Mercurial > repos > iuc > data_manager_snpsift_dbnsfp
annotate data_manager/data_manager_snpsift_dbnsfp.py @ 8:752439501b04 draft default tip
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit ad15c8e96a0633f1939266d63fd17978afa7d9b7
author | iuc |
---|---|
date | Thu, 09 Feb 2023 14:55:23 +0000 |
parents | fe8a9ab8daf9 |
children |
rev | line source |
---|---|
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
1 #!/usr/bin/env python |
7
fe8a9ab8daf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
6
diff
changeset
|
2 from __future__ import print_function |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
3 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
4 import gzip |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
5 import json |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
6 import optparse |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
7 import os |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
8 import os.path |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
9 import re |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
10 import shutil |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
11 import sys |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
12 import urllib |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
13 import zipfile |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
14 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
15 from pysam import ctabix |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
16 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
17 """ |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
18 # Install dbNSFP databases |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
19 # from DbNsfp site |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
20 # Download dbNSFP database |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
21 $ wget ftp://dbnsfp:dbnsfp@dbnsfp.softgenetics.com/dbNSFPv2.4.zip |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
22 # Uncompress |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
23 $ unzip dbNSFP2.4.zip |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
24 # Create a single file version |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
25 $ (head -n 1 dbNSFP2.4_variant.chr1 ; cat dbNSFP2.4_variant.chr* | grep -v "^#") > dbNSFP2.4.txt |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
26 # Compress using block-gzip algorithm |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
27 bgzip dbNSFP2.4.txt |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
28 # Create tabix index |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
29 tabix -s 1 -b 2 -e 2 dbNSFP2.4.txt.gz |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
30 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
31 data_table: |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
32 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
33 <table name="snpsift_dbnsfps" comment_char="#"> |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
34 <columns>key, build, name, value, annotations</columns> |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
35 <file path="tool-data/snpsift_dbnsfps.loc" /> |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
36 </table> |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
37 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
38 #id build description path annotations |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
39 #GRCh37_dbNSFP2.4 GRCh37 GRCh37 dbNSFP2.4 /depot/snpeff/dbNSFP2.4.gz SIFT_pred,Uniprot_acc |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
40 #GRCh38_dbNSFP2.7 GRCh38 GRCh38 dbNSFP2.7 /depot/snpeff/dbNSFP2.7.gz SIFT_pred,Uniprot_acc |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
41 """ |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
42 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
43 data_table = 'snpsift_dbnsfps' |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
44 softgenetics_url = 'ftp://dbnsfp:dbnsfp@dbnsfp.softgenetics.com/' |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
45 dbNSFP_file_pat = '(dbNSFP(.*)_variant|dbscSNV(.*)).chr(.*)' |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
46 tokenize = re.compile(r'(\d+)|(\D+)').findall |
5
09f9bfb2b33b
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit a4b0969b33a68a0ea9ba12291f6694aec24f13ed
iuc
parents:
4
diff
changeset
|
47 dbNSFP_name_pat = r'dbNSFP(v|_light)?(\d*).*?' |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
48 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
49 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
50 def get_nsfp_genome_version(name): |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
51 genome_version = 'hg19' |
5
09f9bfb2b33b
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit a4b0969b33a68a0ea9ba12291f6694aec24f13ed
iuc
parents:
4
diff
changeset
|
52 dbNSFP_name_pat = r'(dbscSNV|dbNSFP(v|_light)?)(\d*).*?' |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
53 m = re.match(dbNSFP_name_pat, name) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
54 if m: |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
55 (base, mid, ver) = m.groups() |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
56 if base == 'dbscSNV': |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
57 genome_version = 'hg19' |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
58 else: |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
59 genome_version = 'hg38' if ver == '3' else 'hg19' if ver == '2' else 'hg18' |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
60 return genome_version |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
61 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
62 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
63 def get_annotations(gzip_path): |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
64 annotations = None |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
65 try: |
7
fe8a9ab8daf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
6
diff
changeset
|
66 with gzip.open(gzip_path, 'r') as fh: |
fe8a9ab8daf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
6
diff
changeset
|
67 buf = fh.read(10000) |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
68 lines = buf.splitlines() |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
