Mercurial > repos > sanbi-uwc > data_manager_novocraft_index_builder

annotate data_manager/novocraft_index_builder.py @ 18:2b89ba1c0057 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit c8c46a5600bb091d701b8cf78f80a50c6b6812f4
author sanbi-uwc
date Tue, 21 Jun 2016 10:12:02 -0400
parents db293ee25be5
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
1 #!/usr/bin/env python
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
2 # Z. Mashologu (SANBI-UWC)
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
3 # import dict as dict
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
4 from __future__ import print_function
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
5 import os
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
6 import sys
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
7 import logging
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
8 import argparse
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
9 import shlex
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
10 from subprocess import check_call, CalledProcessError
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
11
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
12 log = logging.getLogger(__name__)
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
13
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
14 from json import loads, dumps
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
15
6
8dd4b25716d1 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 05004076adb510abc08f6c1e28ba22b2c1a29534
sanbi-uwc
parents: 5
diff changeset
16 DEFAULT_DATA_TABLE_NAME = "novocraft_index"
4
c276a826fc4b planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 1e4e16c747ca6ef261d3307f47a09ff1d49756a1
sanbi-uwc
parents: 0
diff changeset
17
18
2b89ba1c0057 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit c8c46a5600bb091d701b8cf78f80a50c6b6812f4
sanbi-uwc
parents: 17
diff changeset
18 def get_dbkey_id_name(params):
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
19 # TODO: ensure sequence_id is unique and does not already appear in location file
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
20 sequence_id = params['param_dict']['sequence_id']
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
21 sequence_name = params['param_dict']['sequence_name']
18
2b89ba1c0057 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit c8c46a5600bb091d701b8cf78f80a50c6b6812f4
sanbi-uwc
parents: 17
diff changeset
22 sequence_desc = params['param_dict']['sequence_desc']
2b89ba1c0057 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit c8c46a5600bb091d701b8cf78f80a50c6b6812f4
sanbi-uwc
parents: 17
diff changeset
23 if not sequence_desc:
2b89ba1c0057 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit c8c46a5600bb091d701b8cf78f80a50c6b6812f4
sanbi-uwc
parents: 17
diff changeset
24 sequence_desc = sequence_name
2b89ba1c0057 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit c8c46a5600bb091d701b8cf78f80a50c6b6812f4
sanbi-uwc
parents: 17
diff changeset
25 return sequence_id, sequence_name, sequence_desc
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
26
17
db293ee25be5 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 0745f158bbb0d5d190cc4503157d2d3092ab8cc5
sanbi-uwc
parents: 14
diff changeset
27 def _make_novocraft_index(data_manager_dict, fasta_filename, target_directory, sequence_id, sequence_name, data_table_name=DEFAULT_DATA_TABLE_NAME):
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
28 if os.path.exists(target_directory) and not os.path.isdir(target_directory):
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
29 print("Output directory path already exists but is not a directory: {}".format(target_directory),
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
30 file=sys.stderr)
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
31 elif not os.path.exists(target_directory):
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
32 os.mkdir(target_directory)
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
33
12
283eec05d1de planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit e14dbfe322e3787d1ec9b7467a3677b101784227
sanbi-uwc
parents: 11
diff changeset
34 nix_file = sequence_id + ".nix"
283eec05d1de planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit e14dbfe322e3787d1ec9b7467a3677b101784227
sanbi-uwc
parents: 11
diff changeset
35 index_filename = os.path.join(target_directory, nix_file)
11
a9b725202132 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 9ea1c9c5fdf0ed65170469f809b81ba9e82f91d3
sanbi-uwc
parents: 10
diff changeset
36 cmdline_str = 'novoindex {} {}'.format(index_filename, fasta_filename)
a9b725202132 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 9ea1c9c5fdf0ed65170469f809b81ba9e82f91d3
sanbi-uwc
parents: 10
diff changeset
37 cmdline = shlex.split(cmdline_str)
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
38
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
39 try:
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
40 check_call(cmdline)
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
41 except CalledProcessError:
