annotate data_manager/data_manager_snpEff_download.py @ 9:db4c810682ee draft default tip

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
author iuc
date Tue, 24 Nov 2020 07:27:05 +0000
parents fbb459120c29
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
1 #!/usr/bin/env python
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
2 import json
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
3 import optparse
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
4 import os
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
5 import re
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
6 import subprocess
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
7 import sys
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
8
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
9
4
a9cace1ee2b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 5c6d595ecbf2e4e39c25662a165c9e475e01ecd1-dirty
iuc
parents: 3
diff changeset
10 def fetch_databases(genome_list=None):
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
11 snpDBs = dict()
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
12 databases_path = 'databases.out'
4
a9cace1ee2b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 5c6d595ecbf2e4e39c25662a165c9e475e01ecd1-dirty
iuc
parents: 3
diff changeset
13 args = ['snpEff', 'databases']
9
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
14 with open(databases_path, 'w') as databases_output:
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
15 return_code = subprocess.call(args=args, shell=False, stdout=databases_output.fileno())
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
16 if return_code:
8
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
17 sys.exit(return_code)
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
18 try:
9
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
19 with open(databases_path, 'r') as fh:
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
20 for line in fh:
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
21 fields = line.split('\t')
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
22 if len(fields) >= 2:
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
23 genome_version = fields[0].strip()
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
24 if genome_list and genome_version not in genome_list:
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
25 continue
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
26 if genome_version.startswith("Genome") or genome_version.startswith("-"):
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
27 continue
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
28 description = fields[1].strip()
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
29 snpDBs[genome_version] = description
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
30 except Exception as e:
9
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
31 sys.exit('Error parsing %s %s\n' % (databases_path, str(e)))
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
32 return snpDBs
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
33
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
34
4
a9cace1ee2b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 5c6d595ecbf2e4e39c25662a165c9e475e01ecd1-dirty
iuc
parents: 3
diff changeset
35 def getOrganismNames(genomes, organisms):
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
36 genome_list = genomes.split(',')
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
37 organism_list = organisms.split(',') if organisms else []
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
38 if len(genome_list) != len(organism_list):
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
39 descriptions = []
4
a9cace1ee2b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 5c6d595ecbf2e4e39c25662a165c9e475e01ecd1-dirty
iuc
parents: 3
diff changeset
40 snpDBdict = fetch_databases(genome_list=genome_list)
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
41 for genome in snpDBdict:
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
42 descriptions.append(snpDBdict[genome] if genome in snpDBdict else genome)
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
43 return ','.join(descriptions)
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
44 return organisms
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
45
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
46
4
a9cace1ee2b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 5c6d595ecbf2e4e39c25662a165c9e475e01ecd1-dirty
iuc
parents: 3
diff changeset
47 def getSnpeffVersion():
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
48 snpeff_version = 'SnpEff ?.?'
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
49 stderr_path = 'snpeff.err'
4
a9cace1ee2b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 5c6d595ecbf2e4e39c25662a165c9e475e01ecd1-dirty
iuc
parents: 3
diff changeset
50 args = ['snpEff', '-h']
9
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
51 with open(stderr_path, 'w') as stderr_fh:
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
52 return_code = subprocess.call(args=args, shell=False, stderr=stderr_fh.fileno())
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
53 if return_code != 255:
8
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
54 sys.exit(return_code)
9
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
55 with open(stderr_path) as fh:
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
56 for line in fh:
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
57 m = re.match(r'^[Ss]npEff version (SnpEff)\s*(\d+\.\d+).*$', line)
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
58 if m:
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
59 snpeff_version = m.groups()[0] + m.groups()[1]
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
60 break
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
61 return snpeff_version
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
62
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
63
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
64 # Download human database 'hg19'
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
65 # java -jar snpEff.jar download -v hg19
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
66 #
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
67 # <command>java -jar \$SNPEFF_JAR_PATH/snpEff.jar download -c \$JAVA_JAR_PATH/snpEff.config $genomeVersion > $logfile </command>
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
68 #
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
69 # snpEffectPredictor.bin
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
70 # regulation_HeLa-S3.bin
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
71 # regulation_pattern = 'regulation_(.+).bin'
4
a9cace1ee2b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 5c6d595ecbf2e4e39c25662a165c9e475e01ecd1-dirty
iuc
parents: 3
diff changeset
72 def download_database(data_manager_dict, target_directory, genome_version, organism):
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
