Mercurial > repos > peterjc > tmhmm_and_signalp

annotate tools/protein_analysis/psortb.py @ 13:dc958c2a963a draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded v0.2.2, adds missing test files
author peterjc
date Tue, 23 Apr 2013 12:47:49 -0400
parents 99b82a2b1272
children eb6ac44d4b8e
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
11
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
1 #!/usr/bin/env python
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
2 """Wrapper for psortb for use in Galaxy.
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
3
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
4 This script takes exactly six command line arguments - which includes the
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
5 number of threads, and the input protein FASTA filename and output
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
6 tabular filename. It then splits up the FASTA input and calls multiple
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
7 copies of the standalone psortb v3 program, then collates the output.
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
8 e.g. Rather than this,
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
9
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
10 psort $type -c $cutoff -d $divergent -o long $sequence > $outfile
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
11
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
12 Call this:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
13
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
14 psort $threads $type $cutoff $divergent $sequence $outfile
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
15
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
16 If ommitting -c or -d options, set $cutoff and $divergent to zero or blank.
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
17
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
18 Note that this is somewhat redundant with job-splitting available in Galaxy
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
19 itself (see the SignalP XML file for settings), but both can be applied.
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
20
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
21 Additionally it ensures the header line (with the column names) starts
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
22 with a # character as used elsewhere in Galaxy.
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
23 """
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
24 import sys
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
25 import os
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
26 import tempfile
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
27 from seq_analysis_utils import stop_err, split_fasta, run_jobs, thread_count
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
28
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
29 FASTA_CHUNK = 500
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
30
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
31 if "-v" in sys.argv or "--version" in sys.argv:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
32 """Return underlying PSORTb's version"""
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
33 sys.exit(os.system("psort --version"))
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
34
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
35 if len(sys.argv) != 8:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
36 stop_err("Require 7 arguments, number of threads (int), type (e.g. archaea), "
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
37 "output (e.g. terse/normal/long), cutoff, divergent, input protein "
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
38 "FASTA file & output tabular file")
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
39
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
40 num_threads = thread_count(sys.argv[1], default=4)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
41 org_type = sys.argv[2]
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
42 out_type = sys.argv[3]
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
43 cutoff = sys.argv[4]
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
44 if cutoff.strip() and float(cutoff.strip()) != 0.0:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
45 cutoff = "-c %s" % cutoff
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
46 else:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
47 cutoff = ""
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
48 divergent = sys.argv[5]
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
49 if divergent.strip() and float(divergent.strip()) != 0.0:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
50 divergent = "-d %s" % divergent
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
51 else:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
52 divergent = ""
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
53 fasta_file = sys.argv[6]
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
54 tabular_file = sys.argv[7]
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
55
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
56 if out_type == "terse":
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
57 header = ['SeqID', 'Localization', 'Score']
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
58 elif out_type == "normal":
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
59 stop_err("Normal output not implemented yet, sorry.")
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
60 elif out_type == "long":
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
61 if org_type == "-n":
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
62 #Gram negative bacteria
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
63 header = ['SeqID', 'CMSVM-_Localization', 'CMSVM-_Details', 'CytoSVM-_Localization', 'CytoSVM-_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
64 'ECSVM-_Localization', 'ECSVM-_Details', 'ModHMM-_Localization', 'ModHMM-_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
65 'Motif-_Localization', 'Motif-_Details', 'OMPMotif-_Localization', 'OMPMotif-_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
66 'OMSVM-_Localization', 'OMSVM-_Details', 'PPSVM-_Localization', 'PPSVM-_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
67 'Profile-_Localization', 'Profile-_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
68 'SCL-BLAST-_Localization', 'SCL-BLAST-_Details', 'SCL-BLASTe-_Localization', 'SCL-BLASTe-_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
69 'Signal-_Localization', 'Signal-_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
70 'Cytoplasmic_Score', 'CytoplasmicMembrane_Score', 'Periplasmic_Score', 'OuterMembrane_Score',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
71 'Extracellular_Score', 'Final_Localization', 'Final_Localization_Details', 'Final_Score',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
72 'Secondary_Localization', 'PSortb_Version']
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
73 elif org_type == "-p":
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
74 #Gram positive bacteria
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
75 header = ['SeqID', 'CMSVM+_Localization', 'CMSVM+_Details', 'CWSVM+_Localization', 'CWSVM+_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
76 'CytoSVM+_Localization', 'CytoSVM+_Details', 'ECSVM+_Localization', 'ECSVM+_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
77 'ModHMM+_Localization', 'ModHMM+_Details', 'Motif+_Localization', 'Motif+_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
78 'Profile+_Localization', 'Profile+_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
79 'SCL-BLAST+_Localization', 'SCL-BLAST+_Details', 'SCL-BLASTe+_Localization', 'SCL-BLASTe+_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
80 'Signal+_Localization', 'Signal+_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
81 'Cytoplasmic_Score', 'CytoplasmicMembrane_Score', 'Cellwall_Score',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