69 headers = lines[0].split('\t') |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
70 annotations = ','.join([x.strip() for x in headers[4:]]) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
71 except Exception as e: |
7
fe8a9ab8daf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
6
diff
changeset
|
72 sys.exit('Error Reading annotations %s : %s' % (gzip_path, e)) |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
73 return annotations |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
74 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
75 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
76 def tabix_file(input_fname, output_fname): |
7
fe8a9ab8daf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
6
diff
changeset
|
77 print("tabix_file: %s -> %s" % (input_fname, output_fname)) |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
78 ctabix.tabix_compress(input_fname, output_fname, force=True) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
79 # Column indices are 0-based. |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
80 ctabix.tabix_index(output_fname, seq_col=0, start_col=1, end_col=1) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
81 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
82 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
83 def natural_sortkey(string): |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
84 return tuple(int(num) if num else alpha for num, alpha in tokenize(string)) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
85 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
86 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
87 def download_dbnsfp_database(url, output_file): |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
88 dbnsfp_tsv = None |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
89 file_path = 'downloaded_file' |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
90 urllib.urlretrieve(url, file_path) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
91 with zipfile.ZipFile(file_path, 'r') as my_zip: |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
92 dbnsfp_tsv = output_file if output_file else 'dbnsfp_tsv' |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
93 wtr = open(dbnsfp_tsv, 'w') |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
94 allfiles = [info.filename for info in my_zip.infolist()] |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
95 files = [f for f in allfiles if re.match(dbNSFP_file_pat, f)] |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
96 files = sorted(files, key=natural_sortkey) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
97 for j, file in enumerate(files): |
2
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
98 tempfiles = [] |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
99 tempfiles.append(file + "_%d" % len(tempfiles)) |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
100 tfh = open(tempfiles[-1], 'w') |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
101 lastpos = None |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
102 fh = my_zip.open(file, 'rU') |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
103 for i, line in enumerate(fh): |
2
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
104 if i == 0: |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
105 if j == 0: |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
106 wtr.write(line) |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
107 continue |
2
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
108 else: |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
109 pos = int(line.split('\t')[1]) |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
110 if lastpos and pos < lastpos: |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
111 tfh.close() |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
112 tempfiles.append(file + "_%d" % len(tempfiles)) |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
113 tfh = open(tempfiles[-1], 'w') |
7
fe8a9ab8daf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
6
diff
changeset
|
114 print("%s [%d] pos: %d < %d" % (file, i, pos, lastpos), file=sys.stderr) |
2
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
115 lastpos = pos |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
116 tfh.write(line) |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
117 tfh.close() |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
118 if len(tempfiles) == 1: |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
119 with open(tempfiles[0], 'r') as tfh: |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
120 wtr.writelines(tfh.readlines()) |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
121 else: |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
122 tfha = [open(temp, 'r') for temp in tempfiles] |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
123 lines = [tfh.readline() for tfh in tfha] |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
124 curpos = [int(line.split('\t')[1]) for line in lines] |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
125 while len(tfha) > 0: |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
126 k = curpos.index(min(curpos)) |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
127 wtr.write(lines[k]) |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
128 line = tfha[k].readline() |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
129 if line: |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
130 lines[k] = line |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
131 curpos[k] = int(line.split('\t')[1]) |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
132 else: |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
133 tfha[k].close() |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
134 del tfha[k] |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
135 del lines[k] |
3d4cd0e3891f
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 6eea2aa4bb66b329385e1f7324dafae805b555c8
iuc
parents:
0
diff
changeset
|
136 del curpos[k] |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
137 return dbnsfp_tsv |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
138 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
139 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
