Mercurial > repos > devteam > data_manager_fetch_genome_dbkeys_all_fasta
annotate data_manager/data_manager_fetch_genome_all_fasta_dbkeys.py @ 4:60994ca04177 draft
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
author | iuc |
---|---|
date | Tue, 04 Apr 2017 17:13:31 -0400 |
parents | 86fa71e9b427 |
children | b1bc53e9bbc5 |
rev | line source |
---|---|
0 | 1 #!/usr/bin/env python |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
2 # Dan Blankenberg |
0 | 3 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
4 import bz2 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
5 import gzip |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
6 import optparse |
0 | 7 import os |
8 import shutil | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
9 import sys |
0 | 10 import tarfile |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
11 import tempfile |
0 | 12 import zipfile |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
13 from ftplib import FTP |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
14 from json import dumps, loads |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
15 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
16 try: |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
17 # For Python 3.0 and later |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
18 from io import BytesIO as StringIO |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
19 from io import UnsupportedOperation |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
20 from urllib.request import urlopen |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
21 except ImportError: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
22 # Fall back to Python 2 imports |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
23 from StringIO import StringIO |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
24 from urllib2 import urlopen |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
25 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
26 UnsupportedOperation = AttributeError |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
27 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
28 CHUNK_SIZE = 2 ** 20 # 1mb |
0 | 29 |
30 | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
31 def cleanup_before_exit(tmp_dir): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
32 if tmp_dir and os.path.exists(tmp_dir): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
33 shutil.rmtree(tmp_dir) |
0 | 34 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
35 |
0 | 36 def stop_err(msg): |
37 sys.stderr.write(msg) | |
38 sys.exit(1) | |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
39 |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
40 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
41 def get_dbkey_dbname_id_name(params, dbkey_description=None): |
0 | 42 dbkey = params['param_dict']['dbkey_source']['dbkey'] |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
43 # TODO: ensure sequence_id is unique and does not already appear in location file |
0 | 44 sequence_id = params['param_dict']['sequence_id'] |
45 if not sequence_id: | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
46 sequence_id = dbkey # uuid.uuid4() generate and use an uuid instead? |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
47 |
0 | 48 if params['param_dict']['dbkey_source']['dbkey_source_selector'] == 'new': |
49 dbkey_name = params['param_dict']['dbkey_source']['dbkey_name'] | |
50 if not dbkey_name: | |
51 dbkey_name = dbkey | |
52 else: | |
53 dbkey_name = None | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
54 |
0 | 55 sequence_name = params['param_dict']['sequence_name'] |
56 if not sequence_name: | |
57 sequence_name = dbkey_description | |
58 if not sequence_name: | |
59 sequence_name = dbkey | |
60 return dbkey, dbkey_name, sequence_id, sequence_name | |
61 | |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
62 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
63 def _get_files_in_ftp_path(ftp, path): |
0 | 64 path_contents = [] |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
65 ftp.retrlines('MLSD %s' % (path), path_contents.append) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
66 return [line.split(';')[-1].lstrip() for line in path_contents] |
0 | 67 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
68 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
69 def _get_stream_readers_for_tar(fh, tmp_dir): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
70 fasta_tar = tarfile.open(fileobj=fh, mode='r:*') |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
71 return [x for x in [fasta_tar.extractfile(member) for member in fasta_tar.getmembers()] if x] |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
72 |
0 | 73 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
74 def _get_stream_readers_for_zip(fh, tmp_dir): |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
75 """ |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
76 Unpacks all archived files in a zip file. |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
77 Individual files will be concatenated (in _stream_fasta_to_file) |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
78 """ |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
79 fasta_zip = zipfile.ZipFile(fh, 'r') |
0 | 80 rval = [] |
81 for member in fasta_zip.namelist(): | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
82 fasta_zip.extract(member, tmp_dir) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
83 rval.append(open(os.path.join(tmp_dir, member), 'rb')) |
0 | 84 return rval |
85 | |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
86 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
87 def _get_stream_readers_for_gzip(fh, tmp_dir): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
88 return [gzip.GzipFile(fileobj=fh, mode='rb')] |
0 | 89 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
90 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
91 def _get_stream_readers_for_bz2(fh, tmp_dir): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
92 return [bz2.BZ2File(fh.name, 'rb')] |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
93 |
0 | 94 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
95 def sort_fasta(fasta_filename, sort_method, params): |
0 | 96 if sort_method is None: |
97 return | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
98 assert sort_method in SORTING_METHODS, ValueError("%s is not a valid sorting option." % sort_method) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
