06 - Feature generation - Structure features.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "14cc9e24",
   "metadata": {},
   "source": [
    "# Next improvement attempts\n",
    "### _Introduce structural information_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "011825cf",
   "metadata": {},
   "outputs": [],
   "source": [
    "# temporary -- all code needed up to this point\n",
    "import pandas as pd\n",
    "import random\n",
    "from random import sample\n",
    "import re\n",
    "import pprint\n",
    "\n",
    "from rdkit import Chem\n",
    "from rdkit.Chem import Descriptors\n",
    "from rdkit.Chem.Descriptors import *\n",
    "\n",
    "#from rdkit.Chem import MCS\n",
    "from rdkit.Chem import rdFMCS\n",
    "from rdkit.Chem import fmcs\n",
    "from rdkit.Chem import Draw\n",
    "from rdkit.Chem import PandasTools\n",
    "import xlsxwriter\n",
    "\n",
    "import EFGs\n",
    "from EFGs import mol2frag,ifg,identify_functional_groups\n",
    "\n",
    "#data = pd.read_csv('hob_data_set_MSMcleanup.csv')\n",
    "data = pd.read_csv('bioavailability_data.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "72f6b6f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# addition of Murcko to data\n",
    "from rdkit.Chem.Scaffolds import MurckoScaffold\n",
    "df = data\n",
    "for index,row in df.iterrows():\n",
    "    drugname = df.loc[index,'Name']\n",
    "    drugSMILES = df.loc[index,'smile']\n",
    "    drugmol = Chem.MolFromSmiles(drugSMILES)\n",
    "    \n",
    "    murcko_core = MurckoScaffold.GetScaffoldForMol(drugmol)\n",
    "    result = Chem.MolToSmiles(murcko_core)\n",
    "    df.loc[index,'murcko'] = result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "f80e26ff",
   "metadata": {},
   "outputs": [],
   "source": [
    "''' # CLEAN SMILES '''\n",
    "import re\n",
    "def clean_smiles(smilesx,option=1):\n",
    "    if option==2:\n",
    "        expr = re.compile(r'[0-9]+\\*')\n",
    "        out = expr.sub('*',smilesx)\n",
    "    else:\n",
    "        expr = re.compile(r'\\(?\\[?[0-9]+\\*\\]?\\)?')\n",
    "        out = expr.sub('',smilesx)\n",
    "    #out = []\n",
    "    #for k in smilesList:\n",
    "        #k2 = expr.sub('',k)\n",
    "        #out.append(k2)\n",
    "    \n",
    "    return out\n",
    "\n",
    "def clean_smiles_from_mol_list(mol_list,option=1):\n",
    "    list_out = []\n",
    "    for x in mol_list:\n",
    "        smilesx = Chem.MolToSmiles(x)\n",
    "        if option==2:\n",
    "            expr = re.compile(r'[0-9]+\\*')\n",
    "            out = expr.sub('*',smilesx)\n",
    "            list_out.append(out)\n",
    "        elif option==3:\n",
    "            expr = re.compile(r'\\(?\\[?[0-9]+\\*\\]?\\)?')\n",
    "            out = expr.sub('',smilesx)\n",
    "            \n",
    "            #out2 = Chem.MolFromSmiles(out)\n",
    "            #out3 = Chem.MolToSmiles(out2)\n",
    "            \n",
    "            expr = re.compile(r'\\([A-z]+\\)')\n",
    "            #out4 = expr.sub('',out3)\n",
    "            out2 = expr.sub('',out)\n",
    "            #out3 = Chem.MolFromSmiles(out2)\n",
    "            #out4 = Chem.MolToSmiles(out3)\n",
    "            #list_out.append(out4)\n",
    "            list_out.append(out2)\n",
    "        else:\n",
    "            expr = re.compile(r'\\(?\\[?[0-9]+\\*\\]?\\)?')\n",
    "            out = expr.sub('',smilesx)\n",
    "            list_out.append(out)\n",
    "    return list_out\n",
    "\n",
    "''' # BRICS FRAGMENTATION '''\n",
    "from rdkit.Chem import BRICS\n",
    "\n",
    "def gimme_BRICS(drugmol,sorting=True):\n",
    "    fragMols = []\n",
    "    #fragSmiles1 = []\n",
    "    #fragSmiles2 = []\n",
    "    \n",
    "    for x in BRICS.BRICSDecompose(drugmol):\n",
    "        fragMols.append(Chem.MolFromSmiles(x))\n",
    "        \n",
    "        #smiles = clean_smiles(x,option=1)\n",
    "        #fragSmiles1.append(smiles)\n",
    "        #smiles = clean_smiles(x,option=2)\n",
    "        #fragSmiles2.append(smiles)\n",
    "        \n",
    "    if sorting:\n",
    "        fragMols.sort(reverse=True,key = lambda x:x.GetNumAtoms())\n",
    "    else:\n",
    "        pass\n",
    "    #return fragMols,fragSmiles1,fragSmiles2\n",
    "    return fragMols"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "acd811fb",
   "metadata": {},
   "source": [
    "### visual example"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "c278af05",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Drug name:  stemazole\n",
      "Drug SMILES code:\t NN=C(S)Nc1ccc(-c2nnc(S)o2)cc1\n",
      "Fragment SMILES raw:\t ['[16*]c1ccc([16*])cc1', '[14*]c1nnc(S)o1', '[5*]NC(S)=NN']\n",
      "Fragment SMILES v1:\t ['c1ccccc1', 'c1nnc(S)o1', 'NC(S)=NN']\n",
      "Fragment SMILES v4test:\t ['c1ccccc1', 'Sc1nnco1', 'NN=C(N)S']\n",
