Mercurial > repos > sanbi-uwc > data_manager_shapeit_reference
annotate data_manager/shapeit_ref.py @ 8:ec2c5e521533 draft default tip
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit 58664d1dbd0725423a50869ecb376f05e59b6449-dirty
author | sanbi-uwc |
---|---|
date | Mon, 17 Sep 2018 16:02:36 -0400 |
parents | f0c7cab2f547 |
children |
rev | line source |
---|---|
0
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
1 #!/usr/bin/env python3 |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
2 |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
3 from __future__ import division, print_function |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
4 import argparse |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
5 import json |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
6 import os |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
7 import os.path |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
8 |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
9 |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
10 def _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry): |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
11 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {}) |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
12 data_manager_dict['data_tables'][data_table_name] = data_manager_dict['data_tables'].get(data_table_name, []) |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
13 data_manager_dict['data_tables'][data_table_name].append(data_table_entry) |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
14 return data_manager_dict |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
15 |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
16 |
1
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
17 _add_data_table_entry.__annotations__ = {'data_manager': dict, 'data_table_name': str, 'data_table_entry': dict, 'return': dict} |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
18 |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
19 |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
20 def assert_prefix_exists(prefix, path, prefix_type): |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
21 prefix_exists = False |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
22 for filename in os.listdir(path): |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
23 if filename.startswith(prefix): |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
24 prefix_exists = True |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
25 break |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
26 else: |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
27 if not prefix_exists: |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
28 exit("Unable to find a file with {} prefix {} in {}".format(prefix_type, prefix, path)) |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
29 |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
30 |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
31 assert_prefix_exists.__annotations__ = {'prefix': str, 'path': str, 'prefix_type': str, 'return': None} |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
32 |
0
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
33 if __name__ == '__main__': |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
34 parser = argparse.ArgumentParser(description='Generate a data manager output for SHAPEIT reference data') |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
35 parser.add_argument('key', help='Short key to identify this reference set (no spaces)') |
4
094bd0114876
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit 06f9376308b504cd2ab7a4ff0126ae70d73faacc
sanbi-uwc
parents:
3
diff
changeset
|
36 parser.add_argument('name', help='Description of reference set') |
0
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
37 parser.add_argument('path', help='Filesystem path to directory containing this reference set') |
1
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
38 parser.add_argument('reference_prefix', help='Filename prefix for the reference (.hap / .legend / .sample) files') |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
39 parser.add_argument('map_prefix', help='Filename prefix for map files in this reference set') |
5
f0c7cab2f547
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit 1e78071ffc35990e3ea5846a6cc07e194f886cf0
sanbi-uwc
parents:
4
diff
changeset
|
40 parser.add_argument('sample_prefix', help='Filename prefix for sample file in this reference set') |
0
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
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') |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
42 args = parser.parse_args() |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
43 |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
44 if not os.path.exists(args.path): |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
45 exit("Unable to find specified path {}".format(args.path)) |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
46 |
1
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
47 assert_prefix_exists(args.reference_prefix, args.path, 'reference') |
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
48 assert_prefix_exists(args.map_prefix, args.path, 'map') |
5
f0c7cab2f547
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit 1e78071ffc35990e3ea5846a6cc07e194f886cf0
sanbi-uwc
parents:
4
diff
changeset
|
49 assert_prefix_exists(args.sample_prefix, args.path, 'sample') |
0
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
50 |
5
f0c7cab2f547
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit 1e78071ffc35990e3ea5846a6cc07e194f886cf0
sanbi-uwc
parents:
4
diff
changeset
|
51 for column in ('key', 'name', 'path', 'reference_prefix', 'map_prefix', 'sample_prefix'): |
0
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
52 value = getattr(args, column) |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
53 if '\t' in value: |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
54 exit("TAB character found in {} argument".format(column)) |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
55 |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
56 data_manager_dict = {} |
4
094bd0114876
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit 06f9376308b504cd2ab7a4ff0126ae70d73faacc
sanbi-uwc
parents:
3
diff
changeset
|
57 data_table_entry = dict(value=args.key, name=args.name, path=args.path, |
5
f0c7cab2f547
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit 1e78071ffc35990e3ea5846a6cc07e194f886cf0
sanbi-uwc
parents:
4
diff
changeset
|
58 reference_prefix=args.reference_prefix, map_prefix=args.map_prefix, |
f0c7cab2f547
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit 1e78071ffc35990e3ea5846a6cc07e194f886cf0
sanbi-uwc
parents:
4
diff
changeset
|
59 sample_prefix=args.sample_prefix) |
0
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
60 _add_data_table_entry(data_manager_dict, 'shapeit_ref', data_table_entry) |
f86c65c60056
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit c9acc04bcb85873e6eb78b4163d1a7daa2679f22
sanbi-uwc
parents:
diff
changeset
|
61 |
1
52f0767fe68d
planemo upload for repository https://github.com/pvanheus/data_manager_shapeit_reference commit b42d6cc8358088a50bc4463367c740c1e9dc71d8
sanbi-uwc
parents:
0
diff
changeset
|
62 args.output_file.write(json.dumps(data_manager_dict, sort_keys=True) + '\n') |