99 return SORTING_METHODS[sort_method](fasta_filename, params) |
0 | 100 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
101 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
102 def _move_and_index_fasta_for_sorting(fasta_filename): |
0 | 103 unsorted_filename = tempfile.NamedTemporaryFile().name |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
104 shutil.move(fasta_filename, unsorted_filename) |
0 | 105 fasta_offsets = {} |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
106 unsorted_fh = open(unsorted_filename) |
0 | 107 while True: |
108 offset = unsorted_fh.tell() | |
109 line = unsorted_fh.readline() | |
110 if not line: | |
111 break | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
112 if line.startswith(">"): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
113 line = line.split(None, 1)[0][1:] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
114 fasta_offsets[line] = offset |
0 | 115 unsorted_fh.close() |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
116 current_order = [_[1] for _ in sorted((_[1], _[0]) for _ in fasta_offsets.items())] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
117 return (unsorted_filename, fasta_offsets, current_order) |
0 | 118 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
119 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
120 def _write_sorted_fasta(sorted_names, fasta_offsets, sorted_fasta_filename, unsorted_fasta_filename): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
121 unsorted_fh = open(unsorted_fasta_filename) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
122 sorted_fh = open(sorted_fasta_filename, 'wb+') |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
123 |
0 | 124 for name in sorted_names: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
125 offset = fasta_offsets[name] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
126 unsorted_fh.seek(offset) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
127 sorted_fh.write(unsorted_fh.readline()) |
0 | 128 while True: |
129 line = unsorted_fh.readline() | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
130 if not line or line.startswith(">"): |
0 | 131 break |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
132 sorted_fh.write(line) |
0 | 133 unsorted_fh.close() |
134 sorted_fh.close() | |
135 | |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
136 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
137 def _sort_fasta_as_is(fasta_filename, params): |
0 | 138 return |
139 | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
140 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
141 def _sort_fasta_lexicographical(fasta_filename, params): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
142 (unsorted_filename, fasta_offsets, current_order) = _move_and_index_fasta_for_sorting(fasta_filename) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
143 sorted_names = sorted(fasta_offsets.keys()) |
0 | 144 if sorted_names == current_order: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
145 shutil.move(unsorted_filename, fasta_filename) |
0 | 146 else: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
147 _write_sorted_fasta(sorted_names, fasta_offsets, fasta_filename, unsorted_filename) |
0 | 148 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
149 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
150 def _sort_fasta_gatk(fasta_filename, params): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
151 (unsorted_filename, fasta_offsets, current_order) = _move_and_index_fasta_for_sorting(fasta_filename) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
152 sorted_names = list(map(str, range(1, 23))) + ['X', 'Y'] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
153 # detect if we have chrN, or just N |
0 | 154 has_chr = False |
155 for chrom in sorted_names: | |
156 if "chr%s" % chrom in current_order: | |
157 has_chr = True | |
158 break | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
159 |
0 | 160 if has_chr: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
161 sorted_names = ["chr%s" % x for x in sorted_names] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
162 sorted_names.insert(0, "chrM") |
0 | 163 else: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
164 sorted_names.insert(0, "MT") |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
165 sorted_names.extend("%s_random" % x for x in sorted_names) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
166 |
0 | 167 existing_sorted_names = [] |
168 for name in sorted_names: | |
169 if name in current_order: | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
170 existing_sorted_names.append(name) |
0 | 171 for name in current_order: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
172 # TODO: confirm that non-canonical names do not need to be sorted specially |
0 | 173 if name not in existing_sorted_names: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
174 existing_sorted_names.append(name) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
175 |
0 | 176 if existing_sorted_names == current_order: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
177 shutil.move(unsorted_filename, fasta_filename) |
0 | 178 else: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
179 _write_sorted_fasta(existing_sorted_names, fasta_offsets, fasta_filename, unsorted_filename) |
0 | 180 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
181 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
182 def _sort_fasta_custom(fasta_filename, params): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
183 (unsorted_filename, fasta_offsets, current_order) = _move_and_index_fasta_for_sorting(fasta_filename) |
0 | 184 sorted_names = [] |
185 for id_repeat in params['param_dict']['sorting']['sequence_identifiers']: | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
186 sorted_names.append(id_repeat['identifier']) |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
187 handle_not_listed = params['param_dict']['sorting']['handle_not_listed_selector'] |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
188 if handle_not_listed.startswith('keep'): |
0 | 189 add_list = [] |
190 for name in current_order: | |
191 if name not in sorted_names: | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
192 add_list.append(name) |
0 | 193 if add_list: |
194 if handle_not_listed == 'keep_append': | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
195 sorted_names.extend(add_list) |