      "Fragment SMILES v2:\t ['[*]c1ccc([*])cc1', '[*]c1nnc(S)o1', '[*]NC(S)=NN']\n",
      "Fragment SMILES v3:\t ['c1ccccc1', 'c1nnco1', 'NC=NN']\n",
      "Murcko Frag. SMILES v3:\t ['c1ccccc1', 'c1nnco1']\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "''' # EXAMPLE '''\n",
    "\n",
    "testmol= Chem.MolFromSmiles(df.loc[1587,'smile'])\n",
    "fragMols = gimme_BRICS(testmol)\n",
    "\n",
    "murckomol = Chem.MolFromSmiles(df.loc[1587,'murcko'])\n",
    "murckofrags = gimme_BRICS(murckomol)\n",
    "\n",
    "fragSmiles1 = clean_smiles_from_mol_list(fragMols,option=1)\n",
    "fragSmiles2 = clean_smiles_from_mol_list(fragMols,option=2)\n",
    "fragSmiles3 = clean_smiles_from_mol_list(fragMols,option=3)\n",
    "\n",
    "simplified_drug_smiles = clean_smiles_from_mol_list([testmol],option=3)[0]\n",
    "simplified_drugmol = Chem.MolFromSmiles(simplified_drug_smiles)\n",
    "\n",
    "fragSmiles4test = []\n",
    "for x in fragSmiles1:\n",
    "    temp = Chem.MolFromSmiles(x)\n",
    "    Chem.SanitizeMol(temp)\n",
    "    fragSmiles4test.append(Chem.MolToSmiles(temp))\n",
    "    \n",
    "murckofragSmiles3 = clean_smiles_from_mol_list(murckofrags,option=3)\n",
    "\n",
    "frags_molFromSmiles1 = [Chem.MolFromSmiles(x) for x in fragSmiles1]\n",
    "frags_molFromSmiles2 = [Chem.MolFromSmiles(x) for x in fragSmiles2]\n",
    "frags_molFromSmiles3 = [Chem.MolFromSmiles(x) for x in fragSmiles3]\n",
    "frags_molFromMurckoFrags = [Chem.MolFromSmiles(x) for x in murckofragSmiles3]\n",
    "frags_molFromSmiles4test = [Chem.MolFromSmiles(x) for x in fragSmiles4test]\n",
    "print(\"Drug name: \",drugname)\n",
    "print(\"Drug SMILES code:\\t\",drugSMILES)\n",
    "print(\"Fragment SMILES raw:\\t\",[Chem.MolToSmiles(x) for x in fragMols])\n",
    "print(\"Fragment SMILES v1:\\t\",fragSmiles1)\n",
    "print(\"Fragment SMILES v4test:\\t\",fragSmiles4test)\n",
    "print(\"Fragment SMILES v2:\\t\",fragSmiles2)\n",
    "print(\"Fragment SMILES v3:\\t\",fragSmiles3)\n",
    "print(\"Murcko Frag. SMILES v3:\\t\",murckofragSmiles3)\n",
    "\n",
    "\n",
    "molsOut = []\n",
    "molsOut.append(drugmol)\n",
    "molsOut.extend(fragMols)\n",
    "molsOut.append(simplified_drugmol)\n",
    "molsOut.extend(frags_molFromSmiles3)\n",
    "molsOut.append(murckomol)\n",
    "molsOut.extend(frags_molFromMurckoFrags)\n",
    "#molsOut.extend(frags_molFromSmiles1)\n",
    "#molsOut.extend(frags_molFromSmiles4test)\n",
    "#molsOut.extend(frags_molFromSmiles2)\n",
    "\n",
    "labels_i = ['drug','raw fragment','raw fragment','raw fragment',\n",
    "            'simplified drug','simplified fragment','simplified fragment','simplified fragment',\n",
    "            'murcko core','murcko fragment','murcko fragment']\n",
    "labels = []\n",
    "for i,j in enumerate(molsOut):\n",
    "    part1 = labels_i[i]\n",
    "    part2 = j.GetNumAtoms()\n",
    "    part3 = j.GetNumHeavyAtoms()\n",
    "    label = f\"{part1} \\n{part2}atoms, {part3}heavy\"\n",
    "    labels.append(label)\n",
    "Draw.MolsToGridImage(molsOut,molsPerRow=4,legends=labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "00254091",
   "metadata": {},
   "source": [
    "### build fragments dictionary\n",
    "* We will make a Drug Dict, and a Frag Dict\n",
    "* Drug Dict (PK = `drug_name`) will be used to store list of BRICS frags\n",
    "* Frag Dict (PK = `frag_smiles`) will store: \n",
    "  * drug_list = list of each `drug_name`\n",
    "  * size = `frag_mol.GetNumHeavyAtoms()`\n",
    "  * frequency = `len(drug_list)`\n",
    "* Each Frag Dict PK will be entered based on the fragment smiles BOTH v1 (simplified) and v2 (specific)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "258fa2af",
   "metadata": {},
   "outputs": [],
   "source": [
    "drug_dict = {}\n",
    "frag_dict = {}\n",
    "fragment_tracker = set()\n",
    "frag_pairs_dict = {}\n",
    "\n",
    "df = data\n",
    "for index,row in df.iterrows():\n",
    "    #if index < 50:\n",
    "    drugname = df.loc[index,'Name']\n",
    "    drugSMILES = df.loc[index,'smile']\n",
    "    drugmol = Chem.MolFromSmiles(drugSMILES)\n",
    "    Mcore_smiles = df.loc[index,'murcko']\n",
    "    \n",
    "    drug_dict[drugname] = {}\n",
    "    drug_dict[drugname]['smiles'] = drugSMILES\n",
    "    drug_dict[drugname]['mol'] = drugmol\n",
    "    drug_dict[drugname]['Mcore_smiles'] = Mcore_smiles\n",
    "    \n",
    "    for atom in drugmol.GetAtoms():\n",
    "        atom.SetIsotope(0)\n",
    "    params = Chem.AdjustQueryParameters()\n",
    "    params.makeBondsGeneric = True\n",
    "    params.makeDummiesQueries = True\n",
    "    drug_patt = Chem.AdjustQueryProperties(drugmol,params)\n",
    "    \n",
    "    drug_dict[drugname]['drug_patt'] = drug_patt\n",
    "    \n",
    "    fragMols = gimme_BRICS(drugmol)\n",
