def make_paired_bnd_records(record, ref_fasta):
chr2 = record.info['CHR2']
if 'END2' in record.info.keys():
end2 = record.info['END2']
elif record.contig == record.info['CHR2']:
end2 = record.pos + record.info['SVLEN']
orig_id = record.id
new_id = orig_id + "_M1"
mate_id = orig_id + "_M2"
strands = record.info['STRANDS']
alt1 = svu.make_bnd_alt(chr2, end2, strands, ref_base=record.ref)
record.alts = (alt1, )
record.id = new_id
record.info['MATEID'] = mate_id
record.info.pop('CHR2')
if 'END2' in record.info.keys():
record.info.pop('END2')
record.info.pop('SVLEN')
mate_record = record.copy()
mate_record.id = mate_id
mate_record.contig = chr2
mate_record.pos = int(end2)
mate_record.ref = get_ref_base(chr2, int(end2), ref_fasta)
mate_record.info['MATEID'] = new_id
# strands are reversed for the mate BND
mate_strands = strands[::-1]
mate_alt = svu.make_bnd_alt(record.contig, record.pos, mate_strands, ref_base=mate_record.ref)
mate_record.alts = (mate_alt, )
return record, mate_record
def standardize_alts(self, std_rec, raw_rec):
"""
Standardize ALT field.
Default behavior is to standardize BND alt to VCF spec and leave
other SVTYPE alts untouched.
"""
# Standardize tloc ALT after SVTYPE and CHR2/END are standardized
if std_rec.info['SVTYPE'] == 'BND':
alt = make_bnd_alt(std_rec.info['CHR2'], std_rec.stop,
std_rec.info['STRANDS'])
stop = std_rec.stop
std_rec.alts = (alt, )
std_rec.stop = stop
return std_rec
def make_reciprocal_translocation_bnds(record, ref_fasta):
chr1 = record.contig
chr1_pos1 = record.pos
chr1_pos1_ref = get_ref_base(chr1, chr1_pos1, ref_fasta)
chr1_pos2 = chr1_pos1 + 1
chr1_pos2_ref = get_ref_base(chr1, chr1_pos2, ref_fasta)
chr2 = record.info['CHR2']
if 'END2' in record.info.keys():
chr2_pos1 = int(record.info['END2'])
else:
chr2_pos1 = record.stop
chr2_pos1_ref = get_ref_base(chr2, chr2_pos1, ref_fasta)
chr2_pos2 = chr2_pos1 + 1
chr2_pos2_ref = get_ref_base(chr2, chr2_pos2, ref_fasta)
orig_id = record.id
m1_id = orig_id + "_M1" # M1: chr1P->chr2 on chr1
m2_id = orig_id + "_M2" # M2: chr1P->chr2 on chr2
m3_id = orig_id + "_M3" # M3: chr1Q->chr2 on chr1
m4_id = orig_id + "_M4" # M3: chr1Q->chr2 on chr2
event_id = orig_id
if record.info['CPX_TYPE'] == "CTX_PQ/QP":
strands_m1 = '+-'
strands_m2 = '-+'
strands_m3 = '-+'
strands_m4 = '+-'
m1_pos1 = chr1_pos1
m1_ref = chr1_pos1_ref
m1_pos2 = chr2_pos2
m2_pos1 = chr2_pos2
m2_ref = chr2_pos2_ref
m2_pos2 = chr1_pos1
m3_pos1 = chr1_pos2
m3_ref = chr1_pos2_ref
m3_pos2 = chr2_pos2
m4_pos1 = chr2_pos1
m4_ref = chr1_pos1_ref
m4_pos2 = chr1_pos1
elif record.info['CPX_TYPE'] == "CTX_PP/QQ":
strands_m1 = '++'
strands_m2 = '++'
strands_m3 = '--'
strands_m4 = '--'
m1_pos1 = chr1_pos1
m1_ref = chr1_pos1_ref
m1_pos2 = chr2_pos1
m2_pos1 = chr2_pos1
m2_ref = chr2_pos1_ref
m2_pos2 = chr1_pos1
m3_pos1 = chr1_pos2
m3_ref = chr1_pos2_ref
m3_pos2 = chr2_pos2
m4_pos1 = chr2_pos2
m4_ref = chr2_pos2_ref
m4_pos2 = chr1_pos2
record.info['EVENT'] = event_id
record.info['SVTYPE'] = 'BND'
record.info.pop('CHR2')
record.info.pop('SVLEN')
m1 = record.copy()
m1.id = m1_id
m1.contig = chr1
m1.pos = m1_pos1
m1.stop = m1_pos1 + 1
m1.ref = m1_ref
m1.alts = (svu.make_bnd_alt(chr2, m1_pos2, strands_m1, ref_base=m1.ref), )
m1.info['MATEID'] = m2_id
m2 = record.copy()
m2.id = m2_id
m2.contig = chr2
m2.pos = m2_pos1
m2.stop = m2_pos1 + 1
m2.ref = m2_ref
m2.alts = (svu.make_bnd_alt(chr1, m2_pos2, strands_m2, ref_base=m2.ref), )
m2.info['MATEID'] = m2_id
m3 = record.copy()
m3.id = m3_id
m3.contig = chr1
m3.pos = m3_pos1
m3.stop = m3_pos1 + 1
m3.ref = m3_ref
m3.alts = (svu.make_bnd_alt(chr2, m3_pos2, strands_m3, ref_base=m3.ref), )
m3.info['MATEID'] = m4_id
m4 = record.copy()
m4.id = m4_id
m4.contig = chr2
m4.pos = m4_pos1
m4.stop = m4_pos1 + 1
m4.ref = m4_ref
m4.alts = (svu.make_bnd_alt(chr1, m4_pos2, strands_m4, ref_base=m4.ref), )
m4.info['MATEID'] = m3_id
return m1, m2, m3, m4