0 | 196 else: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
197 add_list.extend(sorted_names) |
0 | 198 sorted_names = add_list |
199 if sorted_names == current_order: | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
200 shutil.move(unsorted_filename, fasta_filename) |
0 | 201 else: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
202 _write_sorted_fasta(sorted_names, fasta_offsets, fasta_filename, unsorted_filename) |
0 | 203 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
204 |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
205 def _download_file(start, fh): |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
206 tmp = tempfile.NamedTemporaryFile() |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
207 tmp.write(start) |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
208 tmp.write(fh.read()) |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
209 tmp.flush() |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
210 tmp.seek(0) |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
211 return tmp |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
212 |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
213 |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
214 def get_stream_reader(fh, tmp_dir): |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
215 """ |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
216 Check if file is compressed and return correct stream reader. |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
217 If file has to be downloaded, do it now. |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
218 """ |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
219 magic_dict = { |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
220 b"\x1f\x8b\x08": _get_stream_readers_for_gzip, |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
221 b"\x42\x5a\x68": _get_stream_readers_for_bz2, |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
222 b"\x50\x4b\x03\x04": _get_stream_readers_for_zip, |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
223 } |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
224 start_of_file = fh.read(CHUNK_SIZE) |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
225 try: |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
226 fh.seek(0) |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
227 except UnsupportedOperation: # This happens if fh has been created by urlopen |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
228 fh = _download_file(start_of_file, fh) |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
229 try: # Check if file is tar file |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
230 if tarfile.open(fileobj=StringIO(start_of_file)): |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
231 return _get_stream_readers_for_tar(fh, tmp_dir) |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
232 except tarfile.ReadError: |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
233 pass |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
234 for k, v in magic_dict.items(): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
235 if start_of_file.startswith(k): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
236 return v(fh, tmp_dir) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
237 return [fh] |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
238 |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
239 |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
240 def _get_ucsc_download_address(params, dbkey): |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
241 """ |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
242 Check if we can find the correct file for the supplied dbkey on UCSC's FTP server |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
243 """ |
0 | 244 UCSC_FTP_SERVER = 'hgdownload.cse.ucsc.edu' |
245 UCSC_DOWNLOAD_PATH = '/goldenPath/%s/bigZips/' | |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
246 COMPRESSED_EXTENSIONS = ['.tar.gz', '.tgz', '.tar.bz2', '.zip', '.fa.gz', '.fa.bz2'] |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
247 |
0 | 248 email = params['param_dict']['__user_email__'] |
249 if not email: | |
250 email = 'anonymous@example.com' | |
251 | |
252 ucsc_dbkey = params['param_dict']['reference_source']['requested_dbkey'] or dbkey | |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
253 UCSC_CHROM_FA_FILENAMES = ['%s.chromFa' % ucsc_dbkey, 'chromFa', ucsc_dbkey] |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
254 |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
255 ftp = FTP(UCSC_FTP_SERVER) |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
256 ftp.login('anonymous', email) |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
257 |
0 | 258 ucsc_path = UCSC_DOWNLOAD_PATH % ucsc_dbkey |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
259 path_contents = _get_files_in_ftp_path(ftp, ucsc_path) |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
260 ftp.quit() |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
261 |
0 | 262 for ucsc_chrom_fa_filename in UCSC_CHROM_FA_FILENAMES: |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
263 for ext in COMPRESSED_EXTENSIONS: |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
264 if "%s%s" % (ucsc_chrom_fa_filename, ext) in path_contents: |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
265 ucsc_file_name = "%s%s%s" % (ucsc_path, ucsc_chrom_fa_filename, ext) |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
266 return "ftp://%s%s" % (UCSC_FTP_SERVER, ucsc_file_name) |
0 | 267 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
268 raise Exception('Unable to determine filename for UCSC Genome for %s: %s' % (ucsc_dbkey, path_contents)) |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
269 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
270 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
271 def add_fasta_to_table(data_manager_dict, fasta_readers, target_directory, dbkey, dbkey_name, sequence_id, sequence_name, params, **kwds): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
272 fasta_filename = _stream_fasta_to_file(fasta_stream=fasta_readers, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
273 target_directory=target_directory, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
274 sequence_id=sequence_id) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
275 sort_fasta(fasta_filename, params['param_dict']['sorting']['sort_selector'], params) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
