def printFile(table, outfilename):
outStream = open(outfilename, "w")
# printHeader(outStream)
printHeader(outStream)
for item in table:
outStream.write("<%s/%s> \n\ta VP:GeneAnnotation ;\n" % (visumpointGene.GO, item))
outStream.write("\trdfs:subClassOf VP:GeneConcept ;\n")
for symbol in table[item]:
outStream.write("\tVP:GeneSymbol %s;\n" % (visumpointGene.symbol(visumpointGene.strip(symbol))))
for go in table[item][symbol]:
row = table[item][symbol][go]
outStream.write("%sVP:hasReference [\n" % ("\t"))
outStream.write("\t\ta\tVP:GO;\n")
outStream.write(
"\t\tVP:ID %s;\n" % (visumpointGene.urlGO(visumpointGene.strip("GO_" + go[3 : len(go)])))
)
if row[0].startswith("GO_REF"):
outStream.write(
"\t\tVP:GO_REF %s;\n"
% (visumpointGene.urlGO(visumpointGene.strip("GO_" + row[0][7 : len(row[0])])))
)
outStream.write("\t];\n")
outStream.write(".\n\n")
def printClass(outStream, line):
outStream.write("<%s/%s> \n\ta owl:Class ;\n" % (visumpointGene.GO, line[1]))
outStream.write("\tVP:Symbol %s;\n" % (visumpointGene.symbol(visumpointGene.strip(line[2]))))
outStream.write("\tVP:Symbol %s;\n" % (visumpointGene.symbol(visumpointGene.strip(line[2]))))
# outStream.write('\trdfs:label "%s"^^xsd:string ;\n' % (visumpointGene.strip(line[2])))
# outStream.write('\trdfs:subClassOf VP:HUGO ;\n')
outStream.write(".\n\n")
def printDiseaseClass(outStream,line):
outStream.write('<%s/%s> \n\ta %s;\n' % (Namespace, line[0], visumpointGene.VP_PharmGKBDisease))
outStream.write('\trdfs:label "%s"^^xsd:string ;\n' % (visumpointGene.strip(line[1])))
outStream.write('\tskos:prefLabel "%s"^^xsd:string ;\n' % (visumpointGene.strip(line[1])))
trip = ""
useName = ""
if len(line[2]) > 0 :
for name in line[2].split(','):
if name.startswith("\"") and name.endswith("\""):
useName = name
elif name.startswith("\""):
trip = name
elif name.endswith("\""):
useName = (trip + "," +name).strip("\"")
trip = ""
else :
useName = name
if len(useName) > 0:
visumpointGene.printAlternativeName(outStream, "", visumpointGene.strip(useName), visumpointGene.VP_PharmGKB,"\t\t")
useName = ""
db_id_type_strB = "(.+):(.+)\((.+)\[(.+)/(.+)\]"
db_id_PP = r'(.+):(.+)\(([^()]++)\((.*)\)+'
strExp = "(.+):(.+)\((.+)"
if len(line[4]) > 0 :
for name in line[4].split('),'):
#db_id_type_strB = "(.+):(.+)\((.+)\[(.+)/(.+)\]"
try:
match = regex.match(db_id_type_strB, name, regex.M|regex.I)
if (match):
visumpointGene.printCrossReference(outStream, dbReference(match.group(1)), visumpointGene.strip(match.group(2)), visumpointGene.VP_PharmGKB,"\t\t")
else :
#db_id_PP = r'(.+):(.+)\(([^()]++)\((.*)\)+'
try:
match1 = regex.match( db_id_PP, name, re.M|re.I)
if (match1):
visumpointGene.printCrossReference(outStream, dbReference(match1.group(1)), visumpointGene.strip(match1.group(2)), visumpointGene.VP_PharmGKB,"\t\t")
else :
#strExp = "(.+):(.+)\((.+)"
try:
match2 = regex.match( strExp, name, re.M|re.I)
if(match2):
visumpointGene.printCrossReference(outStream, dbReference(match2.group(1)), visumpointGene.strip(match2.group(2)), visumpointGene.VP_PharmGKB,"\t\t")
else :
outStream.write("#ERROR : processing External References : %s\n" % name)
outStream.write("#\t\t %s\n" % line[4])
except:
e = sys.exc_info()[0]