82 'Extracellular_Score', 'Final_Localization', 'Final_Localization_Details', 'Final_Score',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
83 'Secondary_Localization', 'PSortb_Version']
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
84 elif org_type == "-a":
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
85 #Archaea
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
86 header = ['SeqID', 'CMSVM_a_Localization', 'CMSVM_a_Details', 'CWSVM_a_Localization', 'CWSVM_a_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
87 'CytoSVM_a_Localization', 'CytoSVM_a_Details', 'ECSVM_a_Localization', 'ECSVM_a_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
88 'ModHMM_a_Localization', 'ModHMM_a_Details', 'Motif_a_Localization', 'Motif_a_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
89 'Profile_a_Localization', 'Profile_a_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
90 'SCL-BLAST_a_Localization', 'SCL-BLAST_a_Details', 'SCL-BLASTe_a_Localization', 'SCL-BLASTe_a_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
91 'Signal_a_Localization', 'Signal_a_Details',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
92 'Cytoplasmic_Score', 'CytoplasmicMembrane_Score', 'Cellwall_Score',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
93 'Extracellular_Score', 'Final_Localization', 'Final_Localization_Details', 'Final_Score',
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
94 'Secondary_Localization', 'PSortb_Version']
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
95 else:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
96 stop_err("Expected -n, -p or -a for the organism type, not %r" % org_type)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
97 else:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
98 stop_err("Expected terse, normal or long for the output type, not %r" % out_type)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
99
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
100 tmp_dir = tempfile.mkdtemp()
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
101
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
102 def clean_tabular(raw_handle, out_handle):
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
103 """Clean up tabular TMHMM output, returns output line count."""
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
104 global header
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
105 count = 0
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
106 for line in raw_handle:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
107 if not line.strip() or line.startswith("#"):
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
108 #Ignore any blank lines or comment lines
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
109 continue
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
110 parts = [x.strip() for x in line.rstrip("\r\n").split("\t")]
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
111 if parts == header:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
112 #Ignore the header line
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
113 continue
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
114 if not parts[-1] and len(parts) == len(header) + 1:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
115 #Ignore dummy blank extra column, e.g.
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
116 #"...2.0\t\tPSORTb version 3.0\t\n"
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
117 parts = parts[:-1]
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
118 assert len(parts) == len(header), \
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
119 "%i fields, not %i, in line:\n%r" % (len(line), len(header), line)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
120 out_handle.write(line)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
121 count += 1
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
122 return count
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
123
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
124 #Note that if the input FASTA file contains no sequences,
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
125 #split_fasta returns an empty list (i.e. zero temp files).
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
126 fasta_files = split_fasta(fasta_file, os.path.join(tmp_dir, "tmhmm"), FASTA_CHUNK)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
127 temp_files = [f+".out" for f in fasta_files]
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
128 jobs = ["psort %s %s %s -o %s %s > %s" % (org_type, cutoff, divergent, out_type, fasta, temp)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
129 for fasta, temp in zip(fasta_files, temp_files)]
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
130
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
131 def clean_up(file_list):
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
132 for f in file_list:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
133 if os.path.isfile(f):
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
134 os.remove(f)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
135 try:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
136 os.rmdir(tmp_dir)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
137 except:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
138 pass
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
139
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
140 if len(jobs) > 1 and num_threads > 1:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
141 #A small "info" message for Galaxy to show the user.
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
142 print "Using %i threads for %i tasks" % (min(num_threads, len(jobs)), len(jobs))
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
143 results = run_jobs(jobs, num_threads)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
144 for fasta, temp, cmd in zip(fasta_files, temp_files, jobs):
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
145 error_level = results[cmd]
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
146 if error_level:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
147 try:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
148 output = open(temp).readline()
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
149 except IOError:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
150 output = ""
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
151 clean_up(fasta_files + temp_files)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
152 stop_err("One or more tasks failed, e.g. %i from %r gave:\n%s" % (error_level, cmd, output),
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
153 error_level)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
154 del results
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
155 del jobs
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
156
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
157 out_handle = open(tabular_file, "w")
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
158 out_handle.write("#%s\n" % "\t".join(header))
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
159 count = 0
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
160 for temp in temp_files:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
161 data_handle = open(temp)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
162 count += clean_tabular(data_handle, out_handle)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
163 data_handle.close()
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
164 if not count:
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
165 clean_up(fasta_files + temp_files)
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
166 stop_err("No output from psortb")
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
167 out_handle.close()
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
168 print "%i records" % count
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
169
99b82a2b1272 Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
peterjc
parents:
diff changeset
170 clean_up(fasta_files + temp_files)