    "    fragSmiles1 = clean_smiles_from_mol_list(fragMols,option=1)\n",
    "    fragSmiles2 = clean_smiles_from_mol_list(fragMols,option=2)\n",
    "    fragSmiles3 = clean_smiles_from_mol_list(fragMols,option=3)\n",
    "    \n",
    "    drug_dict[drugname]['BRICS_mols'] = fragMols\n",
    "    drug_dict[drugname]['BRICS_smilesV1'] = fragSmiles1\n",
    "    drug_dict[drugname]['BRICS_smilesV2'] = fragSmiles2\n",
    "    drug_dict[drugname]['BRICS_smilesV3'] = fragSmiles3\n",
    "    drug_dict[drugname]['BRICS_count'] = len(fragMols)\n",
    "    \n",
    "    \n",
    "    frag_patts = []\n",
    "    for fragx in fragMols:\n",
    "        frag_patt = fragx\n",
    "        for atom in frag_patt.GetAtoms():\n",
    "            atom.SetIsotope(0)\n",
    "        params = Chem.AdjustQueryParameters()\n",
    "        params.makeBondsGeneric = True\n",
    "        params.makeDummiesQueries = True\n",
    "        frag_patt = Chem.AdjustQueryProperties(frag_patt,params)\n",
    "        frag_patts.append(frag_patt)\n",
    "    \n",
    "    drug_dict[drugname]['frag_patts'] = frag_patts\n",
    "    \n",
    "    import EFGs\n",
    "    from EFGs import mol2frag,ifg,identify_functional_groups\n",
    "    drugFrags_EFGS = mol2frag(drugmol)[0]\n",
    "\n",
    "    drug_dict[drugname]['frags_EFGS'] = drugFrags_EFGS\n",
    "    \n",
    "    # set empty lists for frag pairings (later)\n",
    "    drug_dict[drugname]['BRICS_pairs_smile'] = []\n",
    "    drug_dict[drugname]['BRICS_pairs_size'] = []\n",
    "    drug_dict[drugname]['BRICS_pairs_frequency'] = []\n",
    "    drug_dict[drugname]['BRICS_pairs_mol'] = []\n",
    "    \n",
    "    # also BRICS decomposition on the Murcko Scaffold, for simplified fragments\n",
    "    murckomol = Chem.MolFromSmiles(df.loc[index,'murcko'])\n",
    "    \n",
    "    for atom in murckomol.GetAtoms():\n",
    "        atom.SetIsotope(0)\n",
    "    params = Chem.AdjustQueryParameters()\n",
    "    params.makeBondsGeneric = True\n",
    "    params.makeDummiesQueries = True\n",
    "    murcko_patt = Chem.AdjustQueryProperties(murckomol,params)\n",
    "    \n",
    "    drug_dict[drugname]['murcko_mol'] = murckomol\n",
    "    drug_dict[drugname]['murcko_patt'] = murcko_patt\n",
    "    \n",
    "    murckofragMols = gimme_BRICS(murckomol)\n",
    "    \n",
    "    murckofrag_patts = []\n",
    "    for fragx in murckofragMols:\n",
    "        for atom in fragx.GetAtoms():\n",
    "            atom.SetIsotope(0)\n",
    "        params = Chem.AdjustQueryParameters()\n",
    "        params.makeBondsGeneric = True\n",
    "        params.makeDummiesQueries = True\n",
    "        frag_patt = Chem.AdjustQueryProperties(fragx,params)\n",
    "        murckofrag_patts.append(frag_patt)\n",
    "    \n",
    "    drug_dict[drugname]['murckoFrag_mols'] = murckofragMols\n",
    "    drug_dict[drugname]['murckoFrag_patts'] = murckofrag_patts\n",
    "    \n",
    "    murckofragSmiles1 = clean_smiles_from_mol_list(murckofragMols,option=1)\n",
    "    murckofragSmiles3 = clean_smiles_from_mol_list(murckofragMols,option=3)\n",
    "    \n",
    "    drug_dict[drugname]['murckoFrag_smiles1'] = murckofragSmiles1\n",
    "    drug_dict[drugname]['murckoFrag_smiles3'] = murckofragSmiles3\n",
    "    drug_dict[drugname]['murckoFrag_count'] = len(murckofragMols)\n",
    "    \n",
    "    murckoFrags_EFGS = mol2frag(murckomol)[0]\n",
    "    drug_dict[drugname]['murckoFrags_EFGS'] = murckoFrags_EFGS\n",
    "    \n",
    "    # categorize Sidechains using difference between All Fragments vs. Murcko (core) Fragments   \n",
    "    sidechains = []\n",
    "    for x in fragSmiles3:\n",
    "        if x in murckofragSmiles3:\n",
    "            pass\n",
    "        else:\n",
    "            sidechains.append(x)\n",
    "    drug_dict[drugname]['sidechain_frags'] = sidechains\n",
    "    \n",
    "    tempset = set()\n",
    "    for L in [fragSmiles1,fragSmiles2,fragSmiles3,murckofragSmiles1,murckofragSmiles3]:\n",
    "        for item in L:\n",
    "            tempset.add(item)\n",
    "    drug_dict[drugname]['all_distinct_frags'] = tempset\n",
    "    \n",
    "\n",
    "    # Create Dictionary for Individual Fragments\n",
    "    for i,j in enumerate(fragMols):\n",
    "        frag_smiles_1 = fragSmiles1[i]\n",
    "        frag_smiles_2 = fragSmiles2[i]\n",
    "        frag_smiles_3 = fragSmiles3[i]\n",
    "        frag_pattern = frag_patts[i]\n",
    "        #frag_size = j.GetNumHeavyAtoms()\n",
    "        \n",
    "        #allfraglist = fragSmiles1[:]\n",
    "        #allfraglist.extend(fragSmiles2)\n",
    "        #allfraglist.extend(fragSmiles3)\n",
    "        # record entries from BOTH keys... simplified (v1) + specific (v2) fragments\n",
    "        for k0 in [(frag_smiles_1,'v1'),(frag_smiles_2,'v2'),(frag_smiles_3,'v3')]:\n",
    "            k = k0[0]\n",
