Mercurial > repos > cstrittmatter > test_eurl_vtec_wgs_pt

annotate scripts/ReMatCh/modules/download.py @ 3:0cbed1c0a762 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
author cstrittmatter
date Tue, 28 Jan 2020 10:42:31 -0500
parents 965517909457
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
1 import os.path
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
2 import multiprocessing
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
3 import sys
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
4 import functools
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
5 import time
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
6 import subprocess
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
7
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
8 try:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
9 import modules.utils as utils
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
10 except ImportError:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
11 from ReMatCh.modules import utils as utils
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
12
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
13
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
14 def get_read_run_info(ena_id):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
15 import urllib.request
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
16
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
17 url = 'http://www.ebi.ac.uk/ena/data/warehouse/filereport?accession=' + ena_id + '&result=read_run'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
18
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
19 read_run_info = None
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
20 try:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
21 url = urllib.request.urlopen(url)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
22 read_run_info = url.read().decode("utf8").splitlines()
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
23 if len(read_run_info) <= 1:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
24 read_run_info = None
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
25 except Exception as error:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
26 print(error)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
27
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
28 return read_run_info
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
29
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
30
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
31 def get_download_information(read_run_info):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
32 header_line = read_run_info[0].split('\t')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
33 info_line = read_run_info[1].split('\t')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
34
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
35 download_information = {'fastq': None, 'submitted': None, 'cram_index': None}
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
36 download_types = ['aspera', 'ftp']
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
37
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
38 for i in range(0, len(header_line)):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
39 header = header_line[i].lower().rsplit('_', 1)
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
40 if header[0] in list(download_information.keys()):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
41 if header[1] in download_types:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
42 if len(info_line[i]) > 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
43 files_path = info_line[i].split(';')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
44 if len(files_path) > 2:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
45 print('WARNING: Were found more files than expected in'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
46 ' {download_information}-{download_types} download'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
47 ' links!'.format(download_information=header[0], download_types=header[1]))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
48 if download_information[header[0]] is None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
49 download_information[header[0]] = {}
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
50 download_information[header[0]][header[1]] = files_path
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
51
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
52 return download_information
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
53
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
54
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
55 def get_sequencing_information(read_run_info):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
56 header_line = read_run_info[0].split('\t')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
57 info_line = read_run_info[1].split('\t')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
58
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
59 sequencing_information = {'run_accession': None, 'instrument_platform': None, 'instrument_model': None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
60 'library_layout': None, 'library_source': None, 'extra_run_accession': None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
61 'nominal_length': None, 'read_count': None, 'base_count': None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
62 'date_download': time.strftime("%Y-%m-%d")}
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
63
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
64 for i in range(0, len(header_line)):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
65 header = header_line[i].lower()
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
66 if header in list(sequencing_information.keys()):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
67 if len(info_line[i]) > 0:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
68 sequencing_information[header] = info_line[i]
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
69
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
70 if len(read_run_info) > 2:
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
71 extra_run_accession = []
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
72 for i in range(2, len(read_run_info)):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
73 info = read_run_info[i].split('\t')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
74 for j in range(0, len(header_line)):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
75 header = header_line[j].lower()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
76 if header == 'run_accession':
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
77 if len(info[j]) > 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
78 extra_run_accession.append(info[j])
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
79 if len(extra_run_accession) >= 1:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
80 sequencing_information['extra_run_accession'] = ','.join(extra_run_accession)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
81
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
82 return sequencing_information
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
83
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
84
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
85 @utils.trace_unhandled_exceptions
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
86 def download_with_aspera(aspera_file_path, aspera_key, outdir, pickle_prefix, sra, ena_id):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
87 command = ['ascp', '-QT', '-l', '300m', '', '-i', aspera_key, '', outdir]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
