type=str)
longislnd_group = main_parser.add_argument_group("LongISLND options")
longislnd_group.add_argument("--longislnd_options",
help="LongISLND options",
default="")
args = main_parser.parse_args()
args.java = utils.get_java(args.java)
utils.JAVA_XMX = utils.JAVA_XMX + args.java_max_mem
makedirs([args.log_dir, args.out_dir])
# Setup logging
FORMAT = '%(levelname)s %(asctime)-15s %(name)-20s %(message)s'
loglevel = get_loglevel(args.loglevel)
if not args.log_to_stderr:
logging.basicConfig(filename=os.path.join(args.log_dir, "varsim.log"),
filemode="w",
level=loglevel,
format=FORMAT)
else:
logging.basicConfig(level=loglevel, format=FORMAT)
simulator = None if args.disable_sim else args.simulator
simulator_opts = ""
if args.simulator == "dwgsim":
simulator_opts = "-e {1},{2} -E {1},{2} -d {3} -s {4} -1 {5} -2 {5} {6}".format(
args.dwgsim_start_e, args.dwgsim_end_e, args.mean_fragment_size,
args.sd_fragment_size, args.read_length, args.dwgsim_options)
elif args.simulator == "art":
art_group.add_argument("--art_options", help="ART command-line options", default="")
pbsim_group = main_parser.add_argument_group("PBSIM options")
pbsim_group.add_argument("--model_qc", metavar="model_qc", help="PBSIM QC model", default=None, type=str)
longislnd_group = main_parser.add_argument_group("LongISLND options")
longislnd_group.add_argument("--longislnd_options", help="LongISLND options", default="")
args = main_parser.parse_args()
utils.JAVA_XMX = utils.JAVA_XMX + args.java_max_mem
makedirs([args.log_dir, args.out_dir])
# Setup logging
FORMAT = '%(levelname)s %(asctime)-15s %(name)-20s %(message)s'
loglevel = get_loglevel(args.loglevel)
if not args.log_to_stderr:
logging.basicConfig(filename=os.path.join(args.log_dir, "varsim.log"), filemode="w", level=loglevel, format=FORMAT)
else:
logging.basicConfig(level=loglevel, format=FORMAT)
simulator = None if args.disable_sim else args.simulator
simulator_opts = ""
if args.simulator == "dwgsim":
simulator_opts = "-e {1},{2} -E {1},{2} -d {3} -s {4} -1 {5} -2 {5} {6}".format(args.dwgsim_start_e, args.dwgsim_end_e, args.mean_fragment_size, args.sd_fragment_size, args.read_length, args.dwgsim_options)
elif args.simulator == "art":
profile_opts = "-1 {} -2 {}".format(args.profile_1, args.profile_2) if (args.profile_1 and args.profile_2) else ""
simulator_opts = "-p -l {} -m {} -s {} {} {}".format(args.read_length, args.mean_fragment_size, args.sd_fragment_size, profile_opts, args.art_options)
elif args.simulator == "longislnd":
simulator_opts = args.longislnd_options
elif args.simulator == "pbsim":
def process(args):
'''
main
:param args:
:return:
'''
# Setup logging
FORMAT = '%(levelname)s %(asctime)-15s %(name)-20s %(message)s'
loglevel = utils.get_loglevel(args.loglevel)
logging.basicConfig(level=loglevel, format=FORMAT)
global LOGGER
LOGGER = logging.getLogger(__name__)
LOGGER.info('running {}'.format(' '.join(sys.argv)))
dup_mode = None
if args.mode == 'first_duplicate':
dup_mode = utils.COMBINE_KEEP_FIRST_DUPLICATE
elif args.mode == 'all_duplicate':
dup_mode = utils.COMBINE_KEEP_ALL_DUPLICATE
elif args.mode == 'no_duplicate':
dup_mode = utils.COMBINE_KEEP_NO_DUPLICATE
else:
raise ValueError
"""
scenarios:
vcf
vcf.gz
vcf.gz + vcf.gz.tbi
"""
input_vcfs = args.vcfs
for i in range(len(args.vcfs)):
current_vcf = args.vcfs[i]
if current_vcf.endswith(".gz") and os.path.isfile(current_vcf +
".tbi"):
input_vcfs[i] = current_vcf
elif current_vcf.endswith(".gz"):
LOGGER.info('indexing {}'.format(current_vcf))
utils.index_vcf_gz(current_vcf)
input_vcfs[i] = current_vcf
else:
LOGGER.info('sort and index {}'.format(current_vcf))
input_vcfs[i] = utils.sort_and_compress(current_vcf,
mode=2,
overwrite=args.overwrite)
output_vcf = args.output_prefix + '.vcf'
if input_vcfs and len(input_vcfs) == 1:
output_vcf = output_vcf + '.gz'
output_vcf_idx = output_vcf + '.tbi'
if (not args.overwrite) and \
(os.path.isfile(output_vcf) or os.path.isfile(output_vcf_idx)):
LOGGER.warn(
'{} or {} exists, use --overwrite otherwise do nothing.'.