print( "Error: %s" % e )
except:
e = sys.exc_info()[0]
print( "Error: %s" % e )
except:
e = sys.exc_info()[0]
print( "Error: %s" % e )
outStream.write('\t. # %s %s\n\n' % (line[1], line[0]))
def printDrugClass(outStream,line, table):
drugName = visumpointGene.strip(line[1])
outStream.write('<%s/%s> \n\ta %s;\n' % (Namespace, line[0], visumpointGene.VP_PharmGKBDrug))
try :
outStream.write('\trdfs:label "%s"^^xsd:string ;\n' % (drugName))
if(len(line[7])):
outStream.write('\t%s %s;\n' % (visumpointGene.VP_SMILES, visumpointGene.smiles(line[7])))
if len(line[2]) > 0 :
for name in line[2].split(','):
visumpointGene.printGenericName(outStream, "", visumpointGene.strip(name), visumpointGene.VP_PharmGKB,"\t")
if len(line[3]) > 0 :
for name in line[2].split(','):
visumpointGene.printTradeName(outStream, "", visumpointGene.strip(name), visumpointGene.VP_PharmGKB,"\t")
try:
if len(line[6]) > 0 :
for xref in line[6].split(','):
try :
db = xref.split(":")[0]
if db == "go":
id = xref.split(":")[2]
elif db =="url":
id = xref.split(":")[1] + ":" + xref.split(":")[2]
else:
id = xref.split(":")[1]
visumpointGene.printCrossReference(outStream,dbReference(db),visumpointGene.strip(id),visumpointGene.VP_PharmGKB, "\t")
except :
outStream.write('# ERROR PROCESSING printDrugClass - %s %s %s\n\n' % (line[1], line[0], xref))
except:
outStream.write('# ERROR PROCESSING printDrugClass - %s %s\n\n' % (line[1], line[0]))
if drugName in table:
for gene in table[drugName]:
if not gene == NO_GENE:
visumpointGene.printHasGeneRelationship(outStream, visumpointGene.VP_Unknown , visumpointGene.strip(gene), visumpointGene.VP_PharmGKB,"\t")
except:
outStream.write('# ERROR PROCESSING printDrugClass - %s %s\n\n' % (line[1], line[0]))
outStream.write('\t. # %s %s\n\n' % (line[1], line[0]))
def printGeneClass(outStream,line):
outStream.write('<%s/%s> \n\ta %s;\n' % (Namespace, line[0], visumpointGene.VP_PharmGKBGene))
outStream.write('\trdfs:label "%s"^^xsd:string ;\n' % (visumpointGene.strip(line[3])))
visumpointGene.printCrossReference(outStream, visumpointGene.EntrezGene_ID , visumpointGene.strip(line[1]), visumpointGene.VP_PharmGKB, "\t\t")
visumpointGene.printCrossReference(outStream, visumpointGene.Ensembl_ID , visumpointGene.strip(line[2]), visumpointGene.VP_PharmGKB, "\t\t")
visumpointGene.printHasSymbol(outStream, visumpointGene.Ensembl_ID , visumpointGene.strip(line[4]), visumpointGene.VP_PharmGKB,"\t\t")
try:
if line[0] == "PA134868213":
x = 0
if len(line[10]) and line[10] != 'null' :
outStream.write('\tVP:CPIC "%s"^^xsd:boolean ;\n' % (visumpointGene.trueFalse(visumpointGene.strip(line[10]))))
if len(line[11]) > 0 and line[11] != 'null' :
outStream.write('\tVP:Chromosome "%s"^^xsd:string;\n' % (visumpointGene.strip(line[11])))
if len(line[12]) > 0 and line[12] != 'null' :
outStream.write('\tVP:Start "%s"^^xsd:integer;\n' % (visumpointGene.strip(line[12])))
if len(line[13]) > 0 and line[13] != 'null' :
outStream.write('\tVP:End "%s"^^xsd:integer ;\n' % (visumpointGene.strip(line[13])))
except:
outStream.write('# Error XREF %s\n\n\n' % (line[0]))
try:
if len(line[9]) > 0 :
for xref in line[9].split(','):
db =""
id = ""
try:
db = xref.split(":")[0]
if db == "go":