88 if not sra:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
89 command[4] = '-P33001'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
90 command[7] = str('era-fasp@' + aspera_file_path)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
91 pickle = pickle_prefix + '.' + aspera_file_path.rsplit('/', 1)[1]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
92 else:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
93 command[7] = 'anonftp@ftp.ncbi.nlm.nih.gov:/sra/sra-instant/reads/ByRun/sra/{a}/{b}/{c}/{c}.sra'.format(
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
94 a=ena_id[:3], b=ena_id[:6], c=ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
95 pickle = pickle_prefix + '.' + ena_id
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
96
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
97 run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, False, 3600, True)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
98
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
99 utils.save_variable_to_pickle(run_successfully, outdir, pickle)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
100
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
101
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
102 @utils.trace_unhandled_exceptions
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
103 def download_with_wget(ftp_file_path, outdir, pickle_prefix, sra, ena_id):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
104 command = ['wget', '--tries=1', '', '-O', '']
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
105 if not sra:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
106 command[2] = ftp_file_path
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
107 file_download = ftp_file_path.rsplit('/', 1)[1]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
108 command[4] = os.path.join(outdir, file_download)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
109 pickle = pickle_prefix + '.' + file_download
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
110 else:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
111 command[2] = 'ftp://ftp-trace.ncbi.nih.gov/sra/sra-instant/reads/ByRun/sra/{a}/{b}/{c}/{c}.sra'.format(
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
112 a=ena_id[:3], b=ena_id[:6], c=ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
113 command[4] = os.path.join(outdir, ena_id + '.sra')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
114 pickle = pickle_prefix + '.' + ena_id
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
115 run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, False, 3600, True)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
116
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
117 utils.save_variable_to_pickle(run_successfully, outdir, pickle)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
118
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
119
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
120 @utils.trace_unhandled_exceptions
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
121 def download_with_sra_prefetch(aspera_key, outdir, pickle_prefix, ena_id):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
122 command = ['prefetch', '', ena_id]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
123
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
124 if aspera_key is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
125 _, ascp, _ = utils.run_command_popen_communicate(['which', 'ascp'], False, None, False)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
126 command[1] = '-a {ascp}|{aspera_key}'.format(ascp=ascp.splitlines()[0], aspera_key=aspera_key)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
127
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
128 run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, False, 3600, True)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
129 if run_successfully:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
130 _, prefetch_outdir, _ = utils.run_command_popen_communicate(['echo', '$HOME/ncbi/public/sra'], True, None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
131 False)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
132
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
133 try:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
134 os.rename(os.path.join(prefetch_outdir.splitlines()[0], ena_id + '.sra'),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
135 os.path.join(outdir, ena_id + '.sra'))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
136 except OSError as e:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
137 print('Found the following error:'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
138 '{}'.format(e))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
139
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
140 from shutil import copy as shutil_copy
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
141
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
142 shutil_copy(os.path.join(prefetch_outdir.splitlines()[0], ena_id + '.sra'),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
143 os.path.join(outdir, ena_id + '.sra'))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
144 os.remove(os.path.join(prefetch_outdir.splitlines()[0], ena_id + '.sra'))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
145
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
146 utils.save_variable_to_pickle(run_successfully, outdir, pickle_prefix + '.' + ena_id)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
147
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
148
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
149 @utils.trace_unhandled_exceptions
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
150 def download_with_curl(ftp_file_path, outdir, pickle_prefix, sra, ena_id):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
151 command = ['curl', '--retry', '1', '', '-o', '']
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
152 if not sra:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
153 command[3] = ftp_file_path
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
154 file_download = ftp_file_path.rsplit('/', 1)[1]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
155 command[5] = os.path.join(outdir, file_download)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
156 pickle = pickle_prefix + '.' + file_download
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
157 else:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
158 command[3] = 'ftp://ftp-trace.ncbi.nih.gov/sra/sra-instant/reads/ByRun/sra/{a}/{b}/{c}/{c}.sra'.format(
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
159 a=ena_id[:3], b=ena_id[:6], c=ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
160 command[5] = os.path.join(outdir, ena_id + '.sra')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
161 pickle = pickle_prefix + '.' + ena_id
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
162 run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, False, 3600, True)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
163
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
164 utils.save_variable_to_pickle(run_successfully, outdir, pickle)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
165
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
166
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
167 def get_pickle_run_successfully(directory, pickle_prefix):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
168 run_successfully = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
169 read_pickle = False
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
170
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
171 files = find_files(directory, pickle_prefix, '.pkl')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
