def canonicalize(filename, output_name):
with open(output_name + "tmp.vcf", 'w') as canon_file:
# Left-align indels and output variants as constitutent indels
tenkit.log_subprocess.check_call([
'vcfallelicprimitives', '--keep-info', '-t', 'VCFALLELICPRIMITIVE',
filename
],
stdout=canon_file)
with open(output_name + "tmp2.vcf", 'w') as fixed_vcf:
# the reason we are doing this is because vcfallelicprimitives screws up the vcf format in some places in the info fields
# it changes some tags from KEY=.; to KEY=; which is invalid. bcftools fixes this, but we dont actually want to filter anything
# this should do that
tenkit.log_subprocess.check_call(
['bcftools', 'filter', output_name + "tmp.vcf"], stdout=fixed_vcf)
with open(output_name + "tmp3.vcf", 'w') as unphased_file:
vcf_in = tk_io.VariantFileReader(output_name + "tmp2.vcf")
unsupported_genotype_filter = [(
"UNSUPPORTED_GENOTYPE",
"If genotype field contains '.' we assume that this is due to making a single sample vcf from a multiple sample vcf in which this sample does not contain the variant."
)]
tenx = ('TENX', '0', 'Flag', "called by 10X", None, None)
vcf_out = tk_io.VariantFileWriter(
unphased_file,
template_file=open(output_name + "tmp2.vcf"),
new_info_fields=[tenx],
new_filters=unsupported_genotype_filter)
for record in vcf_in.record_getter():
sample = record.samples[0]
unsupported_genotype = False
try:
if len(sample.gt_alleles) > 0:
genotype1 = sample.gt_alleles[0]
if genotype1 == '.':
unsupported_genotype = True
else:
unsupported_genotype = True
if len(sample.gt_alleles) > 1:
genotype2 = sample.gt_alleles[1]
if genotype2 is '.':
unsupported_genotype = True
except:
unsupported_genotype = True
if unsupported_genotype:
record.FILTER = ["UNSUPPORTED_GENOTYPE"]
vcf_out.write_record(record)
tk_tabix.sort_vcf(output_name + "tmp3.vcf", output_name)
def combine_vcfs(output_filename, input_vcf_filenames):
tmp_filename = output_filename + ".tmp"
for (i, fn) in enumerate(input_vcf_filenames):
if i == 0:
args = 'cat ' + fn
subprocess.check_call(args + " > " + tmp_filename, shell=True)
else:
args = 'grep -v "^#" ' + fn
ret = subprocess.call(args + " >> " + tmp_filename, shell=True)
if ret == 2:
raise Exception("grep call failed: " + args)
# Sort and index the files
tk_tabix.sort_vcf(tmp_filename, output_filename)
tk_tabix.index_vcf(output_filename)
os.remove(tmp_filename)
def join(args, outs, chunk_defs, chunk_outs):
# mapping of cluster ID -> VCFs
to_merge = collections.defaultdict(list)
for o, d in zip(chunk_outs, chunk_defs):
to_merge[d.cluster_id].append(o.variant_subset)
# merge each VCF subset for a cluster
merged_vcfs = []
for cluster_id, vcf_list in to_merge.iteritems():
merged_vcf = martian.make_path('{}.vcf'.format(cluster_id))
tk_io.combine_vcfs(merged_vcf, vcf_list)
merged_vcfs.append(merged_vcf + '.gz')
# final merge to make one combined VCF
tmp = martian.make_path('tmp.vcf')
cmd = ['vcf-merge'] + merged_vcfs
with open(tmp, 'w') as outf:
subprocess.check_call(cmd, stdout=outf)
# Sort and index the files
tk_tabix.sort_vcf(tmp, outs.variants.replace('.gz', ''))
tk_tabix.index_vcf(outs.variants.replace('.gz', ''))
os.remove(tmp)
def main_sort_variants(args, outs):
if args.input is None or args.input == [None]:
outs.default = None
else:
outs.coerce_strings()
# List of inputs
sort_filename = outs.default[0:(len(outs.default)-3)]
if type(args.input) == type([]):
files = [f for f in args.input if os.path.isfile(f)]
if len(files) == 0:
outs.defaut = None
return
cat_filename = outs.default[:-3]
tk_io.combine_vcfs(cat_filename, args.input)
# Single input
else:
if not os.path.exists(args.input):
outs.default = None
return
tk_tabix.sort_vcf(args.input, sort_filename)
tk_tabix.index_vcf(sort_filename)