Mercurial > repos > peterjc > tmhmm_and_signalp

annotate tools/protein_analysis/promoter2.py @ 23:e1996f0f4e85 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"v0.2.13 - Python 3 fix for raising StopIteration"
author peterjc
date Thu, 17 Jun 2021 17:59:33 +0000
parents 238eae32483c
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
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:
diff changeset
1 #!/usr/bin/env python
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:
diff changeset
2 """Wrapper for Promoter 2.0 for use in Galaxy.
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:
diff changeset
3
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:
diff changeset
4 This script takes exactly three command line arguments:
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:
diff changeset
5 * number of threads
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:
diff changeset
6 * an input DNA FASTA filename
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:
diff changeset
7 * output tabular filename.
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:
diff changeset
8
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:
diff changeset
9 It calls the Promoter 2.0 binary (e.g. .../promoter-2.0/bin/promoter_Linux,
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:
diff changeset
10 bypassing the Perl wrapper script 'promoter' which imposes a significant
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:
diff changeset
11 performace overhead for no benefit here (we don't need HTML output for
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:
diff changeset
12 example).
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:
diff changeset
13
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:
diff changeset
14 The main feature is this Python wrapper script parsers the bespoke
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:
diff changeset
15 tabular output from Promoter 2.0 and reformats it into a Galaxy friendly
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:
diff changeset
16 tab separated table.
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:
diff changeset
17
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:
diff changeset
18 Additionally, in order to take advantage of multiple cores the input FASTA
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:
diff changeset
19 file is broken into chunks and multiple copies of promoter run at once.
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:
diff changeset
20 This can be used in combination with the job-splitting available in Galaxy.
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:
diff changeset
21 Note that rewriting the FASTA input file allows us to avoid a bug in
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:
diff changeset
22 promoter 2 with long descriptions in the FASTA header line (over 200
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:
diff changeset
23 characters) which produces stray fragements of the description in the
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:
diff changeset
24 output file, making parsing non-trivial.
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:
diff changeset
25
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:
diff changeset
26 TODO - Automatically extract the sequence containing a promoter prediction?
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:
diff changeset
27 """
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
28
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
29 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
30
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:
diff changeset
31 import commands
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
32 import os
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
33 import sys
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:
diff changeset
34 import tempfile
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
35
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
36 from seq_analysis_utils import run_jobs, split_fasta, thread_count
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:
diff changeset
37
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:
diff changeset
38 FASTA_CHUNK = 500
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:
diff changeset
39
18
eb6ac44d4b8e Suite v0.2.8, record Promoter 2 verion + misc internal updates
peterjc
parents: 9
diff changeset
40 if "-v" in sys.argv or "--version" in sys.argv:
eb6ac44d4b8e Suite v0.2.8, record Promoter 2 verion + misc internal updates
peterjc
parents: 9
diff changeset
41 sys.exit(os.system("promoter -V"))
eb6ac44d4b8e Suite v0.2.8, record Promoter 2 verion + misc internal updates
peterjc
parents: 9
diff changeset
42
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:
diff changeset
43 if len(sys.argv) != 4:
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
44 sys.exit(
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
45 "Require three arguments, number of threads (int), input DNA FASTA "
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
46 "file & output tabular file. Got %i arguments." % (len(sys.argv) - 1)
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
47 )
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:
diff changeset
48
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
49 num_threads = thread_count(sys.argv[3], default=4)
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:
diff changeset
50 fasta_file = os.path.abspath(sys.argv[2])
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:
diff changeset
51 tabular_file = os.path.abspath(sys.argv[3])
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:
diff changeset
52
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:
diff changeset
53 tmp_dir = tempfile.mkdtemp()
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:
diff changeset
54
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
55
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:
diff changeset
56 def get_path_and_binary():
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 path and binary names for promoter tool."""
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
58 platform = commands.getoutput("uname") # e.g. Linux
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:
diff changeset
59 shell_script = commands.getoutput("which promoter")
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:
diff changeset
60 if not os.path.isfile(shell_script):
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
61 sys.exit("ERROR: Missing promoter executable shell script")
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:
diff changeset
62 path = None
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:
diff changeset
63 for line in open(shell_script):
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
64 if line.startswith("setenv"): # could then be tab or space!
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:
diff changeset
65 parts = line.rstrip().split(None, 2)
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:
diff changeset
66 if parts[0] == "setenv" and parts[1] == "PROM":
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:
diff changeset
67 path = parts[2]
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:
diff changeset
68 if not path:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
69 sys.exit("ERROR: Could not find promoter path (PROM) in %r" % shell_script)
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:
diff changeset
70 if not os.path.isdir(path):
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
71 sys.exit("ERROR: %r is not a directory" % path)
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:
diff changeset
72 bin = "%s/bin/promoter_%s" % (path, platform)
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:
diff changeset
73 if not os.path.isfile(bin):
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
74 sys.exit("ERROR: Missing promoter binary %r" % bin)
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:
diff changeset
75 return path, bin
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:
diff changeset
76
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
77
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:
diff changeset
78 def make_tabular(raw_handle, out_handle):
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:
diff changeset
79 """Parse text output into tabular, return query count."""
