{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Double Check Cell Knockouts\n",
    "\n",
    "Double check that there are no E3 ligase knockouts, like CRBN-/- cell lines, in the dataset.\n",
    "\n",
    "NOTE: It requires the downloaded file \"cellosaurus.txt\" from the Cellosaurus database to be stored in the `data` directory."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'MOLT-4': 'MOLT-4',\n",
       " 'H1975': 'H1975/WR',\n",
       " 'PBMCs': 'PH1-PBMCs-hiPSC4F1',\n",
       " 'THP-1': 'THP-1',\n",
       " 'K562': 'EGFP-K562',\n",
       " 'MCF-7': 'MCF-7',\n",
       " 'Hs578t': 'Hs578T-Dox',\n",
       " 'THP': 'THP-1',\n",
       " 'HCC827': 'HCC827',\n",
       " 'MOLM14': 'MOLM-14',\n",
       " 'HeLa': 'HeLa',\n",
       " 'HCT116': 'HCT116-53BPI(+/-)',\n",
       " 'Human THP-1 monocytes': 'THP-1',\n",
       " 'MCF-7 breast cancer cells': 'MCF-7',\n",
       " 'NAMALWA': 'Namalwa',\n",
       " 'K562 CML': 'K562/GM-CSF',\n",
       " 'MOLM-13': 'MOLM-13',\n",
       " 'MOLT4': 'MOLT4/P',\n",
       " 'OPM2': 'OPM-2',\n",
       " 'MV-4-11': 'MV4-11',\n",
       " 'MDA-MB-468': 'MDA-MB-468',\n",
       " 'PC-3': 'PC-3',\n",
       " 'Jurkat': 'Jurkat',\n",
       " 'LNCaP': 'LNCaP',\n",
       " 'MCF7': 'MCF7 AREc32',\n",
       " '22Rv1': '22Rv1',\n",
       " 'AML cells': 'OCI-AML-1',\n",
       " 'H661': 'NCI-H661',\n",
       " 'Hela': 'HeLa',\n",
       " 'VCaP': 'VCaP',\n",
       " 'THP1': 'PSC-THP1',\n",
       " 'PC3': 'HNC PC3',\n",
       " 'hela': 'HeLa',\n",
       " 'Karpas422': 'Karpas-422',\n",
       " 'U251': 'U251-TR3',\n",
       " 'MM1S': 'MM1.S',\n",
       " 'Ramos': 'Ramos',\n",
       " 'MDA-MB-231': 'MDA-MB-231',\n",
       " 'LnCap': 'LNCaP',\n",
       " 'HEK293': 'HEK293',\n",
       " 'MiaPaCa2': 'MIA PaCa-2',\n",
       " 'SR': 'SR',\n",
       " 'OVCAR8': 'OVCAR-8',\n",
       " 'Hella': 'IH-1',\n",
       " 'RS4;11': 'RS4;11',\n",
       " 'HEK293T': 'HEK293T',\n",
       " 'RAMOS': 'Ramos',\n",
       " 'A375': 'A375-C5',\n",
       " 'KYSE520 esophageal cancer cell line': 'KYSE-520',\n",
       " 'NCI-H2030': 'NCI-H2030',\n",
       " 'SW1573': 'SW1573',\n",
       " 'SU-DHL-1': 'SU-DHL-1',\n",
       " 'PA1': 'PA1',\n",
       " 'Ba/F3': 'Ba/F3',\n",
       " 'HT1080': 'HT1080 EGFP',\n",
       " 'HepG2': 'HepG2 hALR',\n",
       " 'MyLa1929': 'MyLa 1929',\n",
       " 'A549': 'A549 Cas9',\n",
       " 'Huh7': 'Huh7 IFITM2-/-',\n",
       " 'RI-1': 'Ri-1',\n",
       " 'EOL-1': 'EoL-1',\n",
       " 'A-204': 'A-204',\n",
       " 'T47D': 'T47D Ad12',\n",
       " 'VCaP AR+': 'VCaP AR-V7/pHag',\n",
       " 'NCI-H661': 'NCI-H661',\n",
       " 'NCI-H838': 'NCI-H838',\n",
       " 'NCI-H2228': 'NCI-H2228',\n",
       " '22RV1': '22Rv1',\n",
       " 'HBL-1': 'HBL-1 [Human AIDS-related non-Hodgkin lymphoma]',\n",
       " 'Mino': 'Mino',\n",
       " 'H3255': 'NCI-H3255',\n",
       " 'XLA': 'XLA-07',\n",
       " 'LnCaP95': 'LNCaP95 clone A7',\n",
       " 'Namalwa': 'Namalwa',\n",
       " 'CA-46': 'CA46',\n",
       " 'SUM149': 'SUM149-Luc',\n",
       " 'MV4-11': 'MV4-11',\n",
       " 'HD-MB03': 'HD-MB03',\n",
       " 'Molm-16': 'MOLM-16',\n",