id = xref.split(":")[2]
elif db =="url":
id = xref.split(":")[1] + ":" + xref.split(":")[2]
else:
id = xref.split(":")[1]
if len(db) > 0 and len(id) > 0:
visumpointGene.printCrossReference(outStream,dbReference(visumpointGene.strip(db)),visumpointGene.strip(id),visumpointGene.VP_PharmGKB, "\t\t")
except:
outStream.write('#Error XREF -> %s - %s\n\n\n' % (line[0], xref))
except:
outStream.write('# Error XREF %s\n\n\n' % (line[0]))
#outStream.write('\t\t\t # %s\n' % (visumpointGene.strip(xref)))
outStream.write('\t. # %s\n\n\n' % (line[0]))
def dbReference(inStr):
str = visumpointGene.strip(inStr.lower())
if str == "snomedct":
return visumpointGene.VP_SnoMedCT
if str == "umls":
return visumpointGene.VP_UMLS
if str == "ndfrt":
return visumpointGene.VP_NDFRT
if str == "meddra":
return visumpointGene.VP_MedDRA
if str == "mesh":
return visumpointGene.VP_Mesh
if str == "humancycgene":
return visumpointGene.VP_HumanCycGene
if str == "alfred":
return visumpointGene.VP_Alfred
if str == "ctd":
return visumpointGene.VP_CTD
if str == "ensembl":
return visumpointGene.VP_Ensembl
if str == "entrezgene":
return visumpointGene.VP_EntrezGene
if str == "genatlas":
return visumpointGene.VP_GenAtlas
if str == "genecard":
return visumpointGene.VP_GeneCard
if str == "go":
return visumpointGene.VP_GO
if str == "hgnc":
return visumpointGene.VP_HGNC
if str == "hugo":
return visumpointGene.VP_HUGO
if str == "huge":
return visumpointGene.VP_HUGE
if str == "dBase":
return visumpointGene.VP_DBase
if str == "hgnc":
return visumpointGene.VP_HGNC
if str == "hugo":
return visumpointGene.VP_HUGO
if str == "modbase":
return visumpointGene.VP_ModBas
if str == "mutdb":
return visumpointGene.VP_MutDb
if str.startswith("refseq"):
return visumpointGene.VP_RefSeq
if str == "omim":
return visumpointGene.VP_OMIM
if str == "uniprotkb":
return visumpointGene.VP_UniProtKb
if str == "ucscgenomebrowser":
return visumpointGene.VP_UCSCGenomeBrowser
if str == "iupharreceptor":
return visumpointGene.VP_IupharReceptor
if str == "url":
return visumpointGene.VP_URL
if str == "ttd":
return visumpointGene.VP_TTD
if str == "bindingdb":
return visumpointGene.VP_BindingDb
if str == "chemspider":
return visumpointGene.VP_ChemSpider
if str == "drugbank":
return visumpointGene.VP_DrugBank
if str == "keggcompound":
return visumpointGene.VP_KeggCompound
if str == "pubchemcompound":
return visumpointGene.VP_PubChemCompound
if str == "pubchemsubstance":
return visumpointGene.VP_SubChemSubstance
if str == "dailymed":
return visumpointGene.VP_DailyMed
if str == "het":
return visumpointGene.VP_HET
if str == "clinicaltrials":
return visumpointGene.VP_ClinicalTrials
if str == "dpd":
return visumpointGene.VP_DPD
if str == "chebi":
return visumpointGene.VP_Chebi
if str == "genbank":
return visumpointGene.VP_Genbank
if str == "ndc":
return visumpointGene.VP_NDC
if str == "iupharligand":
return visumpointGene.VP_Iupharligand
if str == "keggdrug":
return visumpointGene.VP_Keggdrug
if str == "pdb":
return visumpointGene.VP_PDB
if str == "refseqprot":
return visumpointGene.VP_RefSeqProt
return "ERROR %s" % inStr
def printRSIDClass(outStream,line):
outStream.write('<%s/%s>\n' % (Namespace, line[1]))
