Mercurial > repos > peterjc > tmhmm_and_signalp

annotate tools/protein_analysis/seq_analysis_utils.py @ 20:a19b3ded8f33 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
author peterjc
date Thu, 21 Sep 2017 11:35:20 -0400
parents f3ecd80850e2
children 238eae32483c
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
1 """A few useful functions for working with FASTA files and running jobs.
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
2
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
3 This module was originally written to hold common code used in both the TMHMM
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
4 and SignalP wrappers in Galaxy.
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
5
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
6 Given Galaxy currently supports Python 2.4+ this cannot use the Python module
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
7 multiprocessing so the function run_jobs instead is a simple pool approach
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
8 using just the subprocess library.
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
9 """
20
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
10
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
11 from __future__ import print_function
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
12
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
13 import os
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
14 import subprocess
20
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
15 import sys
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
16
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
17 from time import sleep
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
18
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
19 __version__ = "0.0.2"
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
20
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
21 try:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
22 from multiprocessing import cpu_count
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
23 except ImportError:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
24 # Must be under Python 2.5, this is copied from multiprocessing:
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
25 def cpu_count():
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
26 """Returns the number of CPUs in the system."""
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
27 if sys.platform == 'win32':
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
28 try:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
29 num = int(os.environ['NUMBER_OF_PROCESSORS'])
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
30 except (ValueError, KeyError):
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
31 num = 0
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
32 elif 'bsd' in sys.platform or sys.platform == 'darwin':
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
33 comm = '/sbin/sysctl -n hw.ncpu'
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
34 if sys.platform == 'darwin':
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
35 comm = '/usr' + comm
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
36 try:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
37 with os.popen(comm) as p:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
38 num = int(p.read())
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
39 except ValueError:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
40 num = 0
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
41 else:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
42 try:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
43 num = os.sysconf('SC_NPROCESSORS_ONLN')
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
44 except (ValueError, OSError, AttributeError):
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
45 num = 0
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
46
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
47 if num >= 1:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
48 return num
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
49 else:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
50 raise NotImplementedError('cannot determine number of cpus')
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
51
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
52
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
53 def thread_count(command_line_arg, default=1):
20
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
54 """Determine number of threads to use from the command line args."""
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
55 try:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
56 num = int(command_line_arg)
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
57 except ValueError:
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
58 num = default
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
59 if num < 1:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
60 sys.exit("Threads argument %r is not a positive integer" % command_line_arg)
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
61 # Cap this with the pysical limit of the machine,
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
62 try:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
63 num = min(num, cpu_count())
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
64 except NotImplementedError:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
65 pass
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
66 # For debugging,
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
67 # hostname = os.environ.get("HOSTNAME", "this machine")
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
68 # sys.stderr.write("Using %i cores on %s\n" % (num, hostname))
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
69 return num
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
70
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
71
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
72 def fasta_iterator(filename, max_len=None, truncate=None):
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
73 """Simple FASTA parser yielding tuples of (title, sequence) strings."""
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
74 handle = open(filename)
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
75 title, seq = "", ""
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
76 for line in handle:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
77 if line.startswith(">"):
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
78 if title:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
79 if truncate:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
80 seq = seq[:truncate]
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
81 if max_len and len(seq) > max_len:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
82 raise ValueError("Sequence %s is length %i, max length %i"
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
83 % (title.split()[0], len(seq), max_len))
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
84 yield title, seq
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
85 title = line[1:].rstrip()
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
86 seq = ""
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
87 elif title:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
88 seq += line.strip()
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
89 elif not line.strip() or line.startswith("#"):
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
90 # Ignore blank lines, and any comment lines
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
91 # between records (starting with hash).
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
92 pass
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
93 else:
17
e6cc27d182a8 Uploaded v0.2.6, embedded citations and uses $GALAXY_SLOTS
peterjc
parents: 9
diff changeset
94 handle.close()
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
95 raise ValueError("Bad FASTA line %r" % line)
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
96 handle.close()
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
97 if title:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
98 if truncate:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
99 seq = seq[:truncate]
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
100 if max_len and len(seq) > max_len:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
101 raise ValueError("Sequence %s is length %i, max length %i"
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
102 % (title.split()[0], len(seq), max_len))
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
103 yield title, seq
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
104 raise StopIteration
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
105
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
106
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
107 def split_fasta(input_filename, output_filename_base, n=500, truncate=None, keep_descr=False, max_len=None):
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
108 """Split FASTA file into sub-files each of at most n sequences.