172 if files is not None:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
173 for file_found in files:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
174 if run_successfully:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
175 run_successfully = utils.extract_variable_from_pickle(file_found)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
176 read_pickle = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
177
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
178 os.remove(file_found)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
179
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
180 if not read_pickle:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
181 run_successfully = False
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
182
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
183 return run_successfully
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
184
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
185
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
186 def curl_installed():
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
187 command = ['which', 'curl']
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
188 run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, False, None, False)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
189 return run_successfully
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
190
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
191
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
192 def download(download_information_type, aspera_key, outdir, sra, sra_opt, ena_id):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
193 pickle_prefix = 'download'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
194
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
195 run_successfully = False
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
196 download_sra = False
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
197
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
198 if not sra:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
199 if aspera_key is not None and download_information_type['aspera'] is not None:
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
200 pool = multiprocessing.Pool(processes=2)
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
201 for file_download in download_information_type['aspera']:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
202 pool.apply_async(download_with_aspera, args=(file_download, aspera_key, outdir, pickle_prefix, sra,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
203 ena_id,))
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
204 pool.close()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
205 pool.join()
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
206 run_successfully = get_pickle_run_successfully(outdir, pickle_prefix)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
207 if not run_successfully and download_information_type['ftp'] is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
208 if curl_installed():
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
209 pool = multiprocessing.Pool(processes=2)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
210 for file_download in download_information_type['ftp']:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
211 pool.apply_async(download_with_curl, args=(file_download, outdir, pickle_prefix, sra, ena_id,))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
212 pool.close()
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
213 pool.join()
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
214 run_successfully = get_pickle_run_successfully(outdir, pickle_prefix)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
215 if not run_successfully:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
216 pool = multiprocessing.Pool(processes=2)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
217 for file_download in download_information_type['ftp']:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
218 pool.apply_async(download_with_wget, args=(file_download, outdir, pickle_prefix, sra, ena_id,))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
219 pool.close()
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
220 pool.join()
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
221 run_successfully = get_pickle_run_successfully(outdir, pickle_prefix)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
222
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
223 if not run_successfully and (sra or sra_opt):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
224 if aspera_key is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
225 download_with_aspera(None, aspera_key, outdir, pickle_prefix, sra or sra_opt, ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
226 run_successfully = get_pickle_run_successfully(outdir, pickle_prefix)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
227 if not run_successfully:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
228 download_with_sra_prefetch(aspera_key, outdir, pickle_prefix, ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
229 run_successfully = get_pickle_run_successfully(outdir, pickle_prefix)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
230 if not run_successfully:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
231 if curl_installed():
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
232 download_with_curl(None, outdir, pickle_prefix, sra or sra_opt, ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
233 run_successfully = get_pickle_run_successfully(outdir, pickle_prefix)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
234 if not run_successfully:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
235 download_with_wget(None, outdir, pickle_prefix, sra or sra_opt, ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
236 run_successfully = get_pickle_run_successfully(outdir, pickle_prefix)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
237
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
238 if run_successfully:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
239 download_sra = True
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
240
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
241 return run_successfully, download_sra
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
242
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
243
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
244 def download_files(download_information, aspera_key, outdir, download_cram_bam_true, sra, sra_opt, ena_id):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
245 run_successfully = False
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
246 cram_index_run_successfully = False
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
247 download_sra = False
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
248
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
249 if download_information['fastq'] is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
250 run_successfully, download_sra = download(download_information['fastq'], aspera_key, outdir, sra, sra_opt,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
251 ena_id)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
252
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
253 if not run_successfully:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
254 if download_information['submitted'] is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
255 if not download_cram_bam_true:
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
256 cram_bam = False
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
257 for i in download_information['submitted']:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
