Mercurial > repos > devteam > data_manager_gatk_picard_index_builder
annotate data_manager/data_manager_gatk_picard_index_builder.py @ 7:0f85d523291e draft default tip
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit a54e50916ebdb0858b049cdc549a43df801e1a71
author | iuc |
---|---|
date | Tue, 21 May 2024 07:36:15 +0000 |
parents | 6287d0ee459e |
children |
rev | line source |
---|---|
0 | 1 #!/usr/bin/env python |
2 # Dave B. | |
3 # Uses fasta sorting functions written by Dan Blankenberg. | |
4 | |
5 import json | |
6 import optparse | |
7 import os | |
8 import shutil | |
9 import subprocess | |
10 import sys | |
11 import tempfile | |
12 | |
13 CHUNK_SIZE = 2**20 | |
14 DEFAULT_DATA_TABLE_NAME = "fasta_indexes" | |
15 | |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
16 |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
17 def get_id_name(params, dbkey, fasta_description=None): |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
18 # TODO: ensure sequence_id is unique and does not already appear in location file |
0 | 19 sequence_id = params['param_dict']['sequence_id'] |
20 if not sequence_id: | |
21 sequence_id = dbkey | |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
22 |
0 | 23 sequence_name = params['param_dict']['sequence_name'] |
24 if not sequence_name: | |
25 sequence_name = fasta_description | |
26 if not sequence_name: | |
27 sequence_name = dbkey | |
28 return sequence_id, sequence_name | |
29 | |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
30 |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
31 def build_picard_index(data_manager_dict, fasta_filename, target_directory, dbkey, sequence_id, sequence_name, jar, data_table_name=DEFAULT_DATA_TABLE_NAME): |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
32 fasta_base_name = os.path.split(fasta_filename)[-1] |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
33 gatk_sorted_fasta_filename = os.path.join(target_directory, fasta_base_name) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
34 shutil.copy(fasta_filename, gatk_sorted_fasta_filename) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
35 _sort_fasta_gatk(gatk_sorted_fasta_filename) |
0 | 36 sam_index_filename = '%s.fai' % gatk_sorted_fasta_filename |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
37 if not os.path.exists(sam_index_filename): |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
38 sam_command = ['samtools', 'faidx', gatk_sorted_fasta_filename] |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
39 _run_command(sam_command, target_directory) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
40 args = ['java', '-jar', jar, 'R=%s' % gatk_sorted_fasta_filename, 'O=%s.dict' % sequence_id] |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
41 _run_command(args, target_directory) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
42 data_table_entry = dict(value=sequence_id, dbkey=dbkey, name=sequence_name, path=fasta_base_name) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
43 _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry) |
0 | 44 |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
45 |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
46 def _run_command(command, target_directory): |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
47 tmp_stderr = tempfile.NamedTemporaryFile(prefix="tmp-data-manager-gatk_picard_index_builder-stderr") |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
48 proc = subprocess.Popen(args=command, shell=False, cwd=target_directory, stderr=tmp_stderr.fileno()) |
0 | 49 return_code = proc.wait() |
50 if return_code: | |
51 tmp_stderr.flush() | |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
52 tmp_stderr.seek(0) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
53 sys.stderr.write("Error building index:\n") |
0 | 54 while True: |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
55 chunk = tmp_stderr.read(CHUNK_SIZE) |
0 | 56 if not chunk: |
57 break | |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
58 sys.stderr.write(chunk) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
59 sys.exit(return_code) |
0 | 60 tmp_stderr.close() |
61 | |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
62 |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
63 def _add_data_table_entry(data_manager_dict, data_table_name, data_table_entry): |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
64 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {}) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
65 data_manager_dict['data_tables'][data_table_name] = data_manager_dict['data_tables'].get(data_table_name, []) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
66 data_manager_dict['data_tables'][data_table_name].append(data_table_entry) |
0 | 67 return data_manager_dict |
68 | |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
69 |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
70 def _move_and_index_fasta_for_sorting(fasta_filename): |
0 | 71 unsorted_filename = tempfile.NamedTemporaryFile().name |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
72 shutil.move(fasta_filename, unsorted_filename) |
0 | 73 fasta_offsets = {} |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
74 with open(unsorted_filename) as unsorted_fh: |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
75 while True: |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
76 offset = unsorted_fh.tell() |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
77 line = unsorted_fh.readline() |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
78 if not line: |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
79 break |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
80 if line.startswith(">"): |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
81 line = line.split(None, 1)[0][1:] |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
82 fasta_offsets[line] = offset |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
83 current_order = [x[1] for x in sorted((x[1], x[0]) for x in fasta_offsets.items())] |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
84 return (unsorted_filename, fasta_offsets, current_order) |
0 | 85 |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
86 |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
87 def _write_sorted_fasta(sorted_names, fasta_offsets, sorted_fasta_filename, unsorted_fasta_filename): |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
88 with open(unsorted_fasta_filename, 'rb') as unsorted_fh, open(sorted_fasta_filename, 'wb+') as sorted_fh: |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
89 for name in sorted_names: |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
90 offset = fasta_offsets[name] |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
91 unsorted_fh.seek(offset) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
92 sorted_fh.write(unsorted_fh.readline()) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
93 while True: |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
94 line = unsorted_fh.readline() |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
95 if not line or line.startswith(b">"): |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
96 break |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
