def integrator( self, target, splts ) :
if len(splts) != 1 : assert "len isn't right"
for splt in splts :
##TODO generalize this to make it vendor independent, call start column is feature of VCF, not broad???
calls = splt[ broad.CALL_START: ]
base_calls = []
for pat_ix,c in enumerate(calls) :
sc = broad.splitCall(c)
gt = broad.convertGT( sc )
if broad.isMutated( gt ) or broad.noInf( gt ) :
base_calls.append( BaseCall(sc,pat_ix) )
fields = {}
keys = broad.COLUMN_MAP.keys()
for k in keys :
if k == "chrom" :
fields[k] = globes.chromNum( splt[self.indexOf[k]] )
elif k == "info" :
##TODO generalize this to make it vendor independent
dinfo = broad.makeInfoDict( splt[ self.indexOf[k] ] )
fields["AF"] = dinfo["AF"]
elif k == "dbSNP" :
value = splt[self.indexOf[k]]
#when we getFields, 'dbsnp' will be missing and yield null
if value == '.' : pass
#right now just take the first rs number if multiple
elif value.startswith('rs') :
fields[k] = [int(t.strip()[2:]) for t in value.split(';')][0]
else :
print "malformed rs number?", splt
assert False
else :
fields[k] = splt[ self.indexOf[k] ]
#according to: http://www.broadinstitute.org/gsa/wiki/index.php/Understanding_the_Unified_Genotyper's_VCF_files
#ref and alt are always given for the forward strand
fields['strand'] = True
return Variant( fields, base_calls )
def isMutated(self) :
return broad.isMutated( self.GT )