    "            if k not in frag_dict.keys():\n",
    "                temp_list = []\n",
    "                temp_list.append(drugname)\n",
    "                frag_dict[k] = {}\n",
    "                #frag_dict[k]['size'] = frag_size\n",
    "                frag_dict[k]['frequency'] = len(temp_list)\n",
    "                frag_dict[k]['drugs'] = temp_list\n",
    "                frag_dict[k]['frag_mol'] = j\n",
    "                frag_dict[k]['frag_patt'] = frag_pattern\n",
    "                frag_dict[k]['size'] = j.GetNumHeavyAtoms()\n",
    "                templist2 = set()\n",
    "                templist2.add(f\"smiles_{k0[1]}\")\n",
    "                frag_dict[k]['source'] = templist2\n",
    "            else:\n",
    "                frag_dict[k]['drugs'].append(drugname) \n",
    "                frag_dict[k]['frequency'] = len(frag_dict[k]['drugs'])\n",
    "                frag_dict[k]['source'].add(f\"smiles_{k0[1]}\")\n",
    "                #frag_dict[k]['frequency'] = len(frag_dict[frag_smiles_1]['drugs'])\n",
    "                \n",
    "    # ADD  MURCKO CORE TO FRAG_DICT\n",
    "    \n",
    "    k = Mcore_smiles\n",
    "    j = murckomol\n",
    "    frag_size = j.GetNumHeavyAtoms()\n",
    "    if k not in frag_dict.keys():\n",
    "        temp_list = []\n",
    "        temp_list.append(drugname)\n",
    "        frag_dict[k] = {}\n",
    "        frag_dict[k]['size'] = frag_size\n",
    "        frag_dict[k]['frequency'] = len(temp_list)\n",
    "        frag_dict[k]['drugs'] = temp_list\n",
    "        frag_dict[k]['frag_mol'] = j\n",
    "        frag_dict[k]['frag_patt'] = murcko_patt\n",
    "        templist2 = set()\n",
    "        templist2.add(\"murcko_core\")\n",
    "        frag_dict[k]['source'] = templist2\n",
    "    else:\n",
    "        frag_dict[k]['drugs'].append(drugname) \n",
    "        frag_dict[k]['frequency'] = len(frag_dict[k]['drugs'])\n",
    "        frag_dict[k]['source'].add(\"murcko_core\")\n",
    "    \n",
    "    # ALSO ADD  MURCKO FRAGMENTS TO FRAG_DICT\n",
    "    for i,j in enumerate(murckofragMols):\n",
    "        Mfrag_smiles_1 = murckofragSmiles1[i]\n",
    "        Mfrag_smiles_3 = murckofragSmiles3[i]\n",
    "        Mfrag_size = j.GetNumHeavyAtoms()\n",
    "        murcko_frag_pattern = murckofrag_patts[i]\n",
    "        \n",
    "        #allMfraglist = murckofragSmiles1[:]\n",
    "        #allMfraglist.extend(murckofragSmiles3)\n",
    "        #for k in allMfraglist:\n",
    "        for k0 in [(Mfrag_smiles_1,\"v1\"),(Mfrag_smiles_3,\"v3\")]:\n",
    "            k = k0[0]\n",
    "            if k not in frag_dict.keys():\n",
    "                temp_list = []\n",
    "                temp_list.append(drugname)\n",
    "                frag_dict[k] = {}\n",
    "                frag_dict[k]['size'] = frag_size\n",
    "                frag_dict[k]['frequency'] = len(temp_list)\n",
    "                frag_dict[k]['drugs'] = temp_list\n",
    "                frag_dict[k]['frag_mol'] = j\n",
    "                frag_dict[k]['frag_patt'] = murcko_frag_pattern\n",
    "                templist2 = set()\n",
    "                templist2.add(f\"murckofrag_{k0[1]}\")\n",
    "                frag_dict[k]['source'] = templist2\n",
    "            else:\n",
    "                frag_dict[k]['drugs'].append(drugname) \n",
    "                frag_dict[k]['frequency'] = len(frag_dict[k]['drugs'])\n",
    "                frag_dict[k]['source'].add(f\"murckofrag_{k0[1]}\")\n",
    "    \n",
    "\n",
    "    # Create Dictionary for Fragment pairings\n",
    "    for i1,j1 in enumerate(fragSmiles3): #use fragSmiles3 because that's most simplified version\n",
    "        frag_x = j1\n",
    "        size_x = fragMols[i1].GetNumHeavyAtoms()\n",
    "\n",
    "        if j1 not in frag_pairs_dict.keys():\n",
    "            frag_pairs_dict[j1] = {}\n",
    "\n",
    "        for i2,j2 in enumerate(fragSmiles3):\n",
    "            if i2 == i1:\n",
    "                pass\n",
    "            #elif j2 in frag_pairs_dict.keys():\n",
    "            #    pass\n",
    "            else:\n",
    "                frag_y = j2\n",
    "                size_y = fragMols[i2].GetNumHeavyAtoms()\n",
    "\n",
    "                if j2 not in frag_pairs_dict[j1].keys():\n",
    "                    temp_list = []\n",
    "                    temp_list.append(drugname)\n",
    "                    frag_pairs_dict[j1][j2] = {}\n",
    "                    frag_pairs_dict[j1][j2]['pair_size'] = (size_x + size_y)\n",
    "                    frag_pairs_dict[j1][j2]['frequency'] = len(temp_list)\n",
    "                    frag_pairs_dict[j1][j2]['drugs'] = temp_list\n",
    "                    \n",
    "                else:\n",
    "                    frag_pairs_dict[j1][j2]['drugs'].append(drugname)\n",