276 if dbkey_name: |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
277 add_dbkey_to_table(data_manager_dict=data_manager_dict, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
278 target_directory=target_directory, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
279 dbkey=dbkey, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
280 dbkey_name=dbkey_name, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
281 fasta_filename=fasta_filename) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
282 _add_data_table_entry(data_manager_dict, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
283 data_table_entry=dict(value=sequence_id, dbkey=dbkey, name=sequence_name, path=os.path.basename(fasta_filename)), |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
284 data_table_name='all_fasta') |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
285 return fasta_filename |
0 | 286 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
287 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
288 def add_dbkey_to_table(data_manager_dict, target_directory, dbkey, dbkey_name, fasta_filename): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
289 # do len calc here |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
290 len_base_name = "%s.len" % (dbkey) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
291 compute_fasta_length(fasta_filename, os.path.join(target_directory, len_base_name), keep_first_word=True) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
292 dbkey_dict = dict(value=dbkey, name=dbkey_name, len_path=len_base_name) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
293 _add_data_table_entry(data_manager_dict, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
294 data_table_entry=dbkey_dict, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
295 data_table_name='__dbkeys__') |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
296 |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
297 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
298 def download_from_ucsc(params, dbkey, tmp_dir, **kwds): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
299 url = _get_ucsc_download_address(params, dbkey) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
300 return get_stream_reader(urlopen(url), tmp_dir) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
301 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
302 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
303 def download_from_ncbi(params, tmp_dir, **kwds): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
304 NCBI_DOWNLOAD_URL = 'http://togows.dbcls.jp/entry/ncbi-nucleotide/%s.fasta' # FIXME: taken from dave's genome manager...why some japan site? |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
305 requested_identifier = params['param_dict']['reference_source']['requested_identifier'] |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
306 url = NCBI_DOWNLOAD_URL % requested_identifier |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
307 return get_stream_reader(urlopen(url), tmp_dir) |
0 | 308 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
309 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
310 def download_from_url(params, tmp_dir, **kwds): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
311 """ |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
312 Download a file from a URL and return a list of filehandles from which to read the data. |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
313 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
314 >>> url = 'https://github.com/mvdbeek/tools-devteam/raw/data_manager/data_managers/data_manager_fetch_genome_dbkeys_all_fasta/test-data/test.tar' |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
315 >>> params = {'param_dict': {'reference_source': {'user_url': url}}} |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
316 >>> tmp_dir = tempfile.mkdtemp() |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
317 >>> fh = download_from_url(params=params, tmp_dir=tmp_dir)[0][0] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
318 >>> assert fh.readline().startswith('>FBtr0304171') |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
319 >>> url = 'https://github.com/mvdbeek/tools-devteam/raw/data_manager/data_managers/data_manager_fetch_genome_dbkeys_all_fasta/test-data/test.tar.bz2' |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
320 >>> params = {'param_dict': {'reference_source': {'user_url': url}}} |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
321 >>> fh = download_from_url(params=params, tmp_dir=tmp_dir)[0][0] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
322 >>> assert fh.readline().startswith('>FBtr0304171') |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
323 >>> url = 'https://github.com/mvdbeek/tools-devteam/raw/data_manager/data_managers/data_manager_fetch_genome_dbkeys_all_fasta/test-data/test.tar.gz' |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
324 >>> params = {'param_dict': {'reference_source': {'user_url': url}}} |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
325 >>> fh = download_from_url(params=params, tmp_dir=tmp_dir)[0][0] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
326 >>> assert fh.readline().startswith('>FBtr0304171') |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
327 >>> url = 'https://github.com/mvdbeek/tools-devteam/raw/data_manager/data_managers/data_manager_fetch_genome_dbkeys_all_fasta/test-data/test.zip' |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
328 >>> params = {'param_dict': {'reference_source': {'user_url': url}}} |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
329 >>> fh = download_from_url(params=params, tmp_dir=tmp_dir)[0][0] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
330 >>> assert fh.readline().startswith('>FBtr0304171') |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
331 >>> url = 'https://raw.githubusercontent.com/galaxyproject/tools-devteam/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta/test-data/phiX174.fasta' |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
332 >>> params = {'param_dict': {'reference_source': {'user_url': url}}} |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
333 >>> fh = download_from_url(params=params, tmp_dir=tmp_dir)[0][0] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
334 >>> assert fh.readline().startswith('>phiX174') |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
