def loadGFF_UTR(self,fields,line,transcriptBeginEnd,GFF,
transcripts,readOrder,genes):
exonBegin=int(fields[3])-1
exonEnd=int(fields[4])
exonScore=fields[5]
strand=fields[6]
frame=fields[7]
transcriptId=None
rex=Rex()
if(rex.find('transgrp[:=]\s*(\S+)',line)): transcriptId=rex[1]
elif(rex.find('transcript_id[:=]?\s*"?([^\s";]+)"?',line)):
transcriptId=rex[1]
elif(rex.find('Parent=([^;,\s]+)',line)): transcriptId=rex[1]
geneId=None
if(rex.find('genegrp=(\S+)',line)): geneId=rex[1]
elif(rex.find('gene_id[:=]?\s*"?([^\s\;"]+)"?',line)): geneId=rex[1]
if(transcriptId is None): transcriptId=geneId
if(geneId is None): geneId=transcriptId
if(transcriptId is None):
raise Exception(line+" : no transcript ID found")
if(rex.find("(\S+);$",transcriptId)): transcriptId=rex[1]
if(rex.find("(\S+);$",geneId)): geneId=rex[1]
extra=""
for i in range(8,len(fields)): extra+=fields[i]+" "
if(exonBegin>exonEnd): (exonBegin,exonEnd)=(exonEnd,exonBegin)
transcript=transcripts.get(transcriptId,None)
if(not transcript):
transcripts[transcriptId]=transcript= \
Transcript(transcriptId,strand)
transcript.setStopCodons(self.stopCodons)
transcript.readOrder=readOrder
readOrder+=1
transcript.substrate=fields[0]
transcript.source=fields[1]
if(transcriptBeginEnd.get(transcriptId,None) is not None):
(begin,end)=transcriptBeginEnd[transcriptId]
transcript.setBegin(begin)
transcript.setEnd(end)
else:
transcript.setBegin(exonBegin)
transcript.setEnd(exonEnd)
transcript.geneId=geneId
gene=genes.get(geneId,None)
if(gene is None):
genes[geneId]=gene=Gene(); gene.setId(geneId)
transcript.setGene(gene)
exon=Exon(exonBegin,exonEnd,transcript)
exon.extraFields=extra
if(transcript.rawExons is not None):
exon.frame=frame
exon.score=exonScore
exon.type=fields[2]
transcript.rawExons.append(exon)
elif(not transcript.exonOverlapsExon(exon)):
exon.frame=frame
exon.score=exonScore
exon.type=fields[2]
transcript.UTR.append(exon) # OK -- we sort later
gene.addTranscript(transcript)
def loadGFF_UTR(self,fields,line,transcriptBeginEnd,GFF,
transcripts,readOrder,genes):
exonBegin=int(fields[3])-1
exonEnd=int(fields[4])
exonScore=fields[5]
strand=fields[6]
frame=fields[7]
transcriptId=None
rex=Rex()
if(rex.find('transgrp[:=]\s*(\S+)',line)): transcriptId=rex[1]
elif(rex.find('transcript_id[:=]?\s*"?([^\s";]+)"?',line)):
transcriptId=rex[1]
elif(rex.find('Parent=([^;,\s]+)',line)): transcriptId=rex[1]
geneId=None
if(rex.find('genegrp=(\S+)',line)): geneId=rex[1]
elif(rex.find('gene_id[:=]?\s*"?([^\s\;"]+)"?',line)): geneId=rex[1]
if(transcriptId is None): transcriptId=geneId
if(geneId is None): geneId=transcriptId
if(transcriptId is None):
raise Exception(line+" : no transcript ID found")
if(rex.find("(\S+);$",transcriptId)): transcriptId=rex[1]
if(rex.find("(\S+);$",geneId)): geneId=rex[1]
extra=""
for i in range(8,len(fields)): extra+=fields[i]+" "
if(exonBegin>exonEnd): (exonBegin,exonEnd)=(exonEnd,exonBegin)
transcript=transcripts.get(transcriptId,None)
if(not transcript):
transcripts[transcriptId]=transcript= \
Transcript(transcriptId,strand)
transcript.setStopCodons(self.stopCodons)
transcript.readOrder=readOrder
readOrder+=1
transcript.substrate=fields[0]
transcript.source=fields[1]
if(transcriptBeginEnd.find(transcriptId,None) is not None):
(begin,end)=transcriptBeginEnd[transcriptId]
transcript.setBegin(begin)
transcript.setEnd(end)
transcript.geneId=geneId
gene=genes.get(geneId,None)
if(gene is None):
genes[geneId]=gene=Gene(); gene.setId(geneId)
transcript.setGene(gene)
exon=Exon(exonBegin,exonEnd,transcript)
exon.extraFields=extra
if(transcript.rawExons is not None):
exon.frame=frame
exon.score=exonScore
exon.type=fields[2]
transcript.rawExons.append(exon)
elif(not transcript.exonOverlapsExon(exon)):
exon.frame=frame
exon.score=exonScore
exon.type=fields[2]
transcript.UTR.append(exon) # OK -- we sort later
gene.addTranscript(transcript)
def makeExon(self,root):
begin=int(root["begin"])
end=int(root["end"])
exon=Exon(begin,end,None)
exon.strand=root["strand"]
exon.frame=root["frame"]
exon.type=root["type"]
exon.score=root["score"]
exon.substrate=root["substrate"]
extra=root["extra"]
exon.extraFields=""
for key in extra:
exon.extraFields+=key+"="+extra[key]+";"
return exon
def makeExon(self, root):
begin = int(root["begin"])
end = int(root["end"])
exon = Exon(begin, end, None)
exon.strand = root["strand"]
exon.frame = root["frame"]
exon.type = root["type"]
exon.score = root["score"]
exon.substrate = root["substrate"]
extra = root["extra"]
exon.extraFields = ""
for key in extra:
exon.extraFields += key + "=" + extra[key] + ";"
return exon