258 if download_information['submitted'][i][0].endswith(('.cram', '.bam')):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
259 cram_bam = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
260 break
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
261 if not cram_bam:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
262 run_successfully, download_sra = download(download_information['submitted'], aspera_key, outdir,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
263 False, False, ena_id)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
264
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
265 elif download_cram_bam_true:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
266 run_successfully, download_sra = download(download_information['submitted'], aspera_key, outdir, False,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
267 False, ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
268 if run_successfully and download_information['cram_index'] is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
269 cram_index_run_successfully = download(download_information['cram_index'], aspera_key, outdir,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
270 False, False, ena_id)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
271
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
272 if not run_successfully and (sra or sra_opt):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
273 run_successfully, download_sra = download(download_information['fastq'], aspera_key, outdir, True, sra_opt,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
274 ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
275
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
276 return run_successfully, cram_index_run_successfully, download_sra
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
277
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
278
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
279 def sort_alignment(alignment_file, output_file, sort_by_name_true, threads):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
280 out_format_string = os.path.splitext(output_file)[1][1:].lower()
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
281 command = ['samtools', 'sort', '-o', output_file, '-O', out_format_string, '', '-@', str(threads), alignment_file]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
282 if sort_by_name_true:
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
283 command[6] = '-n'
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
284 run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, False, None, True)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
285
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
286 if not run_successfully:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
287 output_file = None
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
288
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
289 return run_successfully, output_file
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
290
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
291
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
292 def alignment_to_fastq(alignment_file, threads, pair_end_type):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
293 fastq_basename = os.path.splitext(alignment_file)[0]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
294 outfiles = None
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
295 bam_file = fastq_basename + '.temp.bam'
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
296 # sort cram
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
297 run_successfully, bam_file = sort_alignment(alignment_file, bam_file, True, threads)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
298 if run_successfully:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
299 command = ['samtools', 'fastq', '', bam_file]
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
300 if pair_end_type.lower() == 'paired':
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
301 command[2] = '-1 ' + str(fastq_basename + '_1.fq') + ' -2 ' + str(fastq_basename + '_2.fq')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
302 elif pair_end_type == 'single':
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
303 command[2] = '-0 ' + str(fastq_basename + '.fq')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
304
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
305 run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, False, None, True)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
306 if run_successfully:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
307 if pair_end_type.lower() == 'paired':
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
308 outfiles = [str(fastq_basename + '_1.fq'), str(fastq_basename + '_2.fq')]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
309 elif pair_end_type.lower() == 'single':
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
310 outfiles = [str(fastq_basename + '.fq')]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
311
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
312 if bam_file is not None and os.path.isfile(bam_file):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
313 os.remove(bam_file)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
314
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
315 return run_successfully, outfiles
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
316
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
317
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
318 def formart_fastq_headers(in_fastq_1, in_fastq_2):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
319
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
320 out_fastq_1 = in_fastq_1 + '.temp'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
321 out_fastq_2 = in_fastq_2 + '.temp'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
322 writer_in_fastq_1 = open(out_fastq_1, 'wt')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
323 writer_in_fastq_2 = open(out_fastq_2, 'wt')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
324 outfiles = [out_fastq_1, out_fastq_2]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
325 with open(in_fastq_1, 'rtU') as reader_in_fastq_1, open(in_fastq_2, 'rtU') as reader_in_fastq_2:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
326 plus_line = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
327 quality_line = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
328 number_reads = 0
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
329 for in_1, in_2 in zip(reader_in_fastq_1, reader_in_fastq_2):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
330 if len(in_1) > 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
331 in_1 = in_1.splitlines()[0]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
332 in_2 = in_2.splitlines()[0]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
333 if in_1.startswith('@') and plus_line and quality_line:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
334 if in_1 != in_2:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
335 sys.exit('The PE fastq files are not aligned properly!')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
336 in_1 += '/1' + '\n'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
337 in_2 += '/2' + '\n'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
338 writer_in_fastq_1.write(in_1)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
339 writer_in_fastq_2.write(in_2)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
340 plus_line = False
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
341 quality_line = False
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
342 elif in_1.startswith('+') and not plus_line:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