visumpointGene.printCrossReference(outStream,visumpointGene.VP_dbSNP,visumpointGene.strip(line[0]),visumpointGene.VP_PharmGKB,"\t\t")
outStream.write('\t. # %s %s\n\n' % (line[1], line[0]))
def printHUGOClass(outStream,line):
outStream.write('<%s/%s> \n\ta VP:HUGO ;\n' % (HUGO_NAMESPACE, (line[0])[5:len(line[0])]))
outStream.write('\trdfs:label "%s"^^xsd:string ;\n' % (stripStuff(line[2])))
#outStream.write('\trdfs:subClassOf VP:HUGO ;\n')
outStream.write('\tskos:prefLabel "%s"^^xsd:string ;\n' % (stripStuff(line[2])))
#outStream.write('\trdfs:subClassOf hugo:%s ;\n' % (line[0]))
#outStream.write('\thugo:Accession_Numbers BC022009"^^xsd:string ;\n')
outStream.write('\tVP:ApprovedName "%s"^^xsd:string ;\n' % (stripStuff(line[2])))
outStream.write('\tVP:ApprovedSymbol %s;\n' % (visumpointGene.symbol(visumpointGene.strip(line[1]))))
outStream.write('\tVP:ApprovedSymbolString "%s"^^xsd:string;\n' % (visumpointGene.strip(line[1])))
#outStream.write('\thugo:CCDS_IDs "CCDS31071.1"^^xsd:string;\n')
#outStream.write('\thugo:COSMIC_ID "RGS7"^^rdf:XMLLiteral ;\n')
#outStream.write('\thugo:COSMIC_LINK "<a href=\"http://www.sanger.ac.ukz/perl/genetics/CGP/cosmic?action=gene&ln=RGS7\">COSMIC</a>"^^rdf:XMLLiteral ;\n')
outStream.write('\tVP:Chromosome_Sym "%s"^^xsd:string ;\n' % (line[10]))
outStream.write('\tVP:Chromosome "%s"^^xsd:string;\n' % (line[10]))
outStream.write('\tSO:SO_0000105 %s;\n' % (getChromosome(line[10]))) # Chromosome Arm http://en.wikipedia.org/wiki/Locus_(genetics)
outStream.write('\tSO:SO_0000341 %s;\n' % (getChromosome(line[10]))) # Band
outStream.write('\tSO:SO_0000001 %s;\n' % (getChromosome(line[10]))) # Region
if len(line[11]) > 0:
outStream.write('\tVP:ApprovalDate "%s"^^xsd:date ;\n' % (line[11]))
if len(line[12]) > 0:
outStream.write('\tVP:DateModified "%s"^^xsd:date ;\n'% (line[12]))
if len(line[13]) > 0:
outStream.write('\tVP:DateSymbolChanged "%s"^^xsd:date ;\n' % (line[13]))
if len(line[14]) > 0:
outStream.write('\tVP:DateNameChanged "%s"^^xsd:date ;\n' % (line[14]))
#outStream.write('\thugo:Ensembl_ID__mapped_data_supplied_by_Ensembl "ENSG00000182901"^^xsd:string ;\n')
#outStream.write('\thugo:Entrez_Gene_ID "6000"^^xsd:string ;\n')
#outStream.write('\thugo:Entrez_Gene_ID__mapped_data_supplied_by_NCBI "6000"^^xsd:string ;\n')
#outStream.write('\thugo:GDB_ID__mapped_data "GDB:5912686"^^xsd:string ;\n')
#outStream.write('\thugo:Gene_Family_Tag "RGS"^^xsd:string ;\n')
#outStream.write('\thugo:HGNC_ID "HGNC:10003"^^xsd:string ;\n')
outStream.write('\tVP:hasLocusGroup %s;\n' % (LocusGroup(line[5])))
#outStream.write('\thugo:Locus_Specific_Databases "<strong></strong>"^^xsd:string ;\n')
outStream.write('\tVP:hasLocusType %s ;\n' % (LocusType(line[4])))
#outStream.write('\thugo:Mouse_Genome_Database_ID "MGI:1346089"^^xsd:string ;\n')
#outStream.write('\thugo:Mouse_Genome_Database_ID__mapped_data_supplied_by_MGI "MGI:1346089"^^xsd:string ;\n')
#outStream.write('\thugo:OMIM_ID__mapped_data_supplied_by_NCBI "602517"^^xsd:string ;\n')
if len(line[6]) > 0 :
for previousSymbol in line[6].split(','):
outStream.write('\tVP:PreviousSymbol %s ;\n ' % (visumpointGene.symbol(visumpointGene.strip(previousSymbol))))