    "                    frag_pairs_dict[j1][j2]['frequency'] = len(frag_pairs_dict[j1][j2]['drugs'])\n",
    "                \n",
    "                drug_dict[drugname]['BRICS_pairs_smile'].append((frag_x,frag_y))\n",
    "                drug_dict[drugname]['BRICS_pairs_size'].append((size_x + size_y))\n",
    "                drug_dict[drugname]['BRICS_pairs_frequency'].append(len(frag_pairs_dict[j1][j2]['drugs']))\n",
    "                drug_dict[drugname]['BRICS_pairs_mol'].append((fragMols[i1],fragMols[i2]))\n",
    "    #else:\n",
    "    #    pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "9271d596",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Recalculate Fragment Sizes, record to Frag DIct\n",
    "for frag_smiles in frag_dict.keys():\n",
    "    resize = frag_dict[frag_smiles]['frag_mol'].GetNumHeavyAtoms()\n",
    "    recordedSize = frag_dict[frag_smiles]['size']\n",
    "    #frag_dict[frag_smiles]['size (old)'] = recordedSize\n",
    "    frag_dict[frag_smiles]['size'] = resize"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "5c42fc9c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# backup!!\n",
    "# save frag_dict again\n",
    "import pickle\n",
    "filename = 'frag_dict__final.pickle'\n",
    "with open(filename, \"wb\") as handle:\n",
    "    pickle.dump(frag_dict, handle, protocol=pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "4e68d094",
   "metadata": {},
   "outputs": [],
   "source": [
    "'''\n",
    "# NEW VERSION - last version only searched simpler types of fragments, getting druglist3. This is druglist4 now.\n",
    "'''\n",
    "# Recalculate Drug/Fragment Matches, record to Frag DIct\n",
    "matches = []\n",
    "\n",
    "fragList = enumerate([x for x in frag_dict.keys()])\n",
    "drugList = [x for x in drug_dict.keys()]\n",
    "              \n",
    "for i,j in fragList:\n",
    "    \n",
    "    frag_dict[j]['druglist4'] = []\n",
    "    \n",
    "    #if frag_dict[j]['source'].intersection({'murckofrag_v3','smiles_v3','murcko_core'}) != set():\n",
    "    fragPattern = frag_dict[j]['frag_patt']\n",
    "\n",
    "    for index,key in enumerate(drugList):\n",
    "        drugmol = drug_dict[key]['mol']\n",
    "\n",
    "        drugname = key\n",
    "        drugmol = drug_dict[drugname]['mol']\n",
    "        drug_patt = drug_dict[drugname]['drug_patt']\n",
    "        drug_murcko_patt = drug_dict[drugname]['murcko_patt']\n",
    "\n",
    "        if drug_patt.HasSubstructMatch(fragPattern):\n",
    "            frag_dict[j]['druglist4'].append(key)\n",
    "            frag_dict[j]['frequency4'] = len(frag_dict[j]['druglist4'])\n",
    "\n",
    "        # New snippet - if drug pattern doesnt match fragment, check drug murcko as well\n",
    "        elif drug_murcko_patt.HasSubstructMatch(fragPattern):\n",
    "            if key not in set(frag_dict[j]['druglist4']):\n",
    "                frag_dict[j]['druglist4'].append(key)\n",
    "                frag_dict[j]['frequency4'] = len(frag_dict[j]['druglist4'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "24dcc6f6",
   "metadata": {},
   "outputs": [],
   "source": [
    "'''\n",
    "# NEW - using frag_dict 'druglist4', let's update drug_dict 'fragments4'\n",
    "'''\n",
    "frags = [(n,idx) for (n,idx) in enumerate(frag_dict.keys())]\n",
    "\n",
    "for idx,frag in frags:\n",
    "    \n",
    "    frag_size = frag_dict[frag]['size']\n",
    "    frag_freq = len(frag_dict[frag]['druglist4'])\n",
    "    \n",
    "    frag_info = (frag, frag_size, frag_freq)\n",
    "    \n",
    "    drug_info = []\n",
    "    for drug in frag_dict[frag]['druglist4']:\n",
    "        \n",
    "        if 'fragments4' not in drug_dict[drug].keys():\n",
    "            \n",
    "            drug_dict[drug]['fragments4'] = []\n",
    "            drug_dict[drug]['fragments4'].append( frag_info )\n",
    "            \n",
    "            drug_dict[drug]['fragments4_short'] = []\n",
    "            drug_dict[drug]['fragments4_short'].append( frag )\n",
    "            \n",
    "        else:\n",
    "            drug_dict[drug]['fragments4'].append( frag_info )\n",
    "            drug_dict[drug]['fragments4_short'].append( frag )\n",
    "        \n",
    "        \n",
    "        drug_row_index     =  df[df['Name'] == drug].index[0]\n",
    "        drug_BA_value      =  df['value (oral BA %)'][drug_row_index]\n",
    "        \n",
    "        drug_info.append( (drug, drug_BA_value) )\n",
    "        \n",
    "        drug_dict[drug]['bioavailability'] = drug_BA_value\n",
    "    \n",
    "    # save new data in frag_dict\n",
    "    if 'parent_drug_info' not in frag_dict[frag].keys(): \n",
    "        frag_dict[frag]['parent_drug_info'] = drug_info\n",
    "    else:\n",