335 """ |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
336 urls = filter(bool, [x.strip() for x in params['param_dict']['reference_source']['user_url'].split('\n')]) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
337 return [get_stream_reader(urlopen(url), tmp_dir) for url in urls] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
338 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
339 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
340 def download_from_history(params, tmp_dir, **kwds): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
341 # TODO: allow multiple FASTA input files |
0 | 342 input_filename = params['param_dict']['reference_source']['input_fasta'] |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
343 if isinstance(input_filename, list): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
344 fasta_readers = [get_stream_reader(open(filename, 'rb'), tmp_dir) for filename in input_filename] |
0 | 345 else: |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
346 fasta_readers = get_stream_reader(open(input_filename), tmp_dir) |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
347 return fasta_readers |
0 | 348 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
349 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
350 def copy_from_directory(data_manager_dict, params, target_directory, dbkey, dbkey_name, sequence_id, sequence_name, tmp_dir): |
0 | 351 input_filename = params['param_dict']['reference_source']['fasta_filename'] |
352 create_symlink = params['param_dict']['reference_source']['create_symlink'] == 'create_symlink' | |
353 if create_symlink: | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
354 data_table_entries = _create_symlink(input_filename, target_directory, dbkey, dbkey_name, sequence_id, sequence_name) |
0 | 355 else: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
356 if isinstance(input_filename, list): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
357 fasta_readers = [get_stream_reader(open(filename, 'rb'), tmp_dir) for filename in input_filename] |
0 | 358 else: |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
359 fasta_readers = get_stream_reader(open(input_filename), tmp_dir) |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
360 return fasta_readers |
0 | 361 for data_table_name, data_table_entry in data_table_entries: |
362 if data_table_entry: | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
363 _add_data_table_entry(data_manager_dict, data_table_entry, data_table_name) |
0 | 364 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
365 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
366 def _add_data_table_entry(data_manager_dict, data_table_entry, data_table_name): |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
367 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {}) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
368 data_manager_dict['data_tables'][data_table_name] = data_manager_dict['data_tables'].get('all_fasta', []) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
369 data_manager_dict['data_tables'][data_table_name].append(data_table_entry) |
0 | 370 return data_manager_dict |
371 | |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
372 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
373 def _stream_fasta_to_file(fasta_stream, target_directory, sequence_id, close_stream=True): |
0 | 374 fasta_base_filename = "%s.fa" % sequence_id |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
375 fasta_filename = os.path.join(target_directory, fasta_base_filename) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
376 with open(fasta_filename, 'wb+') as fasta_writer: |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
377 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
378 if isinstance(fasta_stream, list) and len(fasta_stream) == 1: |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
379 fasta_stream = fasta_stream[0] |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
380 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
381 if isinstance(fasta_stream, list): |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
382 last_char = None |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
383 for fh in fasta_stream: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
384 if last_char not in [None, '\n', '\r', b'\n', b'\r']: |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
385 fasta_writer.write(b'\n') |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
386 while True: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
387 data = fh.read(CHUNK_SIZE) |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
388 if data: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
389 fasta_writer.write(data) |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
390 last_char = data[-1] |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
391 else: |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
392 break |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
393 if close_stream: |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
394 fh.close() |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
395 else: |
0 | 396 while True: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
397 data = fasta_stream.read(CHUNK_SIZE) |
0 | 398 if data: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
399 fasta_writer.write(data) |
0 | 400 else: |
401 break | |
402 if close_stream: | |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
403 fasta_stream.close() |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
404 return fasta_filename |
0 | 405 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
406 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
407 def compute_fasta_length(fasta_file, out_file, keep_first_word=False): |
0 | 408 infile = fasta_file |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
409 out = open(out_file, 'w') |
0 | 410 |
411 fasta_title = '' | |
412 seq_len = 0 | |
413 | |
414 first_entry = True | |
415 | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
416 for line in open(infile): |
0 | 417 line = line.strip() |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
418 if not line or line.startswith('#'): |
0 | 419 continue |
420 if line[0] == '>': | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
421 if not first_entry: |
0 | 422 if keep_first_word: |
423 fasta_title = fasta_title.split()[0] | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