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:
diff changeset
80 identifier = None
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:
diff changeset
81 queries = 0
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:
diff changeset
82 for line in raw_handle:
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
83 # print(repr(line))
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:
diff changeset
84 if not line.strip() or line == "Promoter prediction:\n":
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:
diff changeset
85 pass
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:
diff changeset
86 elif line[0] != " ":
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
87 identifier = line.strip().replace("\t", " ").split(None, 1)[0]
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:
diff changeset
88 queries += 1
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:
diff changeset
89 elif line == " No promoter predicted\n":
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
90 # End of a record
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:
diff changeset
91 identifier = None
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:
diff changeset
92 elif line == " Position Score Likelihood\n":
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:
diff changeset
93 assert identifier
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:
diff changeset
94 else:
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:
diff changeset
95 try:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
96 position, score, likelihood = line.strip().split(None, 2)
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:
diff changeset
97 except ValueError:
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
98 print("WARNING: Problem with line: %r" % line)
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:
diff changeset
99 continue
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
100 # sys.exit("ERROR: Problem with line: %r" % line)
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
101 if likelihood not in [
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
102 "ignored",
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
103 "Marginal prediction",
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
104 "Medium likely prediction",
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
105 "Highly likely prediction",
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
106 ]:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
107 sys.exit("ERROR: Problem with line: %r" % line)
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
108 out_handle.write(
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
109 "%s\t%s\t%s\t%s\n" % (identifier, position, score, likelihood)
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
110 )
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:
diff changeset
111 return queries
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
112
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
113
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:
diff changeset
114 working_dir, bin = get_path_and_binary()
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:
diff changeset
115
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:
diff changeset
116 if not os.path.isfile(fasta_file):
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
117 sys.exit("ERROR: Missing input FASTA file %r" % fasta_file)
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:
diff changeset
118
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
119 # Note that if the input FASTA file contains no sequences,
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
120 # split_fasta returns an empty list (i.e. zero temp files).
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
121 # We deliberately omit the FASTA descriptions to avoid a
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
122 # bug in promoter2 with descriptions over 200 characters.
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
123 fasta_files = split_fasta(
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
124 fasta_file, os.path.join(tmp_dir, "promoter"), FASTA_CHUNK, keep_descr=False
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
125 )
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
126 temp_files = [f + ".out" for f in fasta_files]
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
127 jobs = [
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
128 "%s %s > %s" % (bin, fasta, temp) for fasta, temp in zip(fasta_files, temp_files)
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
129 ]
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:
diff changeset
130
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
131
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:
diff changeset
132 def clean_up(file_list):
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:
diff changeset
133 for f in file_list:
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:
diff changeset
134 if os.path.isfile(f):
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:
diff changeset
135 os.remove(f)
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:
diff changeset
136 try:
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:
diff changeset
137 os.rmdir(tmp_dir)
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
138 except Exception:
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:
diff changeset
139 pass
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:
diff changeset
140
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
141
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:
diff changeset
142 if len(jobs) > 1 and num_threads > 1:
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
143 # A small "info" message for Galaxy to show the user.
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
144 print("Using %i threads for %i tasks" % (min(num_threads, len(jobs)), len(jobs)))
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:
diff changeset
145 cur_dir = os.path.abspath(os.curdir)
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:
diff changeset
146 os.chdir(working_dir)
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:
diff changeset
147 results = run_jobs(jobs, num_threads)
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:
diff changeset
148 os.chdir(cur_dir)
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:
diff changeset
149 for fasta, temp, cmd in zip(fasta_files, temp_files, jobs):
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:
diff changeset
150 error_level = results[cmd]
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:
diff changeset
151 if error_level:
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:
diff changeset
152 try:
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:
diff changeset
153 output = open(temp).readline()
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:
diff changeset
154 except IOError:
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:
diff changeset
155 output = ""
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:
diff changeset
156 clean_up(fasta_files + temp_files)
21
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
157 sys.exit(
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
158 "One or more tasks failed, e.g. %i from %r gave:\n%s"
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
159 % (error_level, cmd, output),
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
160 error_level,
238eae32483c "Check this is up to date with all 2020 changes (black etc)"
peterjc
parents: 20
diff changeset
161 )
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:
diff changeset
162
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:
diff changeset
163 del results
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:
diff changeset
164 del jobs
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:
diff changeset
165
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:
diff changeset
166 out_handle = open(tabular_file, "w")
8
976a5f2833cd Uploaded v0.1.1 of the bundle, which fixes an error in the header of the tabular output produced for Promoter 2.0
peterjc
parents: 7
diff changeset
167 out_handle.write("#Identifier\tPosition\tScore\tLikelihood\n")
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:
diff changeset
168 queries = 0
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:
diff changeset
169 for temp in temp_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:
diff changeset
170 data_handle = open(temp)
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:
diff changeset
171 count = make_tabular(data_handle, out_handle)
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:
diff changeset
172 data_handle.close()
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:
diff changeset
173 if not count:
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:
diff changeset
174 clean_up(fasta_files + temp_files)
19
f3ecd80850e2 v0.2.9 Python style improvements
peterjc
parents: 18
diff changeset
175 sys.exit("No output from promoter2")
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:
diff changeset
176 queries += count
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:
diff changeset
177 out_handle.close()
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:
diff changeset
178
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:
diff changeset
179 clean_up(fasta_files + temp_files)
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
180 print("Results for %i queries" % queries)