343 in_1 += '\n'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
344 writer_in_fastq_1.write(in_1)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
345 writer_in_fastq_2.write(in_1)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
346 plus_line = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
347 elif plus_line and not quality_line:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
348 in_1 += '\n'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
349 in_2 += '\n'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
350 writer_in_fastq_1.write(in_1)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
351 writer_in_fastq_2.write(in_2)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
352 writer_in_fastq_1.flush()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
353 writer_in_fastq_2.flush()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
354 number_reads += 1
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
355 quality_line = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
356 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
357 in_1 += '\n'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
358 in_2 += '\n'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
359 writer_in_fastq_1.write(in_1)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
360 writer_in_fastq_2.write(in_2)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
361 return number_reads, outfiles
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
362
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
363
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
364 @utils.trace_unhandled_exceptions
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
365 def gzip_files(file_2_compress, pickle_prefix, outdir):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
366 if file_2_compress.endswith('.temp'):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
367 out_file = os.path.splitext(file_2_compress)[0]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
368 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
369 out_file = file_2_compress
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
370
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
371 command = ['gzip', '--stdout', '--best', file_2_compress, '>', str(out_file + '.gz')]
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
372 run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, True, None, True)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
373 if run_successfully:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
374 os.remove(file_2_compress)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
375
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
376 utils.save_variable_to_pickle(run_successfully, outdir,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
377 str(pickle_prefix + '.' + os.path.basename(file_2_compress)))
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
378
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
379
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
380 def find_files(directory, prefix, suffix):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
381 list_files_found = []
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
382 files = [f for f in os.listdir(directory) if not f.startswith('.') and os.path.isfile(os.path.join(directory, f))]
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
383 for file_found in files:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
384 if file_found.startswith(prefix) and file_found.endswith(suffix):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
385 file_path = os.path.join(directory, file_found)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
386 list_files_found.append(file_path)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
387
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
388 if len(list_files_found) == 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
389 list_files_found = None
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
390
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
391 return list_files_found
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
392
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
393
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
394 def compress_files(fastq_files, outdir, threads):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
395 pickle_prefix = 'compress'
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
396 compressed_fastq_files = None
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
397
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
398 pool = multiprocessing.Pool(processes=threads)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
399 for fastq in fastq_files:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
400 pool.apply_async(gzip_files, args=(fastq, pickle_prefix, outdir,))
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
401 pool.close()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
402 pool.join()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
403
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
404 run_successfully = get_pickle_run_successfully(outdir, pickle_prefix)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
405 if run_successfully:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
406 compressed_fastq_files = find_files(outdir, '', '.gz')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
407
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
408 return run_successfully, compressed_fastq_files
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
409
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
410
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
411 def bam_cram_2_fastq(alignment_file, outdir, threads, pair_end_type):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
412 run_successfully, fastq_files = alignment_to_fastq(alignment_file, threads, pair_end_type)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
413 if run_successfully:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
414 if pair_end_type.lower() == 'paired':
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
415 number_reads, fastq_files = formart_fastq_headers(fastq_files[0], fastq_files[1])
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
416
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
417 run_successfully, fastq_files = compress_files(fastq_files, outdir, threads)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
418
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
419 return run_successfully, fastq_files
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
420
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
421
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
422 def check_correct_links(download_information):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
423 for i in download_information:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
424 if download_information[i] is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
425 if download_information[i]['aspera'] is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
426 for j in range(0, len(download_information[i]['aspera'])):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
427 if download_information[i]['aspera'][j].startswith('fasp.sra.ebi.ac.uk/'):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
428 download_information[i]['aspera'][j] = download_information[i]['aspera'][j].replace(
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
429 'fasp.sra.ebi.ac.uk/', 'fasp.sra.ebi.ac.uk:/', 1)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
430 if download_information[i]['ftp'] is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
431 for j in range(0, len(download_information[i]['ftp'])):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
