def geneInfoH(geneNameH, geneSetH, refSeqSummaryFileName='/Z/Sequence/ucsc_hg19/annot/refSeqSummary.txt', hugoFileName='/Z/Sequence/geneinfo/hugo.txt', \
censusFileName='/Z/Sequence/geneinfo/cancer_gene_census.txt', biocartaFileName='/Z/Sequence/geneinfo/BIOCARTA.gmt', \
goFileName='/Z/Sequence/geneinfo/hugo.txt', keggFileName='/Z/Sequence/geneinfo/hugo.txt'):
geneInfoH = {}
for line in open(refSeqSummaryFileName):
(refSeqId,status,summary) = line[:-1].split('\t')
if refSeqId in geneNameH:
geneName = geneNameH[refSeqId]
if geneName not in geneInfoH:
geneInfoH[geneName] = {}
geneInfoH[geneName]['summary'] = summary
for line in open(hugoFileName):
(geneName,desc,aliases,geneCardName,refSeqIds) = line[:-1].split('\t')
if geneName not in geneInfoH:
geneInfoH[geneName] = {}
geneInfoH[geneName]['desc'] = desc
geneInfoH[geneName]['aliases'] = aliases
geneInfoH[geneName]['refSeqIds'] = refSeqIds
for line in open(censusFileName):
tokL = line[:-1].split('\t')
(geneName,desc,somatic,germline,mutType,translocPartners) = (tokL[0],tokL[1],tokL[7],tokL[8],tokL[12],tokL[13])
if geneName == 'Symbol':
continue
if geneName not in geneInfoH:
geneInfoH[geneName] = {'desc':desc}
geneInfoH[geneName]['census_somatic'] = somatic
geneInfoH[geneName]['census_germline'] = germline
geneInfoH[geneName]['census_mutType'] = mutType
geneInfoH[geneName]['census_translocPartners'] = translocPartners
for geneSetDB in list(geneSetH.keys()):
for (geneSetName,(geneSetDesc,geneNameL)) in geneSetH[geneSetDB].items():
for geneName in geneNameL:
if geneName in geneInfoH:
jkbasic.addHash(geneInfoH[geneName],geneSetDB,(geneSetName,geneSetDesc))
else:
geneInfoH[geneName] = {geneSetDB:[(geneSetName,geneSetDesc)]}
return geneInfoH
def loadRefFlatByGeneName(refFlatFileName='refFlat.txt'):
h = {}
for line in open(refFlatFileName):
r = processBlatLine(line)
jkbasic.addHash(h, r['transName'], r)
return h
def loadRefFlatByChr(refFlatFileName='/%s/D/Sequences/hg19/refFlat_hg19.txt' % (homedir,)):
h = {}
for line in open(refFlatFileName):
r = processBlatLine(line)
jkbasic.addHash(h, r['chrom'], r)
return h