    "        frag_dict[frag]['parent_drug_info'].extend(drug_info)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "b87a93b9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# backup!!\n",
    "# save frag_dict again\n",
    "import pickle\n",
    "filename = 'frag_dict__final.pickle'\n",
    "with open(filename, \"wb\") as handle:\n",
    "    pickle.dump(frag_dict, handle, protocol=pickle.HIGHEST_PROTOCOL)\n",
    "    \n",
    "filename = 'drug_dict__final.pickle'\n",
    "with open(filename, \"wb\") as handle:\n",
    "    pickle.dump(drug_dict, handle, protocol=pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "57c74b0e",
   "metadata": {},
   "source": [
    "* (below) this combines `'drugs'` & `'druglist3'` column of `frag_dict` into `'druglist_final'` \n",
    "* <font color='red'> in latest version, we `'druglist3'` should already be equal to `'druglist_final'` </font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "7ea0dc20",
   "metadata": {},
   "outputs": [],
   "source": [
    "count = 0\n",
    "for idx,frag in frags:\n",
    "    druglist_orig = frag_dict[frag]['drugs']\n",
    "    druglist_v4 = frag_dict[frag]['druglist4']\n",
    "    drugs_consolidated = set()\n",
    "    [drugs_consolidated.add(x) for x in druglist_orig]\n",
    "    [drugs_consolidated.add(x) for x in druglist_v4]\n",
    "    frag_dict[frag]['druglist4_final'] = drugs_consolidated\n",
    "    frag_dict[frag]['frequency4_final'] = len(drugs_consolidated)\n",
    "    \n",
    "    # delete incorrect parent_drug_info entries\n",
    "    if 'parent_drug_info' in frag_dict[frag].keys():\n",
    "        del frag_dict[frag]['parent_drug_info']"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4d0536e2",
   "metadata": {},
   "source": [
    "* (below) this takes newly-combined `'druglist_final'` column of `frag_dict` , <br> and to each drug listed there, adds the current fragment to `'fragments_final'` of `'drug_dict'`\n",
    "* <font color='red'> in latest version, we `'fragments3'`/`'fragments3_short'` should already be equal to `'fragments_final'` </font>"
   ]
  },
  {
   "cell_type": "raw",
   "id": "1787d1fc",
   "metadata": {},
   "source": [
    "#### re-assign fragment lists to drug_dict & restore drug lookup info in frag_dict\n",
    "* after consolidating the fragment lists, need to restore accurate info derived from frag lists\n",
    "  * e.g., fragments list in drug_dict, list of parent_drug_info in fragment dict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "9665e31f",
   "metadata": {},
   "outputs": [],
   "source": [
    "frags = [(n,idx) for (n,idx) in enumerate(frag_dict.keys())]\n",
    "\n",
    "for idx,frag in frags:\n",
    "    drugs_list = frag_dict[frag]['druglist4_final'] \n",
    "    frag_freq = frag_dict[frag]['frequency4_final'] \n",
    "    frag_size = frag_dict[frag]['size']\n",
    "    \n",
    "    frag_info = (frag, frag_size, frag_freq)\n",
    "    \n",
    "    drug_info = []\n",
    "    for drug in drugs_list:\n",
    "        drugdictkey1 = 'fragmentsInfo4_final'\n",
    "        drugdictkey2 = 'fragments4_final'\n",
    "        \n",
    "        if drugdictkey2 not in drug_dict[drug].keys():\n",
    "            \n",
    "            drug_dict[drug][drugdictkey1] = []\n",
    "            drug_dict[drug][drugdictkey1].append( frag_info )\n",
    "            \n",
    "            drug_dict[drug][drugdictkey2] = []\n",
    "            drug_dict[drug][drugdictkey2].append( frag )\n",
    "            \n",
    "            drug_dict[drug]['fragment4_count'] = 1\n",
    "            \n",
    "        else:\n",
    "            drug_dict[drug][drugdictkey1].append( frag_info )\n",
    "            drug_dict[drug][drugdictkey2].append( frag )\n",
    "            drug_dict[drug]['fragment4_count'] = len(drug_dict[drug][drugdictkey2])\n",
    "        \n",
    "        drug_row_index     =  df[df['Name'] == drug].index[0]\n",
    "        drug_BA_value      =  df['value (oral BA %)'][drug_row_index] #drug_dict[drug]['bioavailability']\n",
    "        drug_info.append( (drug, drug_BA_value) )\n",
    "        \n",
    "        # -- record BA value in drug_dict for future reference\n",
    "        drug_dict[drug]['bioavailability'] = drug_BA_value\n",
    "    \n",
    "    # save new data in frag_dict\n",
    "    if 'parent_drug_info' not in frag_dict[frag].keys(): \n",
    "        frag_dict[frag]['parent_drug_info'] = drug_info\n",
    "    else:\n",
    "        frag_dict[frag]['parent_drug_info'].extend(drug_info)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a7481140",
   "metadata": {},
   "source": [
    "# NEXT"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3790ea6f",
   "metadata": {},
   "source": [
    "### update column titles in drug_dict & frag_dict, save, then proceed below w/ updated filename"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "bd13c458",
   "metadata": {},
   "outputs": [],
   "source": [
