Mercurial > repos > yating-l > multi_fasta_glimmer_hmm
annotate fetch_reference_data.py @ 7:0f39f40ef695 draft
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
author | yating-l |
---|---|
date | Wed, 03 May 2017 15:15:36 -0400 |
parents | |
children |
rev | line source |
---|---|
7
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
1 #!/usr/bin/env python |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
2 # ref: https://galaxyproject.org/admin/tools/data-managers/how-to/define/ |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
3 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
4 import sys |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
5 import os |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
6 import tempfile |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
7 import shutil |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
8 import argparse |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
9 import urllib2 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
10 import tarfile |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
11 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
12 from galaxy.util.json import from_json_string, to_json_string |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
13 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
14 CHUNK_SIZE = 2**20 #1mb |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
15 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
16 def cleanup_before_exit(tmp_dir): |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
17 if tmp_dir and os.path.exists(tmp_dir): |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
18 shutil.rmtree(tmp_dir) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
19 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
20 def stop_err(msg): |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
21 sys.stderr.write(msg) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
22 sys.exit(1) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
23 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
24 def get_reference_id_name(params): |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
25 genome_id = params['param_dict']['genome_id'] |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
26 genome_name = params['param_dict']['genome_name'] |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
27 return genome_id, genome_name |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
28 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
29 def download_from_GlimmerHMM(data_manager_dict, params, target_directory, sequence_id, sequence_name ): |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
30 GlimmerHMM_DOWNLOAD_URL = 'ftp://ccb.jhu.edu/pub/software/glimmerhmm/GlimmerHMM-3.0.4.tar.gz' |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
31 GlimmerHMM_TRAINED_DIR = os.path.join('GlimmerHMM', 'trained_dir', sequence_id) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
32 with tarfile.open('GlimmerHMM-3.0.4.tar', mode='r:*') as tar: |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
33 subdir = [ |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
34 tarinfo for tarinfo in tar.getmembers() |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
35 if sequence_id in tarinfo.name |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
36 ] |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
37 tar.extractall(members=subdir) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
38 glimmerhmm_trained_target_dir = os.path.join(target_directory, sequence_id) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
39 shutil.copytree(GlimmerHMM_TRAINED_DIR, glimmerhmm_trained_target_dir) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
40 data_table_entry = dict(value=sequence_id, name=sequence_name, path=glimmerhmm_trained_target_dir) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
41 _add_data_table_entry(data_manager_dict, data_table_entry) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
42 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
43 cleanup_before_exit(GlimmerHMM_TRAINED_DIR) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
44 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
45 def _add_data_table_entry( data_manager_dict, data_table_entry ): |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
46 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} ) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
47 data_manager_dict['data_tables']['reference_data'] = data_manager_dict['data_tables'].get('reference_data', []) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
48 data_manager_dict['data_tables']['reference_data'].append( data_table_entry ) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
49 return data_manager_dict |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
50 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
51 REFERENCE_SOURCE_TO_DOWNLOAD = dict(glimmerhmm=download_from_GlimmerHMM) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
52 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
53 def main(): |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
54 #Parse Command Line |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
55 parser = argparse.ArgumentParser() |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
56 args = parser.parse_args() |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
57 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
58 filename = args[0] |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
59 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
60 params = from_json_string(open(filename).read()) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
61 target_directory = params['output_data'][0]['extra_files_path'] |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
62 os.mkdir(target_directory) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
63 data_manager_dict = {} |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
64 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
65 sequence_id, sequence_name = get_reference_id_name(params) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
66 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
67 #Fetch the FASTA |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
68 REFERENCE_SOURCE_TO_DOWNLOAD[params['param_dict']['file_path']](data_manager_dict, params, target_directory, sequence_id, sequence_name) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
69 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
70 #save info to json file |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
71 open(filename, 'wb').write(to_json_string(data_manager_dict)) |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
72 |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
73 if __name__ == "__main__": |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
74 main() |
0f39f40ef695
planemo upload for repository https://github.com/remimarenco/multi_fasta_glimmerhmm.git commit 59eb9b27c7382ae4736eb7425615e4b575757378
yating-l
parents:
diff
changeset
|
75 |