140 def main(): |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
141 # Parse Command Line |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
142 parser = optparse.OptionParser() |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
143 parser.add_option('-g', '--dbkey', dest='dbkey', action='store', type="string", default=None, help='dbkey genome version') |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
144 parser.add_option('-n', '--db_name', dest='db_name', action='store', type="string", default=None, help='A name for a history snpsiftdbnsfp dataset') |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
145 parser.add_option('-s', '--softgenetics', dest='softgenetics', action='store', type="string", default=None, help='A name for softgenetics dbNSFP file') |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
146 parser.add_option('-H', '--snpsiftdbnsfp', dest='snpsiftdbnsfp', action='store', type="string", default=None, help='A history snpsiftdbnsfp dataset') |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
147 parser.add_option('-T', '--dbnsfp_tabular', dest='dbnsfp_tabular', action='store', type="string", default=None, help='A history dbnsfp_tabular dataset') |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
148 (options, args) = parser.parse_args() |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
149 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
150 filename = args[0] |
7
fe8a9ab8daf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
6
diff
changeset
|
151 with open(filename) as fh: |
fe8a9ab8daf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
6
diff
changeset
|
152 params = json.load(fh) |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
153 target_directory = params['output_data'][0]['extra_files_path'] |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
154 if not os.path.exists(target_directory): |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
155 os.mkdir(target_directory) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
156 data_manager_dict = {} |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
157 genome_version = options.dbkey if options.dbkey else 'unknown' |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
158 dbnsfp_tsv = None |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
159 db_name = None |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
160 bzip_path = None |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
161 if options.softgenetics: |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
162 dbnsfp_url = softgenetics_url + options.softgenetics |
5
09f9bfb2b33b
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit a4b0969b33a68a0ea9ba12291f6694aec24f13ed
iuc
parents:
4
diff
changeset
|
163 db_name = options.db_name if options.db_name else re.sub(r'\.zip$', '', options.softgenetics) |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
164 genome_version = get_nsfp_genome_version(options.softgenetics) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
165 tsv = db_name + '.tsv' |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
166 dbnsfp_tsv = download_dbnsfp_database(dbnsfp_url, tsv) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
167 elif options.dbnsfp_tabular: |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
168 db_name = options.db_name |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
169 dbnsfp_tsv = options.dbnsfp_tabular |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
170 elif options.snpsiftdbnsfp: |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
171 (dirpath, bgzip_name) = os.path.split(options.snpsiftdbnsfp) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
172 idxpath = options.snpsiftdbnsfp + '.tbi' |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
173 shutil.copy(options.snpsiftdbnsfp, target_directory) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
174 shutil.copy(idxpath, target_directory) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
175 bzip_path = os.path.join(target_directory, bgzip_name) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
176 db_name = re.sub('(.txt)?.gz$', '', bgzip_name) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
177 else: |
7
fe8a9ab8daf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
6
diff
changeset
|
178 sys.exit('Either --softgenetics or --dbnsfp_tabular required') |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
179 if dbnsfp_tsv: |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
180 bgzip_name = '%s.txt.gz' % db_name |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
181 bzip_path = os.path.join(target_directory, bgzip_name) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
182 tabix_file(dbnsfp_tsv, bzip_path) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
183 annotations = get_annotations(bzip_path) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
184 # Create the SnpSift dbNSFP Reference Data |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
185 data_table_entry = dict(key='%s_%s' % (genome_version, db_name), build=genome_version, name='%s %s' % (genome_version, db_name), value=bgzip_name, annotations=annotations) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
186 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {}) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
187 data_manager_dict['data_tables'][data_table] = data_manager_dict['data_tables'].get(data_table, []) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
188 data_manager_dict['data_tables'][data_table].append(data_table_entry) |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
189 |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
190 # save info to json file |
7
fe8a9ab8daf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
6
diff
changeset
|
191 with open(filename, 'w') as fh: |
fe8a9ab8daf9
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
6
diff
changeset
|
192 json.dump(data_manager_dict, fh, sort_keys=True) |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
193 |
4
47f276ffcbfc
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
2
diff
changeset
|
194 |
0
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
195 if __name__ == "__main__": |
0e9e3bb5776a
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpsift_dbnsfp commit 5316af00b4a71a7b526cbc9540d5158749cc38e4
iuc
parents:
diff
changeset
|
196 main() |