       " '293T': '293T FOXP3',\n",
       " 'H3122': 'NCI-H3122',\n",
       " 'Karpas 299': 'Karpas-299',\n",
       " 'Kelly': 'Kelly',\n",
       " 'RS4; 11': 'RS4;11',\n",
       " 'A431': 'A431 siYAP',\n",
       " 'HCC-827': 'HCC827-Luc [JCRB]',\n",
       " 'platelets': '2T60',\n",
       " 'Panc Tu-I': 'PancTu-I',\n",
       " 'MM.1S': 'MM.1S-luc',\n",
       " 'NCI-H1568': 'NCI-H1568',\n",
       " 'PBMC': 'BSS1-PBMC-iPS4F4',\n",
       " 'DB': 'DB',\n",
       " 'HT-29': 'HT-29',\n",
       " 'SK-MEL-28': 'SK-MEL-28',\n",
       " 'COLO 205': 'COLO 205',\n",
       " 'PC9': 'PC9-BrM3',\n",
       " 'Huh7.5': 'Huh-7.5',\n",
       " 'WI38': 'WI-38 VA13 subline 2RA',\n",
       " 'KYSE520': 'KYSE-520',\n",
       " 'MV4;11': 'MV4-11',\n",
       " 'TM3': 'TM3',\n",
       " 'MDA-MB-436': 'MDA-MB-436',\n",
       " 'germ': 'G1 [Mouse germ stem cell]',\n",
       " 'SW480': 'SW480',\n",
       " 'MB-MDA-231': 'MDA-MB-231',\n",
       " 'HLE': 'HLE',\n",
       " 'HuH-7': 'Huh-7',\n",
       " 'SNU-423': 'SNU-423',\n",
       " 'HUH-1': 'HuH-1',\n",
       " 'SK-Hep-1': 'SK-HEP-1',\n",
       " 'SNU-387': 'SNU-387',\n",
       " 'HLF': 'HLF',\n",
       " 'SNU-398': 'SNU-398',\n",
       " 'HUCCT1': 'HuCC-T1-G10',\n",
       " 'NCI-H23': 'NCI-H23',\n",
       " 'NCI-H358': 'NCI-H358',\n",
       " 'MIA PaCa-2': 'MIA PaCa-2',\n",
       " 'HCC1937': 'HCC1937',\n",
       " 'Capan-1': 'Capan-1',\n",
       " 'SW620': 'SW620',\n",
       " 'KM12': 'KM12',\n",
       " 'HEL': 'HEL',\n",
       " 'Sk-Mel-28': 'SK-MEL-28',\n",
       " 'RAW 264.7': 'RAW 264.7',\n",
       " 'Pfeiffer': 'Pfeiffer',\n",
       " 'DOHH2': 'DoHH2',\n",
       " 'U937': 'U937(CD59+)',\n",
       " 'Z138': 'Z-138',\n",
       " 'OCI-AML2': 'OCI-AML-2',\n",
       " 'OCI-AML3': 'OCI-AML-3',\n",
       " 'Kasumi-1': 'Kasumi-1',\n",
       " 'NB4': 'NB4',\n",
       " 'H1666': 'NCI-H1666',\n",
       " 'CAL-12T': 'CAL-12T',\n",
       " 'SK-MEL-246': 'SK-MEL-246',\n",
       " 'MOLM-14': 'MOLM-14',\n",
       " 'KYSE-270': 'KYSE-270',\n",
       " 'HCT-116': 'HCT 116',\n",
       " 'COLO-205': 'COLO 205',\n",
       " 'Calu-1': 'Calu-1',\n",
       " 'KU812': 'Ku812',\n",
       " 'BT474': 'BT474 A3',\n",
       " 'MIA PaCa2': 'MIA PaCa-2',\n",
       " 'AsPC-1': 'AsPC-1',\n",
       " 'SK-LU-1': 'SK-LU-1',\n",
       " 'HEI-OC1': 'HEI-OC1',\n",
       " 'Bel-7402': 'BEL-7402',\n",
       " 'SH-SY5Y': 'SH-SY5Y',\n",
       " 'H358': 'NCI-H358',\n",
       " 'H23': 'NCI-H23',\n",
       " 'JeKo-1': 'JeKo-1',\n",
       " 'TMD8': 'TMD8',\n",
       " 'MHH-CALL-4': 'MHH-CALL-4',\n",
       " 'KOPN49': 'KOPN-49',\n",
       " 'MOLM-16': 'MOLM-16',\n",
       " 'SUP-M2': 'SUP-M2',\n",
       " 'CCLP1': 'CC-LP-1',\n",
       " 'KATO III': 'KATO III',\n",
       " 'BaF3 FLT3-ITD': 'Ba/F3',\n",
       " 'HEK293A': 'HEK293A mRFP-ATG9',\n",
       " 'HAP1': 'HAP1',\n",
       " 'ER-positive breast cancer cell lines': 'CGBC 01'}"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "with open('../data/processed/cell2identifier.pkl', 'rb') as f:\n",
    "    cell2identifier = pickle.load(f)\n",