73 # get data_dir from config
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
74 # ---
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
75 # Databases are stored here
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
76 # E.g.: Information for 'hg19' is stored in data_dir/hg19/
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
77 #
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
78 # Note: Since version 2.1 you can use tilde ('~') as first character to refer to your home directory
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
79 data_dir = target_directory
4
a9cace1ee2b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 5c6d595ecbf2e4e39c25662a165c9e475e01ecd1-dirty
iuc
parents: 3
diff changeset
80 args = ['snpEff', 'download', '-dataDir', data_dir, '-v', genome_version]
a9cace1ee2b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 5c6d595ecbf2e4e39c25662a165c9e475e01ecd1-dirty
iuc
parents: 3
diff changeset
81 return_code = subprocess.call(args=args, shell=False)
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
82 if return_code:
8
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
83 sys.exit(return_code)
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
84 # search data_dir/genome_version for files
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
85 regulation_pattern = 'regulation_(.+).bin'
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
86 genome_path = os.path.join(data_dir, genome_version)
4
a9cace1ee2b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 5c6d595ecbf2e4e39c25662a165c9e475e01ecd1-dirty
iuc
parents: 3
diff changeset
87 snpeff_version = getSnpeffVersion()
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
88 key = snpeff_version + '_' + genome_version
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
89 if os.path.isdir(genome_path):
9
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
90 for _, _, files in os.walk(genome_path):
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
91 for fname in files:
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
92 if fname.startswith('snpEffectPredictor'):
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
93 # if snpEffectPredictor.bin download succeeded
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
94 name = genome_version + (' : ' + organism if organism else '')
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
95 data_table_entry = dict(key=key, version=snpeff_version, value=genome_version, name=name, path=data_dir)
8
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
96 _add_data_table_entry(data_manager_dict, 'snpeffv_genomedb', data_table_entry)
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
97 else:
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
98 m = re.match(regulation_pattern, fname)
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
99 if m:
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
100 name = m.groups()[0]
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
101 data_table_entry = dict(key=key, version=snpeff_version, genome=genome_version, value=name, name=name)
8
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
102 _add_data_table_entry(data_manager_dict, 'snpeffv_regulationdb', data_table_entry)
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
103 return data_manager_dict
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
104
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
105
8
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
106 def _add_data_table_entry(data_manager_dict, data_table, data_table_entry):
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
107 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {})
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
108 data_manager_dict['data_tables'][data_table] = data_manager_dict['data_tables'].get(data_table, [])
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
109 data_manager_dict['data_tables'][data_table].append(data_table_entry)
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
110 return data_manager_dict
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
111
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
112
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
113 def main():
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
114 parser = optparse.OptionParser()
8
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
115 parser.add_option('-g', '--genome_version', dest='genome_version', action='store', type="string", default=None, help='genome_version')
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
116 parser.add_option('-o', '--organism', dest='organism', action='store', type="string", default=None, help='organism name')
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
117 (options, args) = parser.parse_args()
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
118
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
119 filename = args[0]
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
120
9
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
121 with open(filename) as fh:
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
122 params = json.load(fh)
8
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
123 target_directory = params['output_data'][0]['extra_files_path']
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
124 os.mkdir(target_directory)
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
125 data_manager_dict = {}
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
126
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
127 # Create SnpEff Reference Data
4
a9cace1ee2b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 5c6d595ecbf2e4e39c25662a165c9e475e01ecd1-dirty
iuc
parents: 3
diff changeset
128 for genome_version, organism in zip(options.genome_version.split(','), getOrganismNames(options.genome_version, options.organism).split(',')):
8
fbb459120c29 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 036feef4f46b5aedabc5080c8fa4bc00d9c30d95"
iuc
parents: 7
diff changeset
129 download_database(data_manager_dict, target_directory, genome_version, organism)
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
130
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
131 # save info to json file
9
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
132 with open(filename, 'w') as fh:
db4c810682ee "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 00d1775d87af54665ec1871f021bcec6b4b7a15a"
iuc
parents: 8
diff changeset
133 json.dump(data_manager_dict, fh, sort_keys=True)
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
134
3
821c34cdacb6 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_snpeff commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents: 0
diff changeset
135
0
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
136 if __name__ == "__main__":
111fc9457ec6 planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/data_managers/data_manager_snpeff commit 88c982c5adcd32b11d98428fc554a4fdfcc19584
iuc
parents:
diff changeset
137 main()