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
42 print("Error building RNA STAR index", file=sys.stderr)
4
c276a826fc4b planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 1e4e16c747ca6ef261d3307f47a09ff1d49756a1
sanbi-uwc
parents: 0
diff changeset
43
17
db293ee25be5 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 0745f158bbb0d5d190cc4503157d2d3092ab8cc5
sanbi-uwc
parents: 14
diff changeset
44 data_table_entry = dict( value=sequence_id, dbkey=sequence_id, name=sequence_name, path=index_filename )
4
c276a826fc4b planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 1e4e16c747ca6ef261d3307f47a09ff1d49756a1
sanbi-uwc
parents: 0
diff changeset
45 _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry )
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
46
4
c276a826fc4b planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 1e4e16c747ca6ef261d3307f47a09ff1d49756a1
sanbi-uwc
parents: 0
diff changeset
47 def _add_data_table_entry( data_manager_dict, data_table_name, data_table_entry ):
c276a826fc4b planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 1e4e16c747ca6ef261d3307f47a09ff1d49756a1
sanbi-uwc
parents: 0
diff changeset
48 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
c276a826fc4b planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 1e4e16c747ca6ef261d3307f47a09ff1d49756a1
sanbi-uwc
parents: 0
diff changeset
49 data_manager_dict['data_tables'][ data_table_name ] = data_manager_dict['data_tables'].get( data_table_name, [] )
c276a826fc4b planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 1e4e16c747ca6ef261d3307f47a09ff1d49756a1
sanbi-uwc
parents: 0
diff changeset
50 data_manager_dict['data_tables'][ data_table_name ].append( data_table_entry )
c276a826fc4b planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 1e4e16c747ca6ef261d3307f47a09ff1d49756a1
sanbi-uwc
parents: 0
diff changeset
51 return data_manager_dict
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
52
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
53 def main():
7
d04a59b76845 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 03f281362d5cf7c4d2a813203b83f5d12eb8ca46
sanbi-uwc
parents: 6
diff changeset
54 parser = argparse.ArgumentParser(description="Generate Novo-craft genome index and JSON describing this")
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
55 parser.add_argument('output_filename')
18
2b89ba1c0057 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit c8c46a5600bb091d701b8cf78f80a50c6b6812f4
sanbi-uwc
parents: 17
diff changeset
56 parser.add_argument('--input_filename')
6
8dd4b25716d1 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 05004076adb510abc08f6c1e28ba22b2c1a29534
sanbi-uwc
parents: 5
diff changeset
57 parser.add_argument('--data_table_name', default='novocraft_index')
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
58 args = parser.parse_args()
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
59
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
60 filename = args.output_filename
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
61
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
62 params = loads(open(filename).read())
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
63 target_directory = params['output_data'][0]['extra_files_path']
14
d053e7b179b5 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 2117346f2705cc3990219e5e99d7feb3f5828b9b
sanbi-uwc
parents: 12
diff changeset
64 os.makedirs(target_directory)
4
c276a826fc4b planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 1e4e16c747ca6ef261d3307f47a09ff1d49756a1
sanbi-uwc
parents: 0
diff changeset
65 data_manager_dict = {}
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
66
18
2b89ba1c0057 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit c8c46a5600bb091d701b8cf78f80a50c6b6812f4
sanbi-uwc
parents: 17
diff changeset
67 sequence_id, sequence_name, sequence_desc = get_dbkey_id_name(params)
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
68
18
2b89ba1c0057 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit c8c46a5600bb091d701b8cf78f80a50c6b6812f4
sanbi-uwc
parents: 17
diff changeset
69 #Make novocraft index
2b89ba1c0057 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit c8c46a5600bb091d701b8cf78f80a50c6b6812f4
sanbi-uwc
parents: 17
diff changeset
70 _make_novocraft_index(data_manager_dict, args.input_filename, target_directory, sequence_id, sequence_name, args.data_table_name or DEFAULT_DATA_TABLE_NAME )
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
71
4
c276a826fc4b planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 1e4e16c747ca6ef261d3307f47a09ff1d49756a1
sanbi-uwc
parents: 0
diff changeset
72 open(filename, 'wb').write(dumps( data_manager_dict ))
0
6cddc1a6e282 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit 3a0b36e3ee66f35f7e2d5e1220cd883a9d528fac
sanbi-uwc
parents:
diff changeset
73
18
2b89ba1c0057 planemo upload for repository https://github.com/zipho/data_manager_novocraft_index_builder commit c8c46a5600bb091d701b8cf78f80a50c6b6812f4
sanbi-uwc
parents: 17
diff changeset
74 if __name__ == "__main__": main()