    "cell2identifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['MOLT-4', 'H1975/WR', 'PH1-PBMCs-hiPSC4F1', 'THP-1', 'EGFP-K562',\n",
       "       'MCF-7', 'Hs578T-Dox', 'HCC827', 'MOLM-14', nan, 'HeLa',\n",
       "       'HCT116-53BPI(+/-)', 'Namalwa', 'K562/GM-CSF', 'MOLM-13',\n",
       "       'MOLT4/P', 'OPM-2', 'MV4-11', 'MDA-MB-468', 'PC-3', 'Jurkat',\n",
       "       'LNCaP', 'MCF7 AREc32', '22Rv1', 'OCI-AML-1', 'NCI-H661', 'VCaP',\n",
       "       'PSC-THP1', 'HNC PC3', 'Karpas-422', 'U251-TR3', 'MM1.S', 'Ramos',\n",
       "       'MDA-MB-231', 'HEK293', 'MIA PaCa-2', 'SR', 'OVCAR-8', 'IH-1',\n",
       "       'RS4;11', 'HEK293T', 'A375-C5', 'KYSE-520', 'NCI-H2030', 'SW1573',\n",
       "       'SU-DHL-1', 'PA1', 'HT1080 EGFP', 'HepG2 hALR', 'MyLa 1929',\n",
       "       'A549 Cas9', 'Huh7 IFITM2-/-', 'Ri-1', 'EoL-1', 'A-204',\n",
       "       'T47D Ad12', 'VCaP AR-V7/pHag', 'NCI-H838', 'NCI-H2228',\n",
       "       'HBL-1 [Human AIDS-related non-Hodgkin lymphoma]', 'Mino',\n",
       "       'NCI-H3255', 'XLA-07', 'LNCaP95 clone A7', 'CA46', 'SUM149-Luc',\n",
       "       'HD-MB03', 'MOLM-16', '293T FOXP3', 'NCI-H3122', 'Karpas-299',\n",
       "       'Kelly', 'A431 siYAP', 'HCC827-Luc [JCRB]', '2T60', 'PancTu-I',\n",
       "       'MM.1S-luc', 'NCI-H1568', 'BSS1-PBMC-iPS4F4', 'DB', 'HT-29',\n",
       "       'SK-MEL-28', 'COLO 205', 'PC9-BrM3', 'Huh-7.5',\n",
       "       'WI-38 VA13 subline 2RA', 'TM3', 'MDA-MB-436',\n",
       "       'G1 [Mouse germ stem cell]', 'SW480', 'HLE', 'Huh-7', 'SNU-423',\n",
       "       'HuH-1', 'SK-HEP-1', 'SNU-387', 'HLF', 'SNU-398', 'HuCC-T1-G10',\n",
       "       'NCI-H23', 'NCI-H358', 'HCC1937', 'Capan-1', 'SW620', 'KM12',\n",
       "       'HEL', 'RAW 264.7', 'Pfeiffer', 'DoHH2', 'U937(CD59+)', 'Z-138',\n",
       "       'OCI-AML-2', 'OCI-AML-3', 'Kasumi-1', 'NB4', 'NCI-H1666',\n",
       "       'CAL-12T', 'SK-MEL-246', 'Ba/F3', 'KYSE-270', 'HCT 116', 'Calu-1',\n",
       "       'Ku812', 'BT474 A3', 'AsPC-1', 'SK-LU-1', 'HEI-OC1', 'BEL-7402',\n",
       "       'SH-SY5Y', 'JeKo-1', 'TMD8', 'MHH-CALL-4', 'KOPN-49', 'SUP-M2',\n",
       "       'CC-LP-1', 'KATO III', 'HEK293A mRFP-ATG9', 'HAP1', 'CGBC 01'],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "protac_df = pd.read_csv('../data/PROTAC-Degradation-DB.csv')\n",
    "cell_lines = protac_df['Cell Line Identifier'].unique()\n",
    "cell_lines"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Each entry in cellosaurus looks as the following:\n",
    "\n",
    "```\n",
    "ID   #W7079\n",
    "AC   CVCL_E549\n",
    "SY   #W7079 REM; REMUS; W7079\n",
    "DR   dbMHC; 48440\n",
    "DR   ECACC; 94022552\n",
    "DR   IHW; IHW09223\n",
    "DR   IPD-IMGT/HLA; 11548\n",
    "DR   Wikidata; Q54480892\n",
    "RX   PubMed=30844424;\n",
    "CC   Part of: 12th International Histocompatibility Workshop (12IHW) cell line panel.\n",
    "CC   Population: Caucasian.\n",
    "CC   HLA typing: A*02:01,02:06; B*40:01:02:01,67:01:01; C*03,07; DPB1*04:01,05:01 (IPD-IMGT/HLA=11548).\n",
    "CC   Transformant: NCBI_TaxID; 10376; Epstein-Barr virus (EBV).\n",
    "CC   Derived from site: In situ; Peripheral blood; UBERON=UBERON_0000178.\n",