outStream.write('\tVP:PreviousSymbolString \"%s\"^^xsd:string ;\n' % (stripStuff(previousSymbol)))
if len(line[7]) > 0 :
for previousNames in line[7].split(','):
outStream.write('\tVP:PreviousName \"%s\"^^xsd:string ;\n' % (stripStuff(previousNames)))
if len(line[8]) > 0 :
for synonyms in line[8].split(','):
outStream.write('\tVP:Synonym %s ;\n' % (visumpointGene.symbol(visumpointGene.strip(synonyms))))
outStream.write('\tVP:SynonymString \"%s\"^^xsd:string ;\n' % (stripStuff(synonyms)))
if len(line[22]) > 0 :
for pubmedID in line[22].split(','):
visumpointGene.printCitation(outStream,visumpointGene.PubMed,stripStuff(stripStuff(pubmedID)),"\t")
if len(line[23]) > 0 :
for RefSeq in line[23].split(','):
visumpointGene.printCrossReference(outStream,visumpointGene.RefSeq,stripStuff(RefSeq),"\t")
if len(line[34]) > 0 :
visumpointGene.printCrossReference(outStream,visumpointGene.RefSeq,stripStuff(line[34]),"\t")
if len(line[32]) > 0 :
for eg in line[32].split(','):
visumpointGene.printCrossReference(outStream,visumpointGene.Ensembl_ID,stripStuff(eg),"\t")
if len(line[15]) > 0 :
for an in line[15].split(','):
outStream.write('\tVP:hasAccessionNumber [\n')
outStream.write('\t\t\ta\tVP:AccessionNumber;\n')
outStream.write('\t\t\tVP:ID "%s"^^xsd:string;\n' % (an))
outStream.write('\t\t];\n')
if len(line[16]) > 0 :
visumpointGene.printCrossReference(outStream,visumpointGene.Enzyme_ID,stripStuff(line[16]),"\t")
if len(line[17]) > 0 :
visumpointGene.printCrossReference(outStream,visumpointGene.EntrezGene_ID,stripStuff(line[17]),"\t")
if len(line[18]) > 0 :
visumpointGene.printCrossReference(outStream,visumpointGene.Ensembl_ID,stripStuff(line[18]),"\t")
if len(line[19]) > 0 :
visumpointGene.printCrossReference(outStream,"MouseGenomeID",stripStuff(line[19]),"\t")
if len(line[29]) > 0 :
for an in line[29].split(','):
visumpointGene.printCrossReference(outStream,"CCSDID",stripStuff(an),"\t")
if len(line[30]) > 0 :
for an in line[30].split(','):
visumpointGene.printCrossReference(outStream,visumpointGene.Vega_ID,stripStuff(an),"\t")
if len(line[33]) > 0 :
visumpointGene.printCrossReference(outStream,visumpointGene.OMIM_ID,stripStuff(line[33]),"\t")
if len(line[36]) > 0 :
visumpointGene.printCrossReference(outStream,visumpointGene.Ensembl_ID,stripStuff(line[36]),"\t")
if len(line[37]) > 0 :
visumpointGene.printCrossReference(outStream,"UCSCID",stripStuff(line[37]),"\t")
if len(line[38]) > 0 :
visumpointGene.printCrossReference(outStream,"MouseGenomeID",stripStuff(line[38]),"\t")
if len(line[39]) > 0 :
visumpointGene.printCrossReference(outStream,"RatGenomeID",stripStuff(line[39]),"\t")
#outStream.write('\thugo:Rat_Genome_Database_ID__mapped_data_supplied_by_RGD "RGD:3570"^^xsd:string ;\n')
#outStream.write('\thugo:Record_Type "Standard"^^xsd:string ;\n')
#outStream.write('\thugo:RefSeq__mapped_data_supplied_by_NCBI "NM_002924"^^xsd:string ;\n')
outStream.write('\tVP:Status %s;\n' % (Status(line[3])))
#outStream.write('\thugo:UCSC_ID__mapped_data_supplied_by_UCSC "uc001hyv.2"^^xsd:string ;\n')
#outStream.write('\thugo:UniProt_ID__mapped_data_supplied_by_UniProt "P49802"^^xsd:string ;\n')
#outStream.write('\thugo:VEGA_IDs "OTTHUMG00000040107"^^xsd:string .\n')
outStream.write('.\n')