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
109
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
110 Returns a list of the filenames used (based on the input filename).
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
111 Each sequence can also be truncated (since we only need the start for
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
112 SignalP), and have its description discarded (since we don't usually
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
113 care about it and some tools don't like very long title lines).
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
114
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
115 If a max_len is given and any sequence exceeds it no temp files are
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
116 created and an exception is raised.
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
117 """
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
118 iterator = fasta_iterator(input_filename, max_len, truncate)
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
119 files = []
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
120 try:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
121 while True:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
122 records = []
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
123 for i in range(n):
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
124 try:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
125 records.append(iterator.next())
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
126 except StopIteration:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
127 break
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
128 if not records:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
129 break
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
130 new_filename = "%s.%i.tmp" % (output_filename_base, len(files))
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
131 handle = open(new_filename, "w")
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
132 if keep_descr:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
133 for title, seq in records:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
134 handle.write(">%s\n" % title)
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
135 for i in range(0, len(seq), 60):
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
136 handle.write(seq[i:i + 60] + "\n")
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
137 else:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
138 for title, seq in records:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
139 handle.write(">%s\n" % title.split()[0])
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
140 for i in range(0, len(seq), 60):
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
141 handle.write(seq[i:i + 60] + "\n")
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
142 handle.close()
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
143 files.append(new_filename)
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
144 # print "%i records in %s" % (len(records), new_filename)
20
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
145 except ValueError as err:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
146 # Max length failure from parser - clean up
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
147 try:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
148 handle.close()
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
149 except Exception:
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
150 pass
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
151 for f in files:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
152 if os.path.isfile(f):
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
153 os.remove(f)
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
154 raise err
7
9b45a8743100 Uploaded v0.1.0, which adds a wrapper for Promoter 2.0 (DNA tool) and enables use of Galaxy's <parallelism> tag for SignalP, TMHMM X Promoter wrappers.
peterjc
parents: 6
diff changeset
155 for f in files:
9b45a8743100 Uploaded v0.1.0, which adds a wrapper for Promoter 2.0 (DNA tool) and enables use of Galaxy's <parallelism> tag for SignalP, TMHMM X Promoter wrappers.
peterjc
parents: 6
diff changeset
156 assert os.path.isfile(f), "Missing split file %r (!??)" % f
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
157 return files
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
158
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
159
20
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
160 def run_jobs(jobs, threads, pause=10, verbose=False, fast_fail=True):
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
161 """Takes list of cmd strings, returns dict with error levels."""
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
162 pending = jobs[:]
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
163 running = []
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
164 results = {}
20
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
165 skipped = []
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
166 if threads == 1:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
167 # Special case this for speed, don't need the waits
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
168 for cmd in jobs:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
169 results[cmd] = subprocess.call(cmd, shell=True)
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
170 return results
20
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
171 failed = False
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
172 while pending or running:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
173 # See if any have finished
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
174 for (cmd, process) in running:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
175 return_code = process.poll() # non-blocking
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
176 if return_code is not None:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
177 results[cmd] = return_code
20
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
178 if return_code:
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
179 failed = True
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
180 running = [(cmd, process) for (cmd, process) in running
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
181 if cmd not in results]
3
f3b373a41f81 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
182 if verbose:
20
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
183 print("%i jobs pending, %i running, %i completed" %
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
184 (len(pending), len(running), len(results)))
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
185 # See if we can start any new threads
20
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
186 if pending and failed and fast_fail:
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
187 # Don't start any more jobs
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
188 if verbose:
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
189 print("Failed, will not start remaining %i jobs" % len(pending))
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
190 skipped = pending
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
191 pending = []
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
192 while pending and len(running) < threads:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
193 cmd = pending.pop(0)
3
f3b373a41f81 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
194 if verbose:
20
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
195 print(cmd)
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
196 process = subprocess.Popen(cmd, shell=True)
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
197 running.append((cmd, process))
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
198 # Loop...
6
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository
peterjc
parents: 3
diff changeset
199 sleep(pause)
3
f3b373a41f81 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
200 if verbose:
20
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
201 print("%i jobs completed" % len(results))
a19b3ded8f33 v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 19
diff changeset
202 assert set(jobs) == set(results).union(skipped)
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
203 return results