def main(argv):
parser = argparse.ArgumentParser(
description=__doc__,
prog='svtest merged-depth',
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument('test_bed', type=str)
parser.add_argument('contig_list', type=str)
parser.add_argument('type', type=str)
parser.add_argument('--baseline-bed',
type=str,
default=None,
help="Baseline bed file to evaluate against")
parser.add_argument(
'--test-hits',
type=str,
help=
"List of test record ids that overlap baseline set (required if using --baseline-bed)"
)
parser.add_argument(
'--baseline-hits',
type=str,
help=
"List of baseline record ids that overlap test set (required if using --baseline-bed)"
)
# Print help if no arguments specified
if len(argv) == 0:
parser.print_help()
sys.exit(1)
args = parser.parse_args(argv)
if (bool(args.baseline_bed) ^ bool(args.test_hits)) or \
(bool(args.baseline_bed) ^ bool(args.baseline_hits)) or \
(bool(args.test_hits) ^ bool(args.baseline_hits)):
raise ValueError(
"Inconsistent arguments specified: --baseline-bed, --test-hits, and --baseline-hits must be specified together."
)
contigs = iou.read_contig_list(args.contig_list)
# Read file
with gzip.open(args.test_bed, mode='rb') as ftest:
if args.baseline_bed is None:
metrics = get_metrics(ftest, None, contigs, args.type,
args.test_hits, args.baseline_hits)
else:
with gzip.open(args.baseline_bed, mode='rb') as fbase:
metrics = get_metrics(ftest, fbase, contigs, args.type,
args.test_hits, args.baseline_hits)
# Write metrics
write_metrics(metrics)
def main(argv):
parser = argparse.ArgumentParser(
description=__doc__,
prog='svtest metrics-file',
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument('metrics_file', type=str)
parser.add_argument('contig_list', type=str)
parser.add_argument('--common', action='store_true')
# Print help if no arguments specified
if len(argv) == 0:
parser.print_help()
sys.exit(1)
args = parser.parse_args(argv)
contigs = iou.read_contig_list(args.contig_list)
tu.test_is_not_empty(contigs, "contigs")
# Read file
df = pd.read_csv(args.metrics_file, sep='\t')
metrics = get_metrics(df, contigs, args.common)
# Write metrics
write_metrics(metrics)
def main(argv):
parser = argparse.ArgumentParser(
description=__doc__,
prog='svtest vcf',
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument('test_vcf', type=str)
parser.add_argument('contig_list', type=str)
parser.add_argument('sample_list', type=str)
parser.add_argument(
'types',
type=str,
help='Comma-delimited list of variant types (case-sensitive)')
parser.add_argument('metric_prefix', type=str)
parser.add_argument(
'--baseline-vcf',
type=str,
help='Baseline vcf to provide evaluation metrics against')
parser.add_argument(
'--baseline-bed',
type=str,
help=
'Baseline bed file to provide evaluation metrics against. Must have header beginning with "'
+ BED_FILE_HEADER_CHAR + '" and the following columns: "' +
'", "'.join([
BED_FILE_CHROM_COL, BED_FILE_START_COL, BED_FILE_END_COL,
BED_FILE_SVTYPE_COL
]) + '"')
parser.add_argument(
'--min-reciprocal-overlap',
type=float,
default=0.5,
help='Minimum reciprocal overlap for validation metrics [0.5]')
parser.add_argument('--padding',
type=int,
default=50,
help='Interval padding for validation metrics [50]')
parser.add_argument(
'--max-warnings',
type=int,
default=50,
help='Maximum number of records to print warnings for [50]')
parser.add_argument('--fp-file',
type=str,
default=None,
help='Write false positives to file')
parser.add_argument('--fn-file',
type=str,
default=None,
help='Write false negatives to file')
parser.add_argument('--fp-pass-file',
type=str,
default=None,
help='Write PASS false positives to file')
parser.add_argument('--fn-pass-file',
type=str,
default=None,
help='Write PASS false negatives to file')
# Print help if no arguments specified
if len(argv) == 0:
parser.print_help()
sys.exit(1)
args = parser.parse_args(argv)
if (args.baseline_vcf is None
and args.baseline_bed is None) and (args.fp_file is not None
or args.fn_file is not None):
raise ValueError(
"FP and FN files cannot be generated if --baseline-vcf and --baseline-bed aren't specified"
)
if args.baseline_vcf is not None and args.baseline_bed is not None:
raise ValueError(
"Cannot specify both --baseline-vcf and --baseline-bed")
types_list = args.types.split(',')
contigs = iou.read_contig_list(args.contig_list)
samples = iou.read_samples_list(args.sample_list)
metrics, fp_intervals, fn_intervals, fp_intervals_pass, fn_intervals_pass = get_metrics(
args.test_vcf, args.baseline_vcf, args.baseline_bed, contigs,
types_list, args.min_reciprocal_overlap, args.padding, samples,
args.metric_prefix, args.max_warnings)
# Write metrics
write_metrics(metrics)
if args.fp_file is not None and fp_intervals is not None:
write_intervals(args.fp_file, fp_intervals)
if args.fn_file is not None and fn_intervals is not None:
write_intervals(args.fn_file, fn_intervals)
if args.fp_pass_file is not None and fp_intervals_pass is not None:
write_intervals(args.fp_pass_file, fp_intervals_pass)
if args.fn_pass_file is not None and fn_intervals_pass is not None:
write_intervals(args.fn_pass_file, fn_intervals_pass)
def main(argv):
parser = argparse.ArgumentParser(
description=__doc__,
prog='svtest vcf',
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument('test_vcf', type=str)
parser.add_argument('contig_list', type=str)
parser.add_argument('sample_list', type=str)
parser.add_argument(
'types',
type=str,
help='Comma-delimited list of variant types (case-sensitive)')
parser.add_argument('metric_prefix', type=str)
parser.add_argument(
'--baseline-vcf',
type=str,
help='Baseline vcf to provide evaluation metrics against')
parser.add_argument(
'--min-reciprocal-overlap',
type=float,
default=0.5,
help='Minimum reciprocal overlap for validation metrics [0.5]')
parser.add_argument('--padding',
type=int,
default=50,
help='Interval padding for validation metrics [50]')
parser.add_argument(
'--max-warnings',
type=int,
default=50,
help='Maximum number of records to print warnings for [50]')
parser.add_argument('--fp-file',
type=str,
default=None,
help='Write false positives to file')
parser.add_argument('--fn-file',
type=str,
default=None,
help='Write false negatives to file')
# Print help if no arguments specified
if len(argv) == 0:
parser.print_help()
sys.exit(1)
args = parser.parse_args(argv)
if args.baseline_vcf is None and (args.fp_file is not None
or args.fn_file is not None):
raise ValueError(
"FP and FN files cannot be generated if --baseline-vcf isn't specified"
)
types_list = args.types.split(',')
contigs = iou.read_contig_list(args.contig_list)
samples = iou.read_samples_list(args.sample_list)
metrics, fp_intervals, fn_intervals = get_metrics(
args.test_vcf, args.baseline_vcf, contigs, types_list,
args.min_reciprocal_overlap, args.padding, samples, args.metric_prefix,
args.max_warnings)
# Write metrics
write_metrics(metrics)
if args.fp_file is not None and fp_intervals is not None:
write_intervals(args.fp_file, fp_intervals)
if args.fn_file is not None and fn_intervals is not None:
write_intervals(args.fn_file, fn_intervals)