Mercurial > repos > iuc > snpeff
annotate gbk2fa.py @ 28:b1b327f475ac draft
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit e4366b6a23223f84039a0590cf0d7079b83d8f84"
| author | iuc |
|---|---|
| date | Wed, 13 Oct 2021 23:29:54 +0000 |
| parents | 718842784732 |
| children | 030fe29d4c47 |
| rev | line source |
|---|---|
|
14
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
1 import argparse |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
2 import bz2 |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
3 import contextlib |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
4 import gzip |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
5 import sys |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
6 |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
7 import magic |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
8 from Bio import SeqIO |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
9 |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
10 parser = argparse.ArgumentParser() |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
11 parser.add_argument("genbank_file", help="GenBank input file. Can be compressed with gzip or bzip2") |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
12 parser.add_argument("fasta_file", help="FASTA output datset") |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
13 parser.add_argument("--remove_version", dest="remove_version", action="store_true", help="Remove version number from NCBI form formatted accession numbers. For example, this would convert 'B000657.2' to 'B000657'") |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
14 args = parser.parse_args() |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
15 |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
16 gbk_filename = args.genbank_file |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
17 fa_filename = args.fasta_file |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
18 |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
19 |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
20 @contextlib.contextmanager |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
21 def get_file_handle(gbk_filename): |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
22 f_type = magic.from_file(args.genbank_file, mime=True) |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
23 if f_type == 'text/plain': |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
24 input_handle = open(gbk_filename, "r") |
|
19
718842784732
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit 999eca8a05f17ae567f99b8ca3394f2105491173
iuc
parents:
14
diff
changeset
|
25 elif f_type == 'application/gzip' or f_type == 'application/x-gzip': |
|
14
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
26 input_handle = gzip.open(gbk_filename, "rt") |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
27 elif f_type == 'application/x-bzip2': |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
28 input_handle = bz2.open(gbk_filename, "rt") |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
29 else: |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
30 sys.exit("Cannot process file of type {}. Only plain, gzip'ed, and bzip2'ed genbank files are accepted ".format(f_type)) |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
31 yield input_handle |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
32 input_handle.close() |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
33 |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
34 |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
35 with get_file_handle(gbk_filename) as input_handle, open(fa_filename, "w") as output_handle: |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
36 |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
37 for seq_record in SeqIO.parse(input_handle, "genbank"): |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
38 if args.remove_version: |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
39 seq_id = seq_record.id.split('.')[0] |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
40 else: |
|
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
41 seq_id = seq_record.id |
|
28
b1b327f475ac
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit e4366b6a23223f84039a0590cf0d7079b83d8f84"
iuc
parents:
19
diff
changeset
|
42 print('Writing FASTA record: {}'.format(seq_id)) |
|
14
a7c106b938dd
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/snpeff commit eea43430ff90fe6b13b295f6d5efb2208401a7ef
iuc
parents:
diff
changeset
|
43 output_handle.write(">{}\n{}\n".format(seq_id, seq_record.seq)) |