432 if '#' in download_information[i]['ftp'][j]:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
433 download_information[i]['ftp'][j] = download_information[i]['ftp'][j].replace('#', '%23')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
434 return download_information
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
435
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
436
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
437 def get_fastq_files(download_dir, cram_index_run_successfully, threads, download_paired_type):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
438 run_successfully = False
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
439 downloaded_files = find_files(download_dir, '', '')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
440 if cram_index_run_successfully:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
441 cram_file = None
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
442 for i in downloaded_files:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
443 if i.endswith('.cram'):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
444 cram_file = i
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
445 run_successfully, downloaded_files = bam_cram_2_fastq(cram_file, download_dir, threads, download_paired_type)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
446 else:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
447 if downloaded_files is not None and len(downloaded_files) > 0:
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
448 run_successfully = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
449
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
450 return run_successfully, downloaded_files
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
451
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
452
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
453 def rename_move_files(list_files, new_name, outdir, download_paired_type):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
454 list_new_files = {}
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
455 run_successfully = False
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
456
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
457 for i in range(0, len(list_files)):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
458 temp_name = utils.rchop(os.path.basename(list_files[i]), 'astq.gz')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
459 if len(temp_name) == len(os.path.basename(list_files[i])):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
460 temp_name = utils.rchop(os.path.basename(list_files[i]), 'q.gz')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
461 if download_paired_type.lower() == 'paired':
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
462 if temp_name.endswith(('_R1_001.f', '_1.f')):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
463 list_new_files[i] = os.path.join(outdir, new_name + '_1.fq.gz')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
464 elif temp_name.endswith(('_R2_001.f', '_2.f')):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
465 list_new_files[i] = os.path.join(outdir, new_name + '_2.fq.gz')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
466 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
467 if not temp_name.endswith(('_R1_001.f', '_R2_001.f')):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
468 list_new_files[i] = os.path.join(outdir, new_name + '.fq.gz')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
469 if temp_name.endswith(('_1.f', '_2.f')):
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
470 print('WARNING: possible single-end file conflict with pair-end (' + list_files[i] + ')!')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
471
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
472 if len(list_new_files) == 2 and download_paired_type.lower() == 'paired':
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
473 run_successfully = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
474 elif len(list_new_files) == 1 and download_paired_type.lower() == 'single':
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
475 run_successfully = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
476
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
477 if run_successfully:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
478 try:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
479 for i in range(0, len(list_files)):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
480 if i not in list_new_files:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
481 if os.path.isfile(list_files[i]):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
482 os.remove(list_files[i])
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
483 else:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
484 os.rename(list_files[i], list_new_files[i])
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
485 list_new_files = list(list_new_files.values())
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
486 except Exception as e:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
487 print(e)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
488 run_successfully = False
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
489
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
490 if not run_successfully:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
491 list_new_files = None
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
492
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
493 return run_successfully, list_new_files
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
494
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
495
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
496 # @utils.trace_unhandled_exceptions
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
497 def rename_header_sra(fastq):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
498 run_successfully = False
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
499 try:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
500 command = ['gawk', '\'{if(NR%4==1) $0=gensub(/\./, \"/\", 2); print}\'', fastq, '|', 'gzip', '-1', '>',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
501 str(fastq + '.gz')]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
502 print('Running: ' + str(' '.join(command)))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
503 return_code = subprocess.call(' '.join(command), shell=True)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
504 if return_code == 0:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
505 run_successfully = True
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
506 else:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
507 print('Something went wrong with command: {command}'.format(command=' '.join(command)))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
508 except Exception as e:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
509 print(e)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
510
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
511 return run_successfully
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
512
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
513
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
514 def sra_2_fastq(download_dir, ena_id):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
515 command = ['fastq-dump', '-I', '-O', download_dir, '--split-files', '{download_dir}{ena_id}.sra'.format(
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
516 download_dir=download_dir, ena_id=ena_id)]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
517 run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, False, 3600, True)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
