Mercurial > repos > eric-rasche > apollo

annotate export.py @ 8:df7a90763b3c draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 37aa8fa02c144220da9a3899644025b8c19697c9
author eric-rasche
date Tue, 27 Jun 2017 04:13:27 -0400
parents f9a6e151b3b4
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
1 #!/usr/bin/env python
7
f9a6e151b3b4 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 52b9e5bf6a6efb09a5cb845ee48703651c644174
eric-rasche
parents: 5
diff changeset
2 from __future__ import print_function
f9a6e151b3b4 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 52b9e5bf6a6efb09a5cb845ee48703651c644174
eric-rasche
parents: 5
diff changeset
3 import argparse
f9a6e151b3b4 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 52b9e5bf6a6efb09a5cb845ee48703651c644174
eric-rasche
parents: 5
diff changeset
4 import json
3
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
5 import sys
7
f9a6e151b3b4 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 52b9e5bf6a6efb09a5cb845ee48703651c644174
eric-rasche
parents: 5
diff changeset
6 from Bio import SeqIO
f9a6e151b3b4 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 52b9e5bf6a6efb09a5cb845ee48703651c644174
eric-rasche
parents: 5
diff changeset
7 from BCBio import GFF
f9a6e151b3b4 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 52b9e5bf6a6efb09a5cb845ee48703651c644174
eric-rasche
parents: 5
diff changeset
8 from webapollo import WAAuth, WebApolloInstance, CnOrGuess, GuessCn
f9a6e151b3b4 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 52b9e5bf6a6efb09a5cb845ee48703651c644174
eric-rasche
parents: 5
diff changeset
9 from future import standard_library
f9a6e151b3b4 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 52b9e5bf6a6efb09a5cb845ee48703651c644174
eric-rasche
parents: 5
diff changeset
10 standard_library.install_aliases()
5
7610987e0c48 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 29795b77c0d5c7894219b018a92c5ee7818096c3
eric-rasche
parents: 3
diff changeset
11 try:
7610987e0c48 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 29795b77c0d5c7894219b018a92c5ee7818096c3
eric-rasche
parents: 3
diff changeset
12 import StringIO as io
7610987e0c48 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 29795b77c0d5c7894219b018a92c5ee7818096c3
eric-rasche
parents: 3
diff changeset
13 except ImportError:
7610987e0c48 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 29795b77c0d5c7894219b018a92c5ee7818096c3
eric-rasche
parents: 3
diff changeset
14 import io
7610987e0c48 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 29795b77c0d5c7894219b018a92c5ee7818096c3
eric-rasche
parents: 3
diff changeset
15
3
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
16
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
17 def export(org_cn, seqs):
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
18 org_data = wa.organisms.findOrganismByCn(org_cn)
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
19
5
7610987e0c48 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 29795b77c0d5c7894219b018a92c5ee7818096c3
eric-rasche
parents: 3
diff changeset
20 data = io.StringIO()
3
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
21
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
22 kwargs = dict(
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
23 exportType='GFF3',
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
24 seqType='genomic',
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
25 exportGff3Fasta=True,
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
26 output="text",
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
27 exportFormat="text",
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
28 organism=org_cn,
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
29 )
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
30
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
31 if len(seqs) > 0:
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
32 data.write(wa.io.write(
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
33 exportAllSequences=False,
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
34 sequences=seqs,
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
35 **kwargs
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
36 ).encode('utf-8'))
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
37 else:
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
38 data.write(wa.io.write(
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
39 exportAllSequences=True,
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
40 sequences=[],
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
41 **kwargs
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
42 ).encode('utf-8'))
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
43
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
44 # Seek back to start
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
45 data.seek(0)
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
46
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
47 records = list(GFF.parse(data))
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
48 if len(records) == 0:
5
7610987e0c48 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 29795b77c0d5c7894219b018a92c5ee7818096c3
eric-rasche
parents: 3
diff changeset
49 print("Could not find any sequences or annotations for this organism + reference sequence")
3
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
50 sys.exit(2)
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
51 else:
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
52 for record in records:
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
53 record.annotations = {}
5
7610987e0c48 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 29795b77c0d5c7894219b018a92c5ee7818096c3
eric-rasche
parents: 3
diff changeset
54 record.features = sorted(record.features, key=lambda x: x.location.start)
3
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
55 if args.gff:
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
56 GFF.write([record], args.gff)
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
57 record.description = ""
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
58 if args.fasta:
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
59 SeqIO.write([record], args.fasta, 'fasta')
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
60
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
61 return org_data
0
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
62
5
7610987e0c48 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 29795b77c0d5c7894219b018a92c5ee7818096c3
eric-rasche
parents: 3
diff changeset
63
0
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
64 if __name__ == '__main__':
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
65 parser = argparse.ArgumentParser(description='Sample script to add an attribute to a feature via web services')
3
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
66 WAAuth(parser)
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
67 CnOrGuess(parser)
0
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
68 parser.add_argument('--gff', type=argparse.FileType('w'))
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
69 parser.add_argument('--fasta', type=argparse.FileType('w'))
3
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
70 parser.add_argument('--json', type=argparse.FileType('w'))
0
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
71
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
72 args = parser.parse_args()
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
73
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
74 wa = WebApolloInstance(args.apollo, args.username, args.password)
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
75
3
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
76 org_cn_list, seqs = GuessCn(args, wa)
0
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
77
3
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
78 org_data = []
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
79 for org_cn in org_cn_list:
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
80 indiv_org_data = export(org_cn, seqs)
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
81 org_data.append(indiv_org_data)
d4ae83dedb14 planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4ac38d0b6dba1183f3e78eb5c224c7051064b4a5
eric-rasche
parents: 0
diff changeset
82 args.json.write(json.dumps(org_data, indent=2))