def __init__(self, call_splt, pat_name, variant_ix=1) :
self.pat_name = pat_name
self.fields = {}
self.fields["GT"] = broad.convertGT( call_splt, variant_ix )
if len(call_splt) > 1 :
(GT,AD,DP,GQ,PL) = call_splt
self.fields["DP"] = int(DP)
self.fields["GQ"] = float(GQ)
(r,m) = [int(t) for t in AD.split(',')]
self.fields["AD_ref"] = r
self.fields["AD_mut"] = m
(aa,ab,bb) = [pow(10, -1*(float(t)/10)) for t in PL.split(',')]
self.fields["PL_AA"] = aa
self.fields["PL_AB"] = ab
self.fields["PL_BB"] = bb
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 )
