annotate data_manager/rnastar_index_builder.py @ 14:70316134a316 draft default tip

planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit fbb3a55395065d6ea8ed23189fa65813b93d52a3
author sanbi-uwc
date Wed, 09 Mar 2016 05:45:58 -0500
parents e905fcc6fd15
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
1 #!/usr/bin/env python
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
2
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
3 from __future__ import print_function
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
4 import argparse
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
5 from subprocess import check_call, CalledProcessError
5
aed097239724 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 9d7fb6d7c47e94fb113e0fe26427086935122ff0
sanbi-uwc
parents: 0
diff changeset
6 from json import load, dump, dumps
aed097239724 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 9d7fb6d7c47e94fb113e0fe26427086935122ff0
sanbi-uwc
parents: 0
diff changeset
7 from os import environ, mkdir, makedirs
0
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
8 from os.path import isdir, exists
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
9 import shlex
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
10 import sys
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
11
12
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
12
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
13 def get_id_name(params, dbkey, fasta_description=None):
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
14 # TODO: ensure sequence_id is unique and does not already appear in location file
0
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
15 sequence_id = params['param_dict']['sequence_id']
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
16 if not sequence_id:
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
17 sequence_id = dbkey
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
18
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
19 sequence_name = params['param_dict']['sequence_name']
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
20 if not sequence_name:
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
21 sequence_name = fasta_description
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
22 if not sequence_name:
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
23 sequence_name = dbkey
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
24 return sequence_id, sequence_name
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
25
12
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
26
0
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
27 def make_rnastar_index(output_directory, fasta_filename):
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
28 if exists(output_directory) and not isdir(output_directory):
12
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
29 print("Output directory path already exists but is not a directory: {}".format(output_directory),
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
30 file=sys.stderr)
0
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
31 elif not exists(output_directory):
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
32 mkdir(output_directory)
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
33
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
34 if 'GALAXY_SLOTS' in environ:
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
35 nslots = environ['GALAXY_SLOTS']
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
36 else:
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
37 nslots = 1
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
38
13
e905fcc6fd15 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 4c88ff739e2fb47369e5256151eb55cc5cb47c9d
sanbi-uwc
parents: 12
diff changeset
39 cmdline_str = 'STAR --runMode genomeGenerate --genomeDir {} --genomeFastaFiles {} --runThreadN {}'.format(
12
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
40 output_directory,
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
41 fasta_filename,
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
42 nslots)
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
43 cmdline = shlex.split(cmdline_str)
0
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
44 try:
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
45 check_call(cmdline)
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
46 except CalledProcessError:
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
47 print("Error building RNA STAR index", file=sys.stderr)
12
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
48 return (output_directory)
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
49
0
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
50
12
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
51 def main():
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
52 parser = argparse.ArgumentParser(description="Generate RNA STAR genome index and JSON describing this")
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
53 parser.add_argument('output_filename')
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
54 parser.add_argument('--fasta_filename')
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
55 parser.add_argument('--fasta_dbkey')
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
56 parser.add_argument('--fasta_description', default=None)
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
57 parser.add_argument('--data_table_name', default='rnastar_index')
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
58 args = parser.parse_args()
0
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
59
12
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
60 filename = args.output_filename
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
61
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
62 params = load(open(filename, 'rb'))
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
63 output_directory = params['output_data'][0]['extra_files_path']
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
64 makedirs(output_directory)
0
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
65
12
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
66 make_rnastar_index(output_directory, args.fasta_filename)
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
67 (sequence_id, sequence_name) = get_id_name(params, args.fasta_dbkey, args.fasta_description)
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
68 data_table_entry = dict(value=sequence_id, dbkey=args.fasta_dbkey, name=sequence_name, path=output_directory)
0
2883a3b7dc56 planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 265fa1966ea606ebccea8c2865043c014959c10c-dirty
sanbi-uwc
parents:
diff changeset
69
12
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
70 output_datatable_dict = dict(data_tables={args.data_table_name: [data_table_entry]})
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
71 open(filename, 'wb').write(dumps(output_datatable_dict))
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
72
a42dbab7917c planemo upload for repository https://github.com/pvanheus/data_manager_rnastar_index_builder commit 599e0c3b37869cc016156f7058e16d6c268bc703
sanbi-uwc
parents: 11
diff changeset
73 if __name__ == "__main__": main()