Mercurial > repos > peterjc > tmhmm_and_signalp

annotate tools/protein_analysis/seq_analysis_utils.py @ 22:e1afa4b0b682 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"This is v0.2.12 with black formating and Python 3 next fix etc"
author peterjc
date Thu, 17 Jun 2021 08:34:58 +0000
parents 238eae32483c
children e1996f0f4e85
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
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
19 if sys.version_info[0] < 3:
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
20 range = xrange # noqa: F821
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
22 __version__ = "0.0.4"
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
23
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
24 try:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
25 from multiprocessing import cpu_count
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
26 except ImportError:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
27 # Must be under Python 2.5, this is copied from multiprocessing:
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
28 def cpu_count():
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
29 """Return the number of CPUs in the system."""
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
30 if sys.platform == "win32":
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
31 try:
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
32 num = int(os.environ["NUMBER_OF_PROCESSORS"])
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
33 except (ValueError, KeyError):
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
34 num = 0
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
35 elif "bsd" in sys.platform or sys.platform == "darwin":
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
36 comm = "/sbin/sysctl -n hw.ncpu"
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
37 if sys.platform == "darwin":
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
38 comm = "/usr" + comm
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
39 try:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
40 with os.popen(comm) as p:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
41 num = int(p.read())
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
42 except ValueError:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
43 num = 0
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
44 else:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
45 try:
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
46 num = os.sysconf("SC_NPROCESSORS_ONLN")
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
47 except (ValueError, OSError, AttributeError):
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
48 num = 0
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
49
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
50 if num >= 1:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
51 return num
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
52 else:
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
53 raise NotImplementedError("cannot determine number of cpus")
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
54
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
55
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
56 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
57 """Determine number of threads to use from the command line args."""
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
58 try:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
59 num = int(command_line_arg)
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
60 except ValueError:
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
61 num = default
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
62 if num < 1:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
63 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
64 # Cap this with the pysical limit of the machine,
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
65 try:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
66 num = min(num, cpu_count())
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
67 except NotImplementedError:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
68 pass
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
69 # For debugging,
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
70 # hostname = os.environ.get("HOSTNAME", "this machine")
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
71 # sys.stderr.write("Using %i cores on %s\n" % (num, hostname))
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
72 return num
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
73
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
74
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
75 def fasta_iterator(filename, max_len=None, truncate=None):
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
76 """Parse FASTA file yielding tuples of (name, sequence)."""
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
77 handle = open(filename)
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
78 title, seq = "", ""
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
79 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
80 if line.startswith(">"):
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
81 if title:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
82 if truncate:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
83 seq = seq[:truncate]
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
84 if max_len and len(seq) > max_len:
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
85 raise ValueError(
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
86 "Sequence %s is length %i, max length %i"
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
87 % (title.split()[0], len(seq), max_len)
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
88 )
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
89 yield title, seq
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
90 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
91 seq = ""
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
92 elif title:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
93 seq += line.strip()
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
94 elif not line.strip() or line.startswith("#"):
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
95 # Ignore blank lines, and any comment lines
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
96 # 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
97 pass
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
98 else:
17
e6cc27d182a8 Uploaded v0.2.6, embedded citations and uses $GALAXY_SLOTS
peterjc
parents: 9
diff changeset
99 handle.close()
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
100 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
101 handle.close()
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
102 if title:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
103 if truncate:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
104 seq = seq[:truncate]
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
105 if max_len and len(seq) > max_len:
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
106 raise ValueError(
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
107 "Sequence %s is length %i, max length %i"
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
108 % (title.split()[0], len(seq), max_len)
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
109 )
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
110 yield title, seq
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
111 raise StopIteration
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
112
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
113
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
114 def split_fasta(
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
115 input_filename,
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
116 output_filename_base,
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
117 n=500,
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
118 truncate=None,
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
119 keep_descr=False,
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
120 max_len=None,
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
121 ):
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
122 """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
123
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
124 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
125 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
126 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
127 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
128
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
129 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
130 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
131 """
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
132 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
133 files = []
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
134 try:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
135 while True:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
136 records = []
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
137 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
138 try:
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
139 records.append(next(iterator))
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
140 except StopIteration:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
141 break
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
142 if not records:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
143 break
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
144 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
145 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
146 if keep_descr:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
147 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
148 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
149 for i in range(0, len(seq), 60):
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
150 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
151 else:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
152 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
153 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
154 for i in range(0, len(seq), 60):
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
155 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
156 handle.close()
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
157 files.append(new_filename)
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
158 # 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
159 except ValueError as err:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
160 # 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
161 try:
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
162 handle.close()
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
163 except Exception:
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
164 pass
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
165 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
166 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
167 os.remove(f)
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
168 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
169 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
170 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
171 return files
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
172
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
173
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
174 def run_jobs(jobs, threads, pause=10, verbose=False, fast_fail=True):
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
175 """Take list of cmd strings, return dict with error levels."""
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
176 pending = jobs[:]
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
177 running = []
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
178 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
179 skipped = []
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
180 if threads == 1:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
181 # Special case this for speed, don't need the waits
9
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
182 for cmd in jobs:
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
183 results[cmd] = subprocess.call(cmd, shell=True)
e52220a9ddad Uploaded v0.1.2
peterjc
parents: 7
diff changeset
184 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
185 failed = False
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
186 while pending or running:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
187 # 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
188 for (cmd, process) in running:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
189 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
190 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
191 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
192 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
193 failed = True
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
194 running = [(cmd, process) for (cmd, process) in running 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
195 if verbose:
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
196 print(
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
197 "%i jobs pending, %i running, %i completed"
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
198 % (len(pending), len(running), len(results))
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
199 )
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
200 # 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
201 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
202 # 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
203 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
204 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
205 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
206 pending = []
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
207 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
208 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
209 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
210 print(cmd)
0
bca9bc7fdaef Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
211 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
212 running.append((cmd, process))
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
213 # Loop...
6
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository
peterjc
parents: 3
diff changeset
214 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
215 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
216 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
217 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
218 return results