Mercurial > repos > pjbriggs > ceas
annotate data_manager/data_manager_ceas_fetch_annotations.py @ 4:4e2883bb058d draft
Synchronise version with one on main toolshed (version of 30/06/2015)
author | pjbriggs |
---|---|
date | Wed, 10 Aug 2016 11:04:42 -0400 |
parents | d9032bb158b9 |
children | e853be203962 |
rev | line source |
---|---|
1 | 1 #!/usr/bin/env python |
2 # | |
3 | |
4 import sys | |
5 import os | |
4
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
6 import subprocess |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
7 import tempfile |
1 | 8 import optparse |
9 import urllib2 | |
10 import gzip | |
4
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
11 import shutil |
1 | 12 |
13 from galaxy.util.json import from_json_string, to_json_string | |
14 | |
15 # Download file from specified URL and put into local subdir | |
16 | |
17 if __name__ == '__main__': | |
18 #Parse Command Line | |
19 parser = optparse.OptionParser() | |
20 options,args = parser.parse_args() | |
21 print "options: %s" % options | |
22 print "args : %s" % args | |
23 if len(args) != 2: | |
24 p.error("Need to supply JSON file name and description text") | |
25 | |
26 # Read the JSON supplied from the data manager tool | |
27 # Results from this program will be returned via the | |
28 # same file | |
29 jsonfile = args[0] | |
30 params = from_json_string(open(jsonfile).read() ) | |
31 print "%s" % params | |
32 | |
33 # Extract the data from the input JSON | |
34 # See https://wiki.galaxyproject.org/Admin/Tools/DataManagers/HowTo/Define?highlight=%28\bAdmin%2FTools%2FDataManagers\b%29 | |
35 # for example of JSON | |
36 # | |
37 # We want the values set in the data manager XML | |
38 dbkey = params['param_dict']['dbkey'] | |
4
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
39 description = args[1].strip() |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
40 identifier = params['param_dict']['unique_id'].strip() |
1 | 41 # Where to put the output file |
42 # Nb we have to make this ourselves, it doesn't exist by default | |
43 target_dir = params['output_data'][0]['extra_files_path'] | |
44 os.mkdir(target_dir) | |
45 | |
4
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
46 method = params['param_dict']['reference_source']['reference_source_selector'] |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
47 |
1 | 48 # Dictionary for returning to data manager |
49 data_manager_dict = {} | |
4
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
50 data_manager_dict['data_tables'] = dict() |
1 | 51 |
52 # Download from URL | |
4
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
53 if method == 'web': |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
54 url = params['param_dict']['reference_source']['annotation_url'] |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
55 print "Downloading: %s" % url |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
56 annotation_file_name = os.path.basename(url) |
1 | 57 annotation_file_path = os.path.join(target_dir,annotation_file_name) |
58 print "Annotation file name: %s" % annotation_file_name | |
59 print "Annotation file path: %s" % annotation_file_path | |
4
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
60 open(annotation_file_path,'wb').write(urllib2.urlopen(url).read()) |
1 | 61 if annotation_file_name.endswith('.gz'): |
62 # Uncompress | |
63 uncompressed_file = annotation_file_path[:-3] | |
64 open(uncompressed_file,'wb').write(gzip.open(annotation_file_path,'rb').read()) | |
65 # Remove gzipped file | |
66 os.remove(annotation_file_path) | |
67 annotation_file_name = os.path.basename(uncompressed_file) | |
68 annotation_file_path = uncompressed_file | |
4
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
69 # Update the identifier and description |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
70 if not identifier: |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
71 identifier = "%s_ceas_web" % dbkey |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
72 if not description: |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
73 description = "%s (%s)" % (os.path.splitext(annotation_file_name)[0],dbkey) |
1 | 74 # Update the output dictionary |
75 data_manager_dict['data_tables']['ceas_annotations'] = { | |
4
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
76 'value': identifier, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
77 'dbkey': dbkey, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
78 'name': description, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
79 'path': annotation_file_name, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
80 } |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
81 elif method == 'server': |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
82 # Pull in a file from the server |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
83 filename = params['param_dict']['reference_source']['annotation_filename'] |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
84 create_symlink = params['param_dict']['reference_source']['create_symlink'] |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
85 print "Canonical gene list file name: %s" % filename |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
86 print "Create symlink: %s" % create_symlink |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
87 target_filename = os.path.join(target_dir,os.path.basename(filename)) |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
88 if create_symlink == 'copy_file': |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
89 shutil.copyfile(filename,target_filename) |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
90 else: |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
91 os.symlink(filename,target_filename) |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
92 # Update the identifier and description |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
93 if not identifier: |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
94 identifier = "%s_%s" % (dbkey, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
95 os.path.splitext(os.path.basename(filename))[0]) |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
96 if not description: |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
97 description = "%s: %s" % (dbkey, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
98 os.path.splitext(os.path.basename(filename))[0]) |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
99 # Update the output dictionary |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
100 data_manager_dict['data_tables']['ceas_annotations'] = { |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
101 'value': identifier, |
1 | 102 'dbkey': dbkey, |
103 'name': description, | |
4
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
104 'path': os.path.basename(filename), |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
105 } |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
106 elif method == 'from_wig': |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
107 # Make a reference file from a wig file |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
108 wig_file = params['param_dict']['reference_source']['wig_file'] |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
109 gene_annotation = params['param_dict']['reference_source']['gene_annotation'] |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
110 target_filename = os.path.join(target_dir,"%s_%s.%s" % (dbkey, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
111 os.path.basename(wig_file), |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
112 gene_annotation)) |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
113 print "Wig file: %s" % wig_file |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
114 print "Gene annotation: %s" % gene_annotation |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
115 print "Output file: %s" % os.path.basename(target_filename) |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
116 # Make a working directory |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
117 working_dir = tempfile.mkdtemp() |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
118 # Collect stderr in a file for reporting later |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
119 stderr_filen = tempfile.NamedTemporaryFile().name |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
120 # Build the command to run |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
121 cmd = "build_genomeBG -d %s -g %s -w %s -o %s" % (dbkey, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
122 gene_annotation, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
123 wig_file, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
124 target_filename) |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
125 print "Running %s" % cmd |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
126 proc = subprocess.Popen(args=cmd,shell=True,cwd=working_dir, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
127 stderr=open(stderr_filen,'wb')) |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
128 proc.wait() |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
129 # Copy stderr to stdout |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
130 with open(stderr_filen,'r') as fp: |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
131 sys.stdout.write(fp.read()) |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
132 # Update identifier and description |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
133 if not identifier: |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
134 identifier = "%s_%s_%s" % (dbkey, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
135 gene_annotation, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
136 os.path.basename(wig_file)) |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
137 if not description: |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
138 description = "%s %s from %s" % (dbkey, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
139 gene_annotation, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
140 os.path.basename(wig_file)) |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
141 # Update the output dictionary |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
142 data_manager_dict['data_tables']['ceas_annotations'] = { |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
143 'value': identifier, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
144 'dbkey': dbkey, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
145 'name': description, |
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
146 'path': os.path.basename(target_filename), |
1 | 147 } |
148 else: | |
4
4e2883bb058d
Synchronise version with one on main toolshed (version of 30/06/2015)
pjbriggs
parents:
1
diff
changeset
|
149 raise NotImplementedError("Method '%s' not implemented" % method) |
1 | 150 |
151 #save info to json file | |
152 open(jsonfile,'wb').write(to_json_string(data_manager_dict)) | |
153 |