format(output_vcf, output_vcf_idx))
else:
shutil.copyfile(input_vcfs[0], output_vcf)
shutil.copyfile(input_vcfs[0], output_vcf_idx)
else:
output_vcf = utils.combine_vcf(output_vcf,
input_vcfs,
duplicate_handling_mode=dup_mode)
LOGGER.info('{} done'.format(output_vcf))
return
def process(args):
'''
main
:param args:
:return:
'''
# Setup logging
FORMAT = '%(levelname)s %(asctime)-15s %(name)-20s %(message)s'
loglevel = utils.get_loglevel(args.loglevel)
if args.log_to_file:
logging.basicConfig(filename=args.log_to_file,
filemode="w",
level=loglevel,
format=FORMAT)
else:
logging.basicConfig(level=loglevel, format=FORMAT)
if len(args.vcfs) > 1:
raise NotImplementedError(
'right now only support one prediction VCF. Quick workaround: src/sort_vcf.sh vcf1 vcf2 > merged.vcf'
)
global LOGGER
LOGGER = logging.getLogger(__name__)
LOGGER.info('working hard ...')
utils.JAVA_XMX = utils.JAVA_XMX + args.java_max_mem
args.out_dir = os.path.abspath(args.out_dir)
args.reference = os.path.abspath(args.reference)
utils.makedirs([args.out_dir])
varsim_prefix = os.path.join(args.out_dir, 'varsim_compare_results')
varsim_comparator = VarSimVCFComparator(
prefix=varsim_prefix,
true_vcf=args.true_vcf,
reference=args.reference,
regions=None,
sample=args.sample,
vcfs=args.vcfs,
exclude_filtered=args.exclude_filtered,
disallow_partial_fp=args.disallow_partial_fp,
match_geno=args.match_geno,
log_to_file=args.log_to_file,
opts=args.vcfcompare_options)
varsim_tp, varsim_fn, varsim_fp = varsim_comparator.get_tp(
), varsim_comparator.get_fn(), varsim_comparator.get_fp()
varsim_tp = utils.sort_and_compress(varsim_tp)
varsim_fn = utils.sort_and_compress(varsim_fn)
varsim_fp = utils.sort_and_compress(varsim_fp)
#run vcfeval
sdf = args.sdf
if not sdf:
LOGGER.info(
"user did not supply SDF-formatted reference, trying to generate one..."
)
sdf = generate_sdf(args.reference, args.log_to_file)
'''for vcfeval
sample column must be present, and not empty
if single-sample vcf, vcfeval doesn't check if samples match in truth and call
in multi-sample vcf, sample name must be specified
right now
'''
vcfeval_prefix = os.path.join(args.out_dir, 'vcfeval_compare_results')
if os.path.exists(vcfeval_prefix):
LOGGER.warn('{0} exists, removing ...'.format(vcfeval_prefix))
shutil.rmtree(vcfeval_prefix)
vcfeval_comparator = RTGVCFComparator(
prefix=vcfeval_prefix,
true_vcf=varsim_fn,
reference=sdf,
regions=None,
sample=args.sample,
vcfs=[varsim_fp],
exclude_filtered=args.exclude_filtered,
match_geno=args.match_geno,
log_to_file=args.log_to_file,
opts=args.vcfeval_options)
vcfeval_tp, vcfeval_tp_predict = vcfeval_comparator.get_tp(
), vcfeval_comparator.get_tp_predict()
augmented_tp, augmented_fn, augmented_fp, augmented_t = merge_results(
outdir=args.out_dir,
varsim_tp=varsim_tp,
varsim_fn=varsim_fn,
vcfeval_tp=vcfeval_tp,
varsim_fp=varsim_fp,
vcfeval_tp_predict=vcfeval_tp_predict)
augmented_tp, augmented_fn, augmented_fp, augmented_t = summarize_results(
os.path.join(args.out_dir, "augmented"),
augmented_tp,
augmented_fn,
augmented_fp,
augmented_t,
var_types=args.var_types,
sv_length=args.sv_length,
regions=args.regions,
bed_either=args.bed_either)
LOGGER.info(
"Variant comparison done.\nTrue positive: {0}\nFalse negative: {1}\nFalse positive: {2}\n"
.format(augmented_tp, augmented_fn, augmented_fp))
def process(args):
'''
main
:param args:
:return:
'''
# Setup logging
FORMAT = '%(levelname)s %(asctime)-15s %(name)-20s %(message)s'
loglevel = utils.get_loglevel(args.loglevel)
if args.log_to_file:
logging.basicConfig(filename=args.log_to_file, filemode="w", level=loglevel, format=FORMAT)
else:
logging.basicConfig(level=loglevel, format=FORMAT)
if len(args.vcfs) > 1:
raise NotImplementedError('right now only support one prediction VCF. Quick workaround: src/sort_vcf.sh vcf1 vcf2 > merged.vcf')
global LOGGER
LOGGER = logging.getLogger(__name__)
LOGGER.info('working hard ...')
