annotate data_manager/shapeit_ref.py @ 0:4f7004475124 draft default tip

planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
author sanbi-uwc
date Tue, 18 Sep 2018 00:54:49 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
1 #!/usr/bin/env python3
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
2
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
3 from __future__ import division, print_function
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
4 import argparse
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
5 import json
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
6 import os
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
7 import os.path
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
8
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
9
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
10 def _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry):
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
11 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {})
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
12 data_manager_dict['data_tables'][data_table_name] = data_manager_dict['data_tables'].get(data_table_name, [])
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
13 data_manager_dict['data_tables'][data_table_name].append(data_table_entry)
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
14 return data_manager_dict
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
15
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
16
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
17 _add_data_table_entry.__annotations__ = {'data_manager': dict, 'data_table_name': str, 'data_table_entry': dict, 'return': dict}
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
18
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
19
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
20 def assert_prefix_exists(prefix, path, prefix_type):
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
21 prefix_exists = False
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
22 for filename in os.listdir(path):
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
23 if filename.startswith(prefix):
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
24 prefix_exists = True
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
25 break
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
26 else:
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
27 if not prefix_exists:
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
28 exit("Unable to find a file with {} prefix {} in {}".format(prefix_type, prefix, path))
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
29
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
30
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
31 assert_prefix_exists.__annotations__ = {'prefix': str, 'path': str, 'prefix_type': str, 'return': None}
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
32
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
33 if __name__ == '__main__':
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
34 parser = argparse.ArgumentParser(description='Generate a data manager output for SHAPEIT reference data')
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
35 parser.add_argument('key', help='Short key to identify this reference set (no spaces)')
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
36 parser.add_argument('name', help='Description of reference set')
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
37 parser.add_argument('path', help='Filesystem path to directory containing this reference set')
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
38 parser.add_argument('reference_prefix', help='Filename prefix for the reference (.hap / .legend / .sample) files')
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
39 parser.add_argument('map_prefix', help='Filename prefix for map files in this reference set')
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
40 parser.add_argument('sample_prefix', help='Filename prefix for sample file in this reference set')
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
41 parser.add_argument('output_file', type=argparse.FileType('w'), help='JSON file used to write data manager values to')
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
42 args = parser.parse_args()
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
43
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
44 if not os.path.exists(args.path):
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
45 exit("Unable to find specified path {}".format(args.path))
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
46
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
47 assert_prefix_exists(args.reference_prefix, args.path, 'reference')
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
48 assert_prefix_exists(args.map_prefix, args.path, 'map')
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
49 assert_prefix_exists(args.sample_prefix, args.path, 'sample')
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
50
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
51 for column in ('key', 'name', 'path', 'reference_prefix', 'map_prefix', 'sample_prefix'):
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
52 value = getattr(args, column)
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
53 if '\t' in value:
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
54 exit("TAB character found in {} argument".format(column))
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
55
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
56 data_manager_dict = {}
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
57 data_table_entry = dict(value=args.key, name=args.name, path=args.path,
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
58 reference_prefix=args.reference_prefix, map_prefix=args.map_prefix,
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
59 sample_prefix=args.sample_prefix)
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
60 _add_data_table_entry(data_manager_dict, 'shapeit2_ref', data_table_entry)
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
61
4f7004475124 planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit aa91508cbd163b7b005d87824ffa997e62d5add6
sanbi-uwc
parents:
diff changeset
62 args.output_file.write(json.dumps(data_manager_dict, sort_keys=True) + '\n')