518 if run_successfully:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
519 files = [os.path.join(download_dir, f) for f in os.listdir(download_dir)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
520 if not f.startswith('.') and os.path.isfile(os.path.join(download_dir, f)) and f.endswith('.fastq')]
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
521
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
522 pool = multiprocessing.Pool(processes=2)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
523 results = []
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
524 p = pool.map_async(rename_header_sra, files, callback=results.extend)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
525 p.wait()
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
526
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
527 run_successfully = all(results)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
528
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
529 return run_successfully
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
530
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
531
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
532 download_timer = functools.partial(utils.timer, name='Download module')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
533
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
534
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
535 @download_timer
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
536 def run_download(ena_id, download_paired_type, aspera_key, outdir, download_cram_bam_true, threads, instrument_platform,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
537 sra, sra_opt):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
538 download_dir = os.path.join(outdir, 'download', '')
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
539 utils.remove_directory(download_dir)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
540 os.mkdir(download_dir)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
541
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
542 run_successfully = False
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
543 downloaded_files = None
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
544 sequencing_information = {'run_accession': None, 'instrument_platform': None, 'instrument_model': None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
545 'library_layout': None, 'library_source': None, 'extra_run_accession': None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
546 'nominal_length': None, 'read_count': None, 'base_count': None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
547 'date_download': time.strftime("%Y-%m-%d")}
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
548
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
549 read_run_info = get_read_run_info(ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
550 if read_run_info is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
551 download_information = get_download_information(read_run_info)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
552 download_information = check_correct_links(download_information)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
553 sequencing_information = get_sequencing_information(read_run_info)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
554
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
555 if instrument_platform.lower() == 'all' or \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
556 (sequencing_information['instrument_platform'] is not None and
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
557 sequencing_information['instrument_platform'].lower() == instrument_platform.lower()):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
558 if download_paired_type.lower() == 'both' or \
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
559 (sequencing_information['library_layout'] is not None and
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
560 sequencing_information['library_layout'].lower() == download_paired_type.lower()):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
561 run_successfully, cram_index_run_successfully, download_sra = download_files(download_information,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
562 aspera_key, download_dir,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
563 download_cram_bam_true,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
564 sra, sra_opt, ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
565 if download_sra:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
566 run_successfully = sra_2_fastq(download_dir, ena_id)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
567 if run_successfully:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
568 run_successfully, downloaded_files = get_fastq_files(download_dir, cram_index_run_successfully,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
569 threads,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
570 sequencing_information['library_layout'])
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
571 if run_successfully and downloaded_files is not None:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
572 run_successfully, downloaded_files = rename_move_files(downloaded_files,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
573 sequencing_information['run_accession'],
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
574 outdir,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
575 sequencing_information['library_layout'])
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
576 else:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
577 if sra or sra_opt:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
578 run_successfully, cram_index_run_successfully, download_sra = download_files({'fastq': None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
579 'submitted': None,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
580 'cram_index': None},
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
581 aspera_key, download_dir,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
582 download_cram_bam_true, sra,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
583 sra_opt, ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
584 if download_sra:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
585 run_successfully = sra_2_fastq(download_dir, ena_id)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
586 if run_successfully:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
587 run_successfully, downloaded_files = get_fastq_files(download_dir, cram_index_run_successfully, threads,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
588 'paired')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
589 if not run_successfully:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
590 run_successfully, downloaded_files = get_fastq_files(download_dir, cram_index_run_successfully,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
591 threads, 'single')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
592 if run_successfully and downloaded_files is not None:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
593 run_successfully, downloaded_files = rename_move_files(downloaded_files, ena_id, outdir, 'paired')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
594 if not run_successfully:
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
595 run_successfully, downloaded_files = rename_move_files(downloaded_files, ena_id, outdir, 'single')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
596
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
597 utils.remove_directory(download_dir)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
598
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
599 return run_successfully, downloaded_files, sequencing_information