def _annotate_vcf_files(all_patterns, known_args, pipeline_args):
# type: (List[str], argparse.Namespace, List[str]) -> str
"""Annotates the VCF files using VEP.
Returns:
The annotated VCF files directory.
"""
options = pipeline_options.PipelineOptions(pipeline_args)
google_cloud_options = options.view_as(pipeline_options.GoogleCloudOptions)
annotate_files_job_name = pipeline_common.generate_unique_name(
_ANNOTATE_FILES_JOB_NAME)
_update_google_cloud_job_name(google_cloud_options, annotate_files_job_name)
with beam.Pipeline(options=options) as p:
_ = (p
| beam.Create(all_patterns)
| 'AnnotateShards' >> beam.ParDo(
annotate_files.AnnotateFile(known_args, pipeline_args)))
if known_args.annotation_fields:
known_args.annotation_fields.append(known_args.vep_info_field)
else:
known_args.annotation_fields = [known_args.vep_info_field]
# TODO(bashir2): The VEP runner by default runs VEP with --allele_number hence
# we turn on this feature here. However, this might be inconsistent with other
# annotation fields that are originally present in input files, if they do not
# have ALLELE_NUM annotation. The fix is to make annotation ALT matching
# smarter to fall back on other matching methods if ALLELE_NUM is not present.
# When this is implemented, we may even consider removing use_allele_num flag
# and always start by checking if ALLELE_NUM is present.
known_args.use_allele_num = True
return vep_runner_util.get_output_pattern(known_args.annotation_output_dir)
def get_output_pattern(self):
# type: () -> str
return vep_runner_util.get_output_pattern(self._output_dir)