97 sorted_fh.write(line) |
0 | 98 |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
99 |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
100 def _int_to_roman(integer): |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
101 if not isinstance(integer, int): |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
102 raise TypeError("expected integer, got %s" % type(integer)) |
0 | 103 if not 0 < integer < 4000: |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
104 raise ValueError("Argument must be between 1 and 3999, got %s" % str(integer)) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
105 ints = (1000, 900, 500, 400, 100, 90, 50, 40, 10, 9, 5, 4, 1) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
106 nums = ('M', 'CM', 'D', 'CD', 'C', 'XC', 'L', 'XL', 'X', 'IX', 'V', 'IV', 'I') |
0 | 107 result = "" |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
108 for i in range(len(ints)): |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
109 count = int(integer / ints[i]) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
110 result += nums[i] * count |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
111 integer -= ints[i] * count |
0 | 112 return result |
113 | |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
114 |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
115 def _sort_fasta_gatk(fasta_filename): |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
116 (unsorted_filename, fasta_offsets, current_order) = _move_and_index_fasta_for_sorting(fasta_filename) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
117 sorted_names = list(map(str, range(1, 100))) + list(map(_int_to_roman, range(1, 100))) + ['X', 'Y', 'M'] |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
118 # detect if we have chrN, or just N |
0 | 119 has_chr = False |
120 for chrom in sorted_names: | |
121 if "chr%s" % chrom in current_order: | |
122 has_chr = True | |
123 break | |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
124 |
0 | 125 if has_chr: |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
126 sorted_names = ["chr%s" % x for x in sorted_names] |
0 | 127 else: |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
128 sorted_names.insert(0, "MT") |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
129 sorted_names.extend(["%s_random" % x for x in sorted_names]) |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
130 |
0 | 131 existing_sorted_names = [] |
132 for name in sorted_names: | |
133 # Append each chromosome only once. | |
134 if name in current_order and name not in existing_sorted_names: | |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
135 existing_sorted_names.append(name) |
0 | 136 for name in current_order: |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
137 # TODO: confirm that non-canonical names do not need to be sorted specially |
0 | 138 if name not in existing_sorted_names: |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
139 existing_sorted_names.append(name) |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
140 |
0 | 141 if existing_sorted_names == current_order: |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
142 shutil.move(unsorted_filename, fasta_filename) |
0 | 143 else: |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
144 _write_sorted_fasta(existing_sorted_names, fasta_offsets, fasta_filename, unsorted_filename) |
0 | 145 |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
146 |
0 | 147 def main(): |
148 parser = optparse.OptionParser() | |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
149 parser.add_option('-f', '--fasta_filename', dest='fasta_filename', action='store', type="string", default=None, help='fasta_filename') |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
150 parser.add_option('-d', '--fasta_dbkey', dest='fasta_dbkey', action='store', type="string", default=None, help='fasta_dbkey') |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
151 parser.add_option('-t', '--fasta_description', dest='fasta_description', action='store', type="string", default=None, help='fasta_description') |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
152 parser.add_option('-n', '--data_table_name', dest='data_table_name', action='store', type="string", default=None, help='data_table_name') |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
153 parser.add_option('-j', '--jar', dest='jar', action='store', type="string", default=None, help='GATK .jar file') |
0 | 154 (options, args) = parser.parse_args() |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
155 |
0 | 156 filename = args[0] |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
157 |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
158 with open(filename) as fh: |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
159 params = json.load(fh) |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
160 target_directory = params['output_data'][0]['extra_files_path'] |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
161 os.mkdir(target_directory) |
0 | 162 data_manager_dict = {} |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
163 |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
164 if options.fasta_dbkey in [None, '', '?']: |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
165 raise Exception('"%s" is not a valid dbkey. You must specify a valid dbkey.' % (options.fasta_dbkey)) |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
166 |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
167 sequence_id, sequence_name = get_id_name(params, dbkey=options.fasta_dbkey, fasta_description=options.fasta_description) |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
168 |
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
169 # build the index |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
170 build_picard_index(data_manager_dict, |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
171 options.fasta_filename, |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
172 target_directory, |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
173 options.fasta_dbkey, |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
174 sequence_id, |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
175 sequence_name, |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
176 options.jar, |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
177 data_table_name=options.data_table_name or DEFAULT_DATA_TABLE_NAME) |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
178 |
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
179 # save info to json file |
4
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
180 with open(filename, 'w') as fh: |
6287d0ee459e
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 02d2967f77e3fa5a18aea63dc84aa9ab418dc165"
iuc
parents:
3
diff
changeset
|
181 json.dump(data_manager_dict, fh, sort_keys=True) |
3
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
182 |
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
183 |
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
184 if __name__ == "__main__": |
b31f1fcb203c
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_gatk_picard_index_builder commit 8652f36a3a3838dca989426961561e81432acf4f
iuc
parents:
0
diff
changeset
|
185 main() |