def DoRec():
global GeneIndexes
AttrDict = gff.GetAttrDict()
CurrGeneIndex = gff.GetIntAttr("gene_index", -1)
if CurrGeneIndex == -1:
gff.WriteRec(sys.stdout)
return
Key = gff.Label + "." + str(CurrGeneIndex)
if Key in GeneIndexes.keys():
NewGeneIndex = GeneIndexes[Key]
else:
NewGeneIndex = len(GeneIndexes)
GeneIndexes[Key] = NewGeneIndex
AttrDict["gene_index"] = NewGeneIndex
gff.SetAttrsFromDict(AttrDict)
gff.WriteRec(sys.stdout)
def DoRec():
global LastGeneIndex
global LastExonStart
global LastLabel
global LastExonEnd
global IntronCounts
global ExonCounts
if gff.Feature != "exon":
return
GeneIndex = gff.GetRequiredIntAttr("gene_index")
Key = gff.Label + "%%%" + str(GeneIndex)
if Key not in ExonCounts.keys():
ExonCounts[Key] = 1
else:
ExonCounts[Key] += 1
# print "ExonCounts[%s] = %u" % (Key, ExonCounts[Key])
Start = gff.Start
End = gff.End
Label = gff.Label
if GeneIndex == LastGeneIndex and gff.Label == LastLabel and LastExonStart != -1:
if LastExonStart != -1:
gff.Label = LastLabel
gff.Start = LastExonEnd + 1
gff.End = Start - 1
gff.Source = "exons2introns"
gff.Feature = "intron"
gff.Frame = "."
gff.Strand = "."
gff.Attrs = "gene_index %u; exons %u-%u,%u-%u;" % (GeneIndex, LastExonStart, LastExonEnd, Start, End)
gff.WriteRec(sys.stdout)
if Key not in IntronCounts.keys():
IntronCounts[Key] = 1
else:
IntronCounts[Key] += 1
# print >> sys.stderr, "IntronCounts[%s] = %u" % (Key, IntronCounts[Key])
LastGeneIndex = GeneIndex
LastExonStart = Start
LastExonEnd = End
LastLabel = Label
def DoRec():
global LastGeneIndex
global ExonStart
global LastLabel
global LastEnd
global Attr
if not IsExonFeature(gff.Feature):
return
GeneIndex = gff.GetRequiredIntAttr("gene_index")
Start = gff.Start
End = gff.End
Feature = gff.Feature
Label = gff.Label
if GeneIndex != LastGeneIndex or gff.Label != LastLabel or gff.Start != LastEnd + 1:
if ExonStart != -1:
gff.Label = LastLabel
gff.Start = ExonStart
gff.End = LastEnd
gff.Source = "cdsutr2exons"
gff.Feature = "exon"
gff.Frame = "."
gff.Strand = "."
gff.Attrs = "gene_index %u; ces %s;" % (LastGeneIndex, Attr)
gff.WriteRec(sys.stdout)
Attr = ""
ExonStart = Start
LastGeneIndex = GeneIndex
LastStart = Start
LastLabel = Label
LastEnd = End
s = "%s:%u-%u" % (Feature, Start, End)
if Attr == "":
Attr = s
else:
Attr += "," + s
def DoRec():
gff.WriteRec(sys.stdout)
GeneHis[GeneIndex] = gff.End
GeneStrands[GeneIndex] = gff.Strand
Hi = -1
GeneLos = {}
GeneHis = {}
GeneStrands = {}
Label = ""
gff.GetRecs(FileName, DoRec)
gff.Source = "gene_lengths"
gff.Feature = "gene"
gff.Score = 0
gff.Frame = "."
TotGeneLength = 0
for GeneIndex in GeneLos.keys():
gff.Start = GeneLos[GeneIndex]
gff.End = GeneHis[GeneIndex]
gff.Strand = GeneStrands[GeneIndex]
gff.Attrs = "gene_index %u;" % GeneIndex
gff.WriteRec(sys.stdout)
TotGeneLength += gff.End - gff.Start + 1
print >> sys.stderr, "Max annot end %10u" % Hi
print >> sys.stderr, "Total gene length %10u" % TotGeneLength
print >> sys.stderr, "Pct %10.1f%%" % (float(TotGeneLength) *
100 / Hi)