    "drugs = [(n,idx) for (n,idx) in enumerate(drug_dict.keys())]\n",
    "for idx,drug in drugs:\n",
    "    #drug_fragset = drug_dict[drug]['fragments_final']\n",
    "    drug_fragset = drug_dict[drug]['fragmentsInfo4_final']\n",
    "    #drug_fragset = drug_dict[drug]['fragmentsInfo4_final']\n",
    "    \n",
    "    fragments_subset = []\n",
    "    fragments_subset2 = []\n",
    "    \n",
    "    for (frag,frag_size,frag_freq) in drug_fragset:\n",
    "        if frag_freq > 2 and frag_size > 2 and frag_size < 25:\n",
    "            #fragments_subset.append((frag,frag_size,frag_freq))\n",
    "            fragments_subset.append(frag)\n",
    "            if frag_freq < 1000 and re.findall('[\\[,\\],\\*,\\@,\\#]',x) == []:\n",
    "                fragments_subset2.append(frag)\n",
    "            \n",
    "    drug_dict[drug]['fragments4_subset'] = fragments_subset\n",
    "    drug_dict[drug]['fragments4_subset2'] = fragments_subset2\n",
    "    drug_dict[drug]['count_frags4_subset'] = len(fragments_subset)\n",
    "    drug_dict[drug]['count_frags4_subset2'] = len(fragments_subset2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "45c2376e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# backup!!\n",
    "# save frag_dict again\n",
    "import pickle\n",
    "filename = 'frag_dict__final.pickle'\n",
    "with open(filename, \"wb\") as handle:\n",
    "    pickle.dump(frag_dict, handle, protocol=pickle.HIGHEST_PROTOCOL)\n",
    "    \n",
    "filename = 'drug_dict__final.pickle'\n",
    "with open(filename, \"wb\") as handle:\n",
    "    pickle.dump(drug_dict, handle, protocol=pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "1c869067",
   "metadata": {},
   "outputs": [],
   "source": [
    "drug_dict_df = pd.DataFrame.from_dict(drug_dict)\n",
    "drug_dict_df = drug_dict_df.transpose()\n",
    "drug_dict_df.reset_index(inplace=True)\n",
    "drug_dict_df = drug_dict_df.rename(columns={'index':'drug_name'})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "25403f4f",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_out = drug_dict_df[['drug_name','smiles','bioavailability',\n",
    "                       #'fragmentsInfo3_final','fragment_count',\n",
    "                       #'fragmentsInfo4_final','fragment4_count',\n",
    "                       #'fragments3_final','fragment_count',\n",
    "                       'fragments4_final','fragment4_count',\n",
    "                       #'fragments3_subset','count_frags3_subset',\n",
    "                       'fragments4_subset','count_frags4_subset',\n",
    "                       #'fragments3_subset2',\n",
    "                       'fragments4_subset2',\n",
    "                       'frags_EFGS','BRICS_smilesV3']]#,'BRICS_count']]\n",
    "#df_out.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "8cd05a30",
   "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>drug_name</th>\n",
       "      <th>drug_smiles</th>\n",
       "      <th>ba_pct</th>\n",
       "      <th>frags_all</th>\n",
       "      <th>num_frags_all</th>\n",
       "      <th>frags_subset</th>\n",
       "      <th>num_frags_subset</th>\n",
       "      <th>frags_subset2</th>\n",
       "      <th>frags_efgs</th>\n",
       "      <th>frags_brics</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>sulfadiazine</td>\n",
       "      <td>Nc1ccc(S(=O)(=O)Nc2ncccn2)cc1</td>\n",
       "      <td>90.0</td>\n",
       "      <td>[S(=O)(=O)c1ccc(N)cc1, [*]S(=O)(=O)c1ccc(N)cc1...</td>\n",
       "      <td>126</td>\n",
       "      <td>[S(=O)(=O)c1ccc(N)cc1, [*]S(=O)(=O)c1ccc(N)cc1...</td>\n",
       "      <td>100</td>\n",
       "      <td>[S(=O)(=O)c1ccc(N)cc1, [*]S(=O)(=O)c1ccc(N)cc1...</td>\n",
       "      <td>[N, NS(=O)=O, c1cncnc1, c1ccccc1]</td>\n",
       "      <td>[S(=O)(=O)c1ccccc1, c1ncccn1, N]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>clofarabine</td>\n",
       "      <td>Nc1nc(Cl)nc2c1ncn2C1OC(CO)C(O)C1F</td>\n",
       "      <td>50.0</td>\n",
       "      <td>[N, [*]N[*], *N*, n1cnc2c(N)nc(Cl)nc21, [*]n1c...</td>\n",
       "      <td>237</td>\n",
       "      <td>[n1cnc2cncnc21, C1OCCC1O, c1ncc2ncn(C3CCCO3)c2...</td>\n",
       "      <td>194</td>\n",
       "      <td>[n1cnc2cncnc21, C1OCCC1O, c1ncc2ncn(C3CCCO3)c2...</td>\n",
       "      <td>[N, Cl, O, O, O, F, c1ncc2nc[nH]c2n1]</td>\n",
       "      <td>[n1cnc2cncnc21, C1OCCC1O, CO]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>sulfamethoxazole</td>\n",
       "      <td>Cc1cc(NS(=O)(=O)c2ccc(N)cc2)no1</td>\n",
       "      <td>99.0</td>\n",
       "      <td>[S(=O)(=O)c1ccc(N)cc1, [*]S(=O)(=O)c1ccc(N)cc1...</td>\n",
       "      <td>164</td>\n",
       "      <td>[S(=O)(=O)c1ccc(N)cc1, [*]S(=O)(=O)c1ccc(N)cc1...</td>\n",
       "      <td>125</td>\n",
       "      <td>[S(=O)(=O)c1ccc(N)cc1, [*]S(=O)(=O)c1ccc(N)cc1...</td>\n",