utils.JAVA_XMX = utils.JAVA_XMX + args.java_max_mem
args.out_dir = os.path.abspath(args.out_dir)
args.reference = os.path.abspath(args.reference)
utils.makedirs([args.out_dir])
varsim_prefix = os.path.join(args.out_dir, 'varsim_compare_results')
varsim_comparator = VarSimVCFComparator(prefix=varsim_prefix, true_vcf = args.true_vcf, reference = args.reference,
regions = None,
sample = args.sample, vcfs = args.vcfs,
exclude_filtered = args.exclude_filtered,
disallow_partial_fp = args.disallow_partial_fp,
match_geno = args.match_geno, log_to_file= args.log_to_file, opts = args.vcfcompare_options)
varsim_tp, varsim_fn, varsim_fp = varsim_comparator.get_tp(), varsim_comparator.get_fn(), varsim_comparator.get_fp()
varsim_tp = utils.sort_and_compress(varsim_tp)
varsim_fn = utils.sort_and_compress(varsim_fn)
varsim_fp = utils.sort_and_compress(varsim_fp)
#run vcfeval
sdf = args.sdf
if not sdf:
LOGGER.info("user did not supply SDF-formatted reference, trying to generate one...")
sdf = generate_sdf(args.reference, args.log_to_file)
'''for vcfeval
sample column must be present, and not empty
if single-sample vcf, vcfeval doesn't check if samples match in truth and call
in multi-sample vcf, sample name must be specified
right now
'''
vcfeval_prefix = os.path.join(args.out_dir, 'vcfeval_compare_results')
if os.path.exists(vcfeval_prefix):
LOGGER.warn('{0} exists, removing ...'.format(vcfeval_prefix))
shutil.rmtree(vcfeval_prefix)
vcfeval_comparator = RTGVCFComparator(prefix=vcfeval_prefix, true_vcf = varsim_fn, reference = sdf,
regions = None,
sample = args.sample, vcfs = [varsim_fp],
exclude_filtered = args.exclude_filtered,
match_geno = args.match_geno, log_to_file= args.log_to_file,
opts = args.vcfeval_options)
vcfeval_tp, vcfeval_tp_predict = vcfeval_comparator.get_tp(), vcfeval_comparator.get_tp_predict()
augmented_tp, augmented_fn, augmented_fp, augmented_t = merge_results(
outdir = args.out_dir,
varsim_tp = varsim_tp, varsim_fn = varsim_fn,
vcfeval_tp = vcfeval_tp, varsim_fp = varsim_fp, vcfeval_tp_predict = vcfeval_tp_predict)
augmented_tp, augmented_fn, augmented_fp, augmented_t = summarize_results(os.path.join(args.out_dir,"augmented"), augmented_tp, augmented_fn, augmented_fp, augmented_t,
var_types= args.var_types, sv_length= args.sv_length, regions = args.regions, bed_either = args.bed_either)
LOGGER.info("Variant comparison done.\nTrue positive: {0}\nFalse negative: {1}\nFalse positive: {2}\n".
format(augmented_tp, augmented_fn, augmented_fp))