Mercurial > repos > peterjc > effectivet3

annotate tools/protein_analysis/effectiveT3.py @ 0:096088373590

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
author peterjc
date Tue, 07 Jun 2011 16:32:23 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
1 #!/usr/bin/env python
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
2 """Wrapper for EffectiveT3 v1.0.1 for use in Galaxy.
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
3
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
4 This script takes exactly five command line arguments:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
5 * model name (e.g. TTSS_STD-1.0.1.jar)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
6 * threshold (selective or sensitive)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
7 * an input protein FASTA filename
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
8 * output tabular filename
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
9
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
10 It then calls the standalone Effective T3 v1.0.1 program (not the
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
11 webservice), and reformats the semi-colon separated output into
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
12 tab separated output for use in Galaxy.
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
13 """
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
14 import sys
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
15 import os
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
16 import subprocess
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
17
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
18 #You may need to edit this to match your local setup,
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
19 effectiveT3_jar = "/opt/EffectiveT3/TTSS_GUI-1.0.1.jar"
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
20
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
21
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
22 def stop_err(msg, error_level=1):
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
23 """Print error message to stdout and quit with given error level."""
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
24 sys.stderr.write("%s\n" % msg)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
25 sys.exit(error_level)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
26
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
27 if len(sys.argv) != 5:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
28 stop_err("Require four arguments: model, threshold, input protein FASTA file & output tabular file")
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
29
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
30 model, threshold, fasta_file, tabular_file = sys.argv[1:]
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
31
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
32 if not os.path.isfile(fasta_file):
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
33 stop_err("Input FASTA file not found: %s" % fasta_file)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
34
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
35 if threshold not in ["selective", "sensitive"] \
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
36 and not threshold.startswith("cutoff="):
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
37 stop_err("Threshold should be selective, sensitive, or cutoff=..., not %r" % threshold)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
38
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
39 def clean_tabular(raw_handle, out_handle):
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
40 """Clean up Effective T3 output to make it tabular."""
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
41 count = 0
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
42 positive = 0
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
43 errors = 0
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
44 for line in raw_handle:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
45 if not line or line.startswith("#") \
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
46 or line.startswith("Id; Description; Score;"):
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
47 continue
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
48 assert line.count(";") >= 3, repr(line)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
49 #Normally there will just be three semi-colons, however the
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
50 #original FASTA file's ID or description might have had
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
51 #semi-colons in it as well, hence the following hackery:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
52 try:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
53 id_descr, score, effective = line.rstrip("\r\n").rsplit(";",2)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
54 #Cope when there was no FASTA description
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
55 if "; " not in id_descr and id_descr.endswith(";"):
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
56 id = id_descr[:-1]
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
57 descr = ""
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
58 else:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
59 id, descr = id_descr.split("; ",1)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
60 except ValueError:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
61 stop_err("Problem parsing line:\n%s\n" % line)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
62 parts = [s.strip() for s in [id, descr, score, effective]]
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
63 out_handle.write("\t".join(parts) + "\n")
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
64 count += 1
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
65 if float(score) < 0:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
66 errors += 1
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
67 if effective.lower() == "true":
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
68 positive += 1
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
69 return count, positive, errors
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
70
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
71 def run(cmd):
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
72 #Avoid using shell=True when we call subprocess to ensure if the Python
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
73 #script is killed, so too is the child process.
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
74 try:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
75 child = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
76 except Exception, err:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
77 stop_err("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
78 #Use .communicate as can get deadlocks with .wait(),
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
79 stdout, stderr = child.communicate()
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
80 return_code = child.returncode
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
81 if return_code:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
82 if stderr and stdout:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
83 stop_err("Return code %i from command:\n%s\n\n%s\n\n%s" % (return_code, err, stdout, stderr))
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
84 else:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
85 stop_err("Return code %i from command:\n%s\n%s" % (return_code, err, stderr))
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
86
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
87 if not os.path.isfile(effectiveT3_jar):
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
88 stop_err("Effective T3 JAR file not found: %s" % effectiveT3_jar)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
89
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
90 effectiveT3_dir = os.path.dirname(effectiveT3_jar)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
91 if not os.path.isdir(effectiveT3_dir):
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
92 stop_err("Effective T3 folder not found: %s" % effectiveT3_dir)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
93
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
94 effectiveT3_model = os.path.join(effectiveT3_dir, "module", model)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
95 if not os.path.isfile(effectiveT3_model):
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
96 stop_err("Effective T3 model JAR file not found: %s" % effectiveT3_model)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
97
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
98 #We will have write access whereever the output should be,
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
99 temp_file = os.path.abspath(tabular_file + ".tmp")
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
100
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
101 #Use absolute paths since will change current directory...
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
102 tabular_file = os.path.abspath(tabular_file)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
103 fasta_file = os.path.abspath(fasta_file)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
104
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
105 cmd = ["java", "-jar", effectiveT3_jar,
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
106 "-f", fasta_file,
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
107 "-m", model,
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
108 "-t", threshold,
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
109 "-o", temp_file,
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
110 "-q"]
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
111
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
112 try:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
113 #Must run from directory above the module subfolder:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
114 os.chdir(effectiveT3_dir)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
115 except:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
116 stop_err("Could not change to Effective T3 folder: %s" % effectiveT3_dir)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
117
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
118 run(cmd)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
119
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
120 if not os.path.isfile(temp_file):
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
121 stop_err("ERROR - No output file from Effective T3")
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
122
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
123 out_handle = open(tabular_file, "w")
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
124 out_handle.write("#ID\tDescription\tScore\tEffective\n")
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
125 data_handle = open(temp_file)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
126 count, positive, errors = clean_tabular(data_handle, out_handle)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
127 data_handle.close()
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
128 out_handle.close()
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
129
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
130 os.remove(temp_file)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
131
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
132 if errors:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
133 print "%i sequences, %i positive, %i errors" \
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
134 % (count, positive, errors)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
135 else:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
136 print "%i/%i sequences positive" % (positive, count)
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
137
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
138 if count and count==errors:
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
139 #Galaxy will still allow them to see the output file
096088373590 Migrated tool version 0.0.7 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
140 stop_err("All your sequences gave an error code")