Mercurial > repos > sanbi-uwc > data_manager_shapeit_reference
view data_manager/shapeit_ref.py @ 3:203133a92000 draft
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit 74cb5dd2b3aaed4a5b26840b451b8f2920e4b9fe-dirty
author | sanbi-uwc |
---|---|
date | Thu, 13 Sep 2018 10:03:16 -0400 |
parents | 52f0767fe68d |
children | 094bd0114876 |
line wrap: on
line source
#!/usr/bin/env python3 from __future__ import division, print_function import argparse import json import os import os.path def _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry): data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {}) data_manager_dict['data_tables'][data_table_name] = data_manager_dict['data_tables'].get(data_table_name, []) data_manager_dict['data_tables'][data_table_name].append(data_table_entry) return data_manager_dict _add_data_table_entry.__annotations__ = {'data_manager': dict, 'data_table_name': str, 'data_table_entry': dict, 'return': dict} def assert_prefix_exists(prefix, path, prefix_type): prefix_exists = False for filename in os.listdir(path): if filename.startswith(prefix): prefix_exists = True break else: if not prefix_exists: exit("Unable to find a file with {} prefix {} in {}".format(prefix_type, prefix, path)) assert_prefix_exists.__annotations__ = {'prefix': str, 'path': str, 'prefix_type': str, 'return': None} if __name__ == '__main__': parser = argparse.ArgumentParser(description='Generate a data manager output for SHAPEIT reference data') parser.add_argument('key', help='Short key to identify this reference set (no spaces)') parser.add_argument('description', help='Description of reference set') parser.add_argument('path', help='Filesystem path to directory containing this reference set') parser.add_argument('reference_prefix', help='Filename prefix for the reference (.hap / .legend / .sample) files') parser.add_argument('map_prefix', help='Filename prefix for map files in this reference set') parser.add_argument('output_file', type=argparse.FileType('w'), help='JSON file used to write data manager values to') args = parser.parse_args() if not os.path.exists(args.path): exit("Unable to find specified path {}".format(args.path)) assert_prefix_exists(args.reference_prefix, args.path, 'reference') assert_prefix_exists(args.map_prefix, args.path, 'map') for column in ('key', 'description', 'path', 'reference_prefix', 'map_prefix'): value = getattr(args, column) if '\t' in value: exit("TAB character found in {} argument".format(column)) data_manager_dict = {} data_table_entry = dict(value=args.key, description=args.description, path=args.path, reference_prefix=args.reference_prefix, map_prefix=args.map_prefix) _add_data_table_entry(data_manager_dict, 'shapeit_ref', data_table_entry) args.output_file.write(json.dumps(data_manager_dict, sort_keys=True) + '\n')