    "CC   Cell type: B-cell; CL=CL_0000236.\n",
    "OX   NCBI_TaxID=9606; ! Homo sapiens (Human)\n",
    "SX   Sex unspecified\n",
    "AG   Age unspecified\n",
    "CA   Transformed cell line\n",
    "DT   Created: 22-10-12; Last updated: 30-01-24; Version: 17\n",
    "//\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load cellosaurus data into a string file\n",
    "with open('data/cellosaurus.txt', 'r') as f:\n",
    "    cellosaurus = f.read()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "cell2cellosaurus = {}\n",
    "for cell_line in cell_lines:\n",
    "    cell2cellosaurus[cell_line] = None\n",
    "    for i, entry in enumerate(cellosaurus.split('//\\n')):\n",
    "        # print(entry)\n",
    "        # print('-' * 80)\n",
    "        if f'ID   {cell_line}' in entry:\n",
    "            cell2cellosaurus[cell_line] = entry\n",
    "\n",
    "            if 'E3' in entry or 'ligase' in entry.lower():\n",
    "                print(cell_line)\n",
    "                print(entry)\n",
    "                print('-' * 80)\n",
    "            break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ID   HCT116-53BPI(+/-)\n",
      "AC   CVCL_1R00\n",
      "DR   cancercelllines; CVCL_1R00\n",
      "DR   RCB; RCB2996\n",
      "DR   Wikidata; Q54882028\n",
      "CC   Population: Caucasian.\n",
      "CC   Knockout cell: Method=Homologous recombination; HGNC; 11999; TP53BP1 (Note=1 of 2 alleles).\n",
      "CC   Sequence variation: Mutation; HGNC; 173; ACVR2A; Simple; p.Lys437Argfs*5 (c.1310delA); dbSNP=rs764719749; Zygosity=Homozygous (from parent cell line).\n",
      "CC   Sequence variation: Mutation; HGNC; 1101; BRCA2; Simple; p.Ile2675Aspfs*6 (c.8021dupA) (c.8021_8022insA); ClinVar=VCV000267050; Zygosity=Heterozygous (from parent cell line).\n",
      "CC   Sequence variation: Mutation; HGNC; 1787; CDKN2A; Simple; p.Arg24Serfs*20 (c.68dupG) (c.68_69insG) (p.G23fs); Zygosity=Heterozygous (from parent cell line).\n",
      "CC   Sequence variation: Mutation; HGNC; 2514; CTNNB1; Simple; p.Ser45del (c.133_135delTCT); ClinVar=VCV000017576; Zygosity=Heterozygous (from parent cell line).\n",
      "CC   Sequence variation: Mutation; HGNC; 3373; EP300; Simple; p.Met1470Cysfs*22 (c.4408delA); Zygosity=Heterozygous (from parent cell line).\n",
      "CC   Sequence variation: Mutation; HGNC; 3373; EP300; Simple; p.Asn1700Thrfs*9 (c.5099delA); Zygosity=Heterozygous (from parent cell line).\n",
      "CC   Sequence variation: Mutation; HGNC; 6407; KRAS; Simple; p.Gly13Asp (c.38G>A); ClinVar=VCV000012580; Zygosity=Heterozygous (from parent cell line).\n",
      "CC   Sequence variation: Mutation; HGNC; 8975; PIK3CA; Simple; p.His1047Arg (c.3140A>G); ClinVar=VCV000013652; Zygosity=Heterozygous (from parent cell line).\n",
      "CC   Sequence variation: Mutation; HGNC; 9277; PPM1D; Simple; p.Leu450Ter (c.1349delT) (p.Leu450fs) (c.1344delT); Zygosity=Heterozygous (from parent cell line).\n",
      "CC   Sequence variation: Mutation; HGNC; 11773; TGFBR2; Simple; p.Lys128Serfs*35 (c.383delA); ClinVar=VCV000477546; Zygosity=Homozygous (from parent cell line).\n",
