], experiment[\"test\"]\n",
    "    train_set      = pd.concat([datasets[t] for t in train_ids]).reset_index(drop=True)\n",
    "    train_set_name = \".\".join(train_ids)\n",
    "    test_set_name  = \".\".join(test_ids)\n",
    "    output_path    = \"./PROSTATA_EXPERIMENTS/train_\"+train_set_name+\"_test_\"+test_set_name\n",
    "    Path(output_path).mkdir(exist_ok=True)           \n",
    "    train_set_filtered = homology_filter(train_set,                                              \n",
    "                                                datasets[test_ids[0]])\n",
    "    \n",
    "    meta_info_ = {\"train_set_size_unfiltered\":train_set.shape[0],\n",
    "                  \"train_set_size_filtered\":  train_set_filtered.shape[0],\n",
    "                  \"test_set_size\":            datasets[test_ids[0]].shape[0],\n",
    "                  \"train_set_ids\":            train_ids,\n",
    "                  \"test_set_ids\" :            test_ids}\n",
    "    meta_info.append(meta_info_)\n",
    "    train_set_filtered.to_csv(output_path+\"/train.csv\")\n",
    "    for test_id in test_ids:\n",
    "        datasets[test_id].to_csv(output_path+f\"/test_{test_id}.csv\")\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "b2c11059",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "S2648/S2648_r & p53 & 8016 / 8036 & 42\n",
      "Q3488 & ssym & 3458 / 3488 & 342\n",
      "Q3488 & p53 & 3488 / 3488 & 42\n",
      "Q3488 & myoglobin & 3488 / 3488 & 134\n",
      "S2648/S2648_r & myoglobin & 7888 / 8036 & 134\n",
      "Q3421 & ssym & 1812 / 3407 & 342\n",
      "prostata & s669 & 13678 / 13678 & 669\n",
      "prostata & hem_test & 13326 / 13678 & 294\n",
      "prostata & oligo_test & 13320 / 13678 & 382\n",
      "Q3421 & ssym & 1812 / 3407 & 342\n"
     ]
    }
   ],
   "source": [
    "for m in meta_info:\n",
    "    print(\"/\".join(m[\"train_set_ids\"][:2]), \"&\", m[\"test_set_ids\"][0], \"&\", m[\"train_set_size_filtered\"],\"/\",m[\"train_set_size_unfiltered\"], \"&\", m[\"test_set_size\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "87f2cd4f",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}