       "      <td>[NS(=O)=O, N, c1ccccc1, c1cnoc1]</td>\n",
       "      <td>[S(=O)(=O)c1ccccc1, c1ccon1, N]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>tolazoline</td>\n",
       "      <td>c1ccc(CC2=NCCN2)cc1</td>\n",
       "      <td>90.0</td>\n",
       "      <td>[N, [*]N[*], *N*, CC1=NCCN1, [*]CC1=NCCN1, c1c...</td>\n",
       "      <td>102</td>\n",
       "      <td>[CC1=NCCN1, [*]CC1=NCCN1, c1ccccc1, [*]c1ccccc...</td>\n",
       "      <td>80</td>\n",
       "      <td>[CC1=NCCN1, [*]CC1=NCCN1, c1ccccc1, [*]c1ccccc...</td>\n",
       "      <td>[N=CN, c1ccccc1]</td>\n",
       "      <td>[CC1=NCCN1, c1ccccc1]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>cotinine</td>\n",
       "      <td>CN1C(=O)CCC1c1cccnc1</td>\n",
       "      <td>97.0</td>\n",
       "      <td>[N, [*]N[*], *N*, CO, [*]CO, C1CCC(=O)N1C, [*]...</td>\n",
       "      <td>175</td>\n",
       "      <td>[c1cccnc1, [*]c1cccnc1, *c1cccnc1, *C1CCC(=O)N...</td>\n",
       "      <td>142</td>\n",
       "      <td>[c1cccnc1, [*]c1cccnc1, *c1cccnc1, *C1CCC(=O)N...</td>\n",
       "      <td>[NC=O, c1ccncc1]</td>\n",
       "      <td>[C1CCC(=O)N1C, c1cccnc1]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          drug_name                        drug_smiles ba_pct  \\\n",
       "0      sulfadiazine      Nc1ccc(S(=O)(=O)Nc2ncccn2)cc1   90.0   \n",
       "1       clofarabine  Nc1nc(Cl)nc2c1ncn2C1OC(CO)C(O)C1F   50.0   \n",
       "2  sulfamethoxazole    Cc1cc(NS(=O)(=O)c2ccc(N)cc2)no1   99.0   \n",
       "3        tolazoline                c1ccc(CC2=NCCN2)cc1   90.0   \n",
       "4          cotinine               CN1C(=O)CCC1c1cccnc1   97.0   \n",
       "\n",
       "                                           frags_all num_frags_all  \\\n",
       "0  [S(=O)(=O)c1ccc(N)cc1, [*]S(=O)(=O)c1ccc(N)cc1...           126   \n",
       "1  [N, [*]N[*], *N*, n1cnc2c(N)nc(Cl)nc21, [*]n1c...           237   \n",
       "2  [S(=O)(=O)c1ccc(N)cc1, [*]S(=O)(=O)c1ccc(N)cc1...           164   \n",
       "3  [N, [*]N[*], *N*, CC1=NCCN1, [*]CC1=NCCN1, c1c...           102   \n",
       "4  [N, [*]N[*], *N*, CO, [*]CO, C1CCC(=O)N1C, [*]...           175   \n",
       "\n",
       "                                        frags_subset num_frags_subset  \\\n",
       "0  [S(=O)(=O)c1ccc(N)cc1, [*]S(=O)(=O)c1ccc(N)cc1...              100   \n",
       "1  [n1cnc2cncnc21, C1OCCC1O, c1ncc2ncn(C3CCCO3)c2...              194   \n",
       "2  [S(=O)(=O)c1ccc(N)cc1, [*]S(=O)(=O)c1ccc(N)cc1...              125   \n",
       "3  [CC1=NCCN1, [*]CC1=NCCN1, c1ccccc1, [*]c1ccccc...               80   \n",
       "4  [c1cccnc1, [*]c1cccnc1, *c1cccnc1, *C1CCC(=O)N...              142   \n",
       "\n",
       "                                       frags_subset2  \\\n",
       "0  [S(=O)(=O)c1ccc(N)cc1, [*]S(=O)(=O)c1ccc(N)cc1...   \n",
       "1  [n1cnc2cncnc21, C1OCCC1O, c1ncc2ncn(C3CCCO3)c2...   \n",
       "2  [S(=O)(=O)c1ccc(N)cc1, [*]S(=O)(=O)c1ccc(N)cc1...   \n",
       "3  [CC1=NCCN1, [*]CC1=NCCN1, c1ccccc1, [*]c1ccccc...   \n",
       "4  [c1cccnc1, [*]c1cccnc1, *c1cccnc1, *C1CCC(=O)N...   \n",
       "\n",
       "                              frags_efgs                       frags_brics  \n",
       "0      [N, NS(=O)=O, c1cncnc1, c1ccccc1]  [S(=O)(=O)c1ccccc1, c1ncccn1, N]  \n",
       "1  [N, Cl, O, O, O, F, c1ncc2nc[nH]c2n1]     [n1cnc2cncnc21, C1OCCC1O, CO]  \n",
       "2       [NS(=O)=O, N, c1ccccc1, c1cnoc1]   [S(=O)(=O)c1ccccc1, c1ccon1, N]  \n",
       "3                       [N=CN, c1ccccc1]             [CC1=NCCN1, c1ccccc1]  \n",
       "4                       [NC=O, c1ccncc1]          [C1CCC(=O)N1C, c1cccnc1]  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_out = df_out.rename(columns={'smiles':'drug_smiles', \n",
    "                                      'bioavailability':'ba_pct',\n",
    "                                      #'fragments3_final':'frags3_all',\n",
    "                                      'fragments4_final':'frags_all',\n",
    "                                      #'fragment_count':'num_frags3_all',\n",
    "                                      'fragment4_count':'num_frags_all',\n",
    "                                'fragments4_subset':'frags_subset',\n",
    "                                'count_frags4_subset':'num_frags_subset',\n",
    "                                'fragments4_subset2':'frags_subset2',\n",
    "                                      'frags_EFGS':'frags_efgs',\n",
    "                                      'BRICS_smilesV3':'frags_brics'})\n",
    "df_out.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "b289fed3",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_out.to_pickle('bioavailabilityData_w_Frags__final.pkl')"
   ]
  }
 ],
 "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": "3.10.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}