def _batch_gvcfs(data, region, vrn_files, ref_file, out_file=None):
"""Perform batching of gVCF files if above recommended input count.
"""
if out_file is None:
out_file = vrn_files[0]
max_batch = int(dd.get_joint_group_size(data))
if len(vrn_files) > max_batch:
out = []
# group to get below the maximum batch size, using 200 as the baseline
for i, batch_vrn_files in enumerate(tz.partition_all(max(max_batch, 200), vrn_files)):
base, ext = utils.splitext_plus(out_file)
batch_out_file = "%s-b%s%s" % (base, i, ext)
out.append(_run_combine_gvcfs(batch_vrn_files, region, ref_file, batch_out_file, data))
return _batch_gvcfs(data, region, out, ref_file)
else:
return vrn_files
def _batch_gvcfs(data, region, vrn_files, ref_file, out_file=None):
"""Perform batching of gVCF files if above recommended input count.
"""
if out_file is None:
out_file = vrn_files[0]
# group to get below the maximum batch size, using 200 as the baseline
max_batch = int(dd.get_joint_group_size(data))
if len(vrn_files) > max_batch:
out = []
num_batches = int(math.ceil(float(len(vrn_files)) / max_batch))
for i, batch_vrn_files in enumerate(tz.partition_all(num_batches, vrn_files)):
base, ext = utils.splitext_plus(out_file)
batch_out_file = "%s-b%s%s" % (base, i, ext)
out.append(run_combine_gvcfs(batch_vrn_files, region, ref_file, batch_out_file, data))
return _batch_gvcfs(data, region, out, ref_file)
else:
return vrn_files