def invtobnd(var):
strands = var.get_info('STRANDS')
strand_dict = dict(x.split(':') for x in strands.split(','))
for o in strand_dict.keys():
if strand_dict[o] == '0':
del (strand_dict[o])
strands = ','.join(['%s:%s' % (o, strand_dict[o]) for o in strand_dict])
var.set_info('STRANDS', strands)
if strands[:2] == '++':
ALT = 'N]' + var.chrom + ':' + str(var.get_info('END')) + ']'
elif strands[:2] == '--':
ALT = '[' + var.chrom + ':' + str(var.get_info('END')) + '[N'
var.set_info('SVTYPE', 'BND')
var.alt = ALT
[tempci, temp95] = [var.get_info('CIPOS'), var.get_info('CIPOS95')]
try:
temppr = var.get_info('PRPOS')
except KeyError:
raise MissingProbabilitiesException('Required tag PRPOS not found.')
var.set_info('CIPOS', var.get_info('CIEND'))
var.set_info('CIEND', tempci)
var.set_info('CIPOS95', var.get_info('CIEND95'))
var.set_info('CIEND95', temp95)
try:
var.set_info('PRPOS', var.get_info('PREND'))
except KeyError:
raise MissingProbabilitiesException('Required tag PREND not found.')
var.set_info('PREND', temppr)
def write_var(var, vcf_out, include_genotypes=False):
v_id = var.var_id
if var.get_info('CIPOS95') != '0,0' or var.get_info('CIEND95') != '0,0':
var.set_info('IMPRECISE', True)
else:
var.set_info('IMPRECISE', False)
if var.get_info('SVTYPE') == 'INV' and ('--:0' in var.get_info('STRANDS')
or '++:0'
in var.get_info('STRANDS')):
invtobnd(var)
if var.alt not in ['<DEL>', '<DUP>', '<INV>', '<INS>']:
var.var_id = str(v_id) + '_1'
var.set_info('EVENT', v_id)
var.set_info('MATEID', str(v_id) + '_2')
var.info.pop('END', None)
var.info.pop('SVLEN', None)
varstring = var.get_var_string(use_cached_gt_string=True)
if not include_genotypes:
varstring = '\t'.join(varstring.split('\t', 10)[:8])
vcf_out.write(varstring + '\n')
new_alt = ''
if var.alt[0] == '[':
new_alt = '[' + var.chrom + ':' + str(var.pos) + '[N'
elif var.alt[0] == ']':
new_alt = 'N[' + var.chrom + ':' + str(var.pos) + '['
elif var.alt[-1] == '[':
new_alt = ']' + var.chrom + ':' + str(var.pos) + ']N'
elif var.alt[-1] == ']':
new_alt = 'N]' + var.chrom + ':' + str(var.pos) + ']'
sep, chrom, pos = parse_bnd_alt_string(var.alt)
var.chrom = chrom
var.pos = int(pos)
var.var_id = str(v_id) + '_2'
var.set_info('MATEID', str(v_id) + '_1')
var.set_info('SECONDARY', True)
var.alt = new_alt
[tempci, temp95] = [var.get_info('CIPOS'), var.get_info('CIPOS95')]
try:
temppr = var.get_info('PRPOS')
except KeyError:
raise MissingProbabilitiesException(
'Required tag PRPOS not found.')
var.set_info('CIPOS', var.get_info('CIEND'))
var.set_info('CIEND', tempci)
var.set_info('CIPOS95', var.get_info('CIEND95'))
var.set_info('CIEND95', temp95)
try:
var.set_info('PRPOS', var.get_info('PREND'))
except KeyError:
raise MissingProbabilitiesException(
'Required tag PREND not found.')
var.set_info('PREND', temppr)
varstring = var.get_var_string(use_cached_gt_string=True)
if not include_genotypes:
varstring = '\t'.join(varstring.split('\t', 10)[:8])
vcf_out.write(varstring + '\n')
else:
varstring = var.get_var_string(use_cached_gt_string=True)
if not include_genotypes:
varstring = '\t'.join(varstring.split('\t', 10)[:8])
vcf_out.write(varstring + '\n')