validation_variants = list()
for variant in ordered_variants:
if 'cosmic_ids' in samples[sample].keys():
cosmic_ids = samples[sample]['cosmic_ids'].split(',')
if variant.cosmic_ids:
for cosmic_id in variant.cosmic_ids:
if cosmic_id in cosmic_ids:
validation_variants.append(variant)
break
if 'rs_ids' in samples[sample].keys():
rs_ids = samples[sample]['rs_ids'].split(',')
if variant.rs_ids:
for rs_id in variant.rs_ids:
if rs_id in rs_ids:
validation_variants.append(variant)
break
if 'amplicons' in samples[sample].keys():
overlapping_amplicons = variant.amplicon_data['amplicon'].split(',')
amplicons = samples[sample]['amplicons'].split(',')
if variant.amplicon_data['amplicon']:
for amplicon in overlapping_amplicons:
if amplicons in amplicons:
validation_variants.append(variant)
break
sys.stdout.write("Retrieved {} total variants\n".format(variants.count()))
sys.stdout.write("Writing {} variants to sample report\n".format(len(validation_variants)))
utils.write_sample_variant_report(args.report, sample, validation_variants, args.variant_callers)
for amplicon in target_amplicons:
coverage_data = SampleCoverage.objects.timeout(None).filter(
SampleCoverage.sample == samples[sample]['sample_name'],
SampleCoverage.amplicon == amplicon,
SampleCoverage.run_id == samples[sample]['run_id'],
SampleCoverage.library_name == samples[sample]['library_name'],
SampleCoverage.program_name == "sambamba"
)
for result in coverage_data:
target_amplicon_coverage[amplicon]['num_reads'] = result.num_reads
target_amplicon_coverage[amplicon]['mean_coverage'] = result.mean_coverage
ordered_variants = variants.order_by('library_name', 'chr', 'pos',
'ref', 'alt').limit(variants.count() + 1000)
filtered_variants = list()
for variant in ordered_variants:
if variant.amplicon_data['amplicon']:
amplicons = variant.amplicon_data['amplicon'].split(',')
for amplicon in amplicons:
if amplicon in target_amplicons:
for caller in callers:
if caller in variant.callers:
filtered_variants.append(variant)
break
sys.stdout.write("Retrieved {} total variants\n".format(variants.count()))
sys.stdout.write("Sending {} variants to reporting\n".format(len(filtered_variants)))
utils.write_sample_variant_report(args.report, sample, filtered_variants, target_amplicon_coverage, callers)
'max_maf': args.max_pop_freq,
'depth': args.depth}
callers = args.variant_callers.split(',')
sys.stdout.write("Processing samples\n")
for sample in samples:
sys.stdout.write("Running Cassandra query for sample {}\n".format(sample))
variants = SampleVariant.objects.timeout(None).filter(SampleVariant.reference_genome == config['genome_version'],
SampleVariant.sample == samples[sample]['sample_name'],
SampleVariant.run_id == samples[sample]['run_id'],
SampleVariant.library_name == samples[sample]['library_name'],
SampleVariant.max_som_aaf >= thresholds['min_saf'],
SampleVariant.max_maf_all <= thresholds['max_maf'],
SampleVariant.max_depth >= thresholds['depth']
).allow_filtering()
ordered_variants = variants.order_by('library_name', 'chr', 'pos',
'ref', 'alt').limit(variants.count() + 1000)
filtered_variants = list()
for variant in ordered_variants:
if variant.amplicon_data['amplicon']:
for caller in callers:
if caller in variant.callers:
filtered_variants.append(variant)
break
sys.stdout.write("Retrieved {} total variants\n".format(variants.count()))
sys.stdout.write("Writing {} variants to sample report\n".format(len(filtered_variants)))
utils.write_sample_variant_report(args.report, sample, filtered_variants, callers)