      "CC   Derived from site: In situ; Colon; UBERON=UBERON_0001155.\n",
      "DI   NCIt; C4910; Colon carcinoma\n",
      "OX   NCBI_TaxID=9606; ! Homo sapiens (Human)\n",
      "HI   CVCL_0291 ! HCT 116\n",
      "SX   Male\n",
      "AG   48Y\n",
      "CA   Cancer cell line\n",
      "DT   Created: 08-07-15; Last updated: 30-01-24; Version: 13\n",
      "\n",
      "--------------------------------------------------------------------------------\n",
      "ID   Huh7 IFITM2-/-\n",
      "AC   CVCL_UZ56\n",
      "DR   cancercelllines; CVCL_UZ56\n",
      "DR   Wikidata; Q94308822\n",
      "RX   PubMed=30723923;\n",
      "CC   Population: Japanese.\n",
      "CC   Knockout cell: Method=CRISPR/Cas9; HGNC; 5413; IFITM2.\n",
      "CC   Sequence variation: Mutation; HGNC; 6307; KDR; Simple; p.Gln472His (c.1416A>T); ClinVar=VCV000134603; Zygosity=Homozygous (from parent cell line).\n",
      "CC   Sequence variation: Mutation; HGNC; 20932; POLD3; Simple; p.Lys109Arg (c.326A>G) (p.Lys70Arg, c.209A>G); Zygosity=Unspecified (from parent cell line).\n",
      "CC   Sequence variation: Mutation; HGNC; 11730; TERT; Simple; c.1-124C>T (c.228C>T) (C228T); Zygosity=Unspecified; Note=In promoter (from parent cell line).\n",
      "CC   Sequence variation: Mutation; HGNC; 11998; TP53; Simple; p.Tyr220Cys (c.659A>G); ClinVar=VCV000127819; Zygosity=Homozygous (from parent cell line).\n",
      "CC   Derived from site: In situ; Liver; UBERON=UBERON_0002107.\n",
      "DI   NCIt; C7956; Adult hepatocellular carcinoma\n",
      "DI   ORDO; Orphanet_210159; Adult hepatocellular carcinoma\n",
      "OX   NCBI_TaxID=9606; ! Homo sapiens (Human)\n",
      "HI   CVCL_0336 ! Huh-7\n",
      "SX   Male\n",
      "AG   57Y\n",
      "CA   Cancer cell line\n",
      "DT   Created: 25-02-19; Last updated: 05-10-23; Version: 10\n",
      "\n",
      "--------------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "for cell_line, cellosaurus_entry in cell2cellosaurus.items():\n",
    "    if pd.isna(cell_line):\n",
    "        continue\n",
    "    if '-/-' in cell_line or '+/-' in cell_line or '+/+' in cell_line or '-/-' in cellosaurus_entry or '+/-' in cellosaurus_entry or '+/+' in cellosaurus_entry:\n",
    "        print(cellosaurus_entry)\n",
    "        print('-' * 80)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "HCT116\n",
      "HCT116-53BPI(+/-)\n",
      "--------------------------------------------------------------------------------\n",
      "Huh7\n",
      "Huh7 IFITM2-/-\n",
      "--------------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "for cell_synonym, cell_id in cell2identifier.items():\n",
    "    if pd.isna(cell_line):\n",
    "        continue\n",
    "    if '-/-' in cell_synonym or '+/-' in cell_synonym or '+/+' in cell_synonym or '-/-' in cell_id or '+/-' in cell_id or '+/+' in cell_id:\n",
    "        print(cell_synonym)\n",
    "        print(cell_id)\n",
    "        print('-' * 80)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "paroutes-env",
   "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": "3.8.19"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}