424 out.write("%s\t%d\n" % (fasta_title[1:], seq_len)) |
0 | 425 else: |
426 first_entry = False | |
427 fasta_title = line | |
428 seq_len = 0 | |
429 else: | |
430 seq_len += len(line) | |
431 | |
432 # last fasta-entry | |
433 if keep_first_word: | |
434 fasta_title = fasta_title.split()[0] | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
435 out.write("%s\t%d\n" % (fasta_title[1:], seq_len)) |
0 | 436 out.close() |
437 | |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
438 |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
439 def _create_symlink(input_filename, target_directory, dbkey, dbkey_name, sequence_id, sequence_name): |
0 | 440 fasta_base_filename = "%s.fa" % sequence_id |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
441 fasta_filename = os.path.join(target_directory, fasta_base_filename) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
442 os.symlink(input_filename, fasta_filename) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
443 |
0 | 444 dbkey_dict = None |
445 if dbkey_name: | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
446 # do len calc here |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
447 len_base_name = "%s.len" % (dbkey) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
448 compute_fasta_length(fasta_filename, os.path.join(target_directory, len_base_name), keep_first_word=True) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
449 dbkey_dict = dict(value=dbkey, name=dbkey_name, len_path=len_base_name) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
450 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
451 return [('__dbkeys__', dbkey_dict), ('all_fasta', dict(value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_filename))] |
0 | 452 |
453 | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
454 REFERENCE_SOURCE_TO_DOWNLOAD = dict(ucsc=download_from_ucsc, ncbi=download_from_ncbi, url=download_from_url, history=download_from_history, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
455 directory=copy_from_directory) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
456 SORTING_METHODS = dict(as_is=_sort_fasta_as_is, lexicographical=_sort_fasta_lexicographical, gatk=_sort_fasta_gatk, custom=_sort_fasta_custom) |
0 | 457 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
458 |
0 | 459 def main(): |
460 parser = optparse.OptionParser() | |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
461 parser.add_option('-d', '--dbkey_description', dest='dbkey_description', action='store', type="string", default=None, help='dbkey_description') |
0 | 462 (options, args) = parser.parse_args() |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
463 |
0 | 464 filename = args[0] |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
465 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
466 params = loads(open(filename).read()) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
467 target_directory = params['output_data'][0]['extra_files_path'] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
468 os.mkdir(target_directory) |
0 | 469 data_manager_dict = {} |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
470 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
471 dbkey, dbkey_name, sequence_id, sequence_name = get_dbkey_dbname_id_name(params, dbkey_description=options.dbkey_description) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
472 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
473 if dbkey in [None, '', '?']: |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
474 raise Exception('"%s" is not a valid dbkey. You must specify a valid dbkey.' % (dbkey)) |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
475 |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
476 # Create a tmp_dir, in case a zip file needs to be uncompressed |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
477 tmp_dir = tempfile.mkdtemp() |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
478 # Fetch the FASTA |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
479 try: |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
480 reference_source = params['param_dict']['reference_source']['reference_source_selector'] |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
481 fasta_readers = REFERENCE_SOURCE_TO_DOWNLOAD[reference_source](data_manager_dict=data_manager_dict, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
482 params=params, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
483 target_directory=target_directory, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
484 dbkey=dbkey, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
485 dbkey_name=dbkey_name, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
486 sequence_id=sequence_id, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
487 sequence_name=sequence_name, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
488 tmp_dir=tmp_dir) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
489 if fasta_readers: |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
490 add_fasta_to_table(data_manager_dict=data_manager_dict, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
491 fasta_readers=fasta_readers, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
492 target_directory=target_directory, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
493 dbkey=dbkey, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
494 dbkey_name=dbkey_name, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
495 sequence_id=sequence_id, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
496 sequence_name=sequence_name, |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
497 params=params) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
498 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
499 finally: |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
500 cleanup_before_exit(tmp_dir) |
4
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
501 # save info to json file |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
502 open(filename, 'wb').write(dumps(data_manager_dict).encode()) |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
503 |
60994ca04177
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
3
diff
changeset
|
504 |
3
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
505 if __name__ == "__main__": |
86fa71e9b427
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_fetch_genome_dbkeys_all_fasta commit 9df6291d7d65c5274c4fd35775108eae58ea6c23
devteam
parents:
1
diff
changeset
|
506 main() |