annotate data_manager/data_manager_fetch_motifs.py @ 6:aa0d1b185070 draft

Uploaded
author jeremyjliu
date Wed, 28 Jan 2015 23:39:30 -0500
parents 6621a6ac8bb4
children b4c2836d2e4e
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
1 #!/usr/bin/env python
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
2 #Dan Blankenberg
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
3
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
4 import sys
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
5 import os
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
6 import tempfile
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
7 import shutil
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
8 import optparse
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
9 import urllib2
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
10 #import uuid
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
11 from ftplib import FTP
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
12 import tarfile
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
13 import zipfile
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
14 import gzip
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
15 import bz2
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
16
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
17 from galaxy.util.json import from_json_string, to_json_string
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
18
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
19 CHUNK_SIZE = 2**20 #1mb
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
20
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
21 def download_motif_databases( data_manager_dict, params, target_directory, motif_db ):
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
22
6
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
23 if motif_db == "pouya":
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
24 BGZ = ['COMPBIO URL HERE',
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
25 "pouya_motifs.bed.bgz", "pouya_bgz", "Pouya Encode Motifs (hg19) BGZ"]
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
26 TBI = ['COMPBIO URL HERE',
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
27 "pouya_motifs.bed.bgz.tbi", "pouya_tbi", "Pouya Encode Motifs (hg19) TBI"]
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
28 elif motif_db == "jaspar":
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
29 BGZ = ['COMPBIO URL HERE',
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
30 "jaspar_jolma_motifs.bed.bgz", "jaspar_bgz", "Jaspar and Jolma Motifs (hg19) BGZ"]
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
31 TBI = ['COMPBIO URL HERE',
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
32 "jaspar_jolma_motifs.bed.bgz.tbi", "jaspar_tbi", "Jaspar and Jolma Motifs (hg19) TBI"]
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
33 elif motif_db == "mouse":
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
34 BGZ = ['COMPBIO URL HERE',
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
35 "mouse_motifs.bed.bgz", "mouse_bgz", "Mouse Motifs (mm9) BGZ"]
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
36 TBI = ['COMPBIO URL HERE',
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
37 "mouse_motifs.bed.bgz.tbi", "mouse_tbi", "Mouse Motifs (mm9) TBI"]
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
38 else:
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
39 BGZ = ['http://gehlenborg.com/wp-content/uploads/motif/pouya_test_motifs.bed.bgz',
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
40 "pouya_test_motifs.bed.bgz", "test_bgz", "Test Pouya Subset (hg19) BGZ"]
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
41 TBI = ['http://gehlenborg.com/wp-content/uploads/motif/pouya_test_motifs.bed.bgz.tbi',
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
42 "pouya_test_motifs.bed.bgz.tbi", "test_tbi", "Test Pouya Subset (hg19) TBI"]
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
43
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
44 bgz_reader = urllib2.urlopen( BGZ[0] )
4
75d825e1b00d Uploaded
jeremyjliu
parents: 2
diff changeset
45 bgz_data_table_entry = _stream_fasta_to_file( bgz_reader, target_directory, params,
6
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
46 BGZ[1], BGZ[2], BGZ[3] )
4
75d825e1b00d Uploaded
jeremyjliu
parents: 2
diff changeset
47 _add_data_table_entry( data_manager_dict, 'motif_databases', bgz_data_table_entry )
75d825e1b00d Uploaded
jeremyjliu
parents: 2
diff changeset
48
6
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
49 tbi_reader = urllib2.urlopen( TBI[0] )
4
75d825e1b00d Uploaded
jeremyjliu
parents: 2
diff changeset
50 tbi_data_table_entry = _stream_fasta_to_file( tbi_reader, target_directory, params,
6
aa0d1b185070 Uploaded
jeremyjliu
parents: 5
diff changeset
51 TBI[1], TBI[2], TBI[3] )
4
75d825e1b00d Uploaded
jeremyjliu
parents: 2
diff changeset
52 _add_data_table_entry( data_manager_dict, 'motif_databases', tbi_data_table_entry )
0
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
53
2
d5faf2b51b07 Uploaded
jeremyjliu
parents: 1
diff changeset
54 def _add_data_table_entry( data_manager_dict, data_table, data_table_entry ):
0
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
55 data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
2
d5faf2b51b07 Uploaded
jeremyjliu
parents: 1
diff changeset
56 data_manager_dict['data_tables'][data_table] = data_manager_dict['data_tables'].get( data_table, [] )
d5faf2b51b07 Uploaded
jeremyjliu
parents: 1
diff changeset
57 data_manager_dict['data_tables'][data_table].append( data_table_entry )
0
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
58 return data_manager_dict
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
59
5
6621a6ac8bb4 Uploaded
jeremyjliu
parents: 4
diff changeset
60 def _stream_fasta_to_file( fasta_stream, target_directory, params,
6621a6ac8bb4 Uploaded
jeremyjliu
parents: 4
diff changeset
61 fasta_base_filename, value, name, close_stream=True ):
0
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
62 fasta_filename = os.path.join( target_directory, fasta_base_filename )
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
63 fasta_writer = open( fasta_filename, 'wb+' )
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
64
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
65 while True:
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
66 buffer = fasta_stream.read(CHUNK_SIZE)
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
67 if not buffer:
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
68 break
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
69
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
70 fasta_writer.write(buffer)
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
71
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
72 fasta_stream.close()
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
73 fasta_writer.close()
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
74
4
75d825e1b00d Uploaded
jeremyjliu
parents: 2
diff changeset
75 return dict( value=value, name=name, path=fasta_base_filename )
0
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
76
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
77 def main():
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
78 #Parse Command Line
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
79 parser = optparse.OptionParser()
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
80 parser.add_option( '-m', '--motif_db', dest='motif_db', action='store', type="string", default=None, help='motif_db' )
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
81 (options, args) = parser.parse_args()
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
82
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
83 filename = args[0]
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
84
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
85 params = from_json_string( open( filename ).read() )
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
86 target_directory = params[ 'output_data' ][0]['extra_files_path']
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
87 os.mkdir( target_directory )
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
88 data_manager_dict = {}
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
89
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
90 #Fetch the Motif Database
1
2ed33f9b9a47 Uploaded
jeremyjliu
parents: 0
diff changeset
91 download_motif_databases( data_manager_dict, params, target_directory, options.motif_db )
0
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
92
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
93 #save info to json file
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
94 open( filename, 'wb' ).write( to_json_string( data_manager_dict ) )
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
95
ba049ce65693 Initial upload
jeremyjliu
parents:
diff changeset
96 if __name__ == "__main__": main()