Mercurial > repos > peterjc > effectivet3

comparison tools/effectiveT3/effectiveT3.py @ 3:b0b927299aee draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded v0.0.11 with automatic dependency installation. The Python wrapper also gives specific errors for partial installation issues.
author peterjc
date Thu, 16 May 2013 04:45:41 -0400
parents
children 1ea715da1879
comparison
equal deleted inserted replaced
2:5a8e09f115f8 3:b0b927299aee
1 #!/usr/bin/env python
2 """Wrapper for EffectiveT3 v1.0.1 for use in Galaxy.
3
4 This script takes exactly five command line arguments:
5 * model name (e.g. TTSS_STD-1.0.1.jar)
6 * threshold (selective or sensitive)
7 * an input protein FASTA filename
8 * output tabular filename
9
10 It then calls the standalone Effective T3 v1.0.1 program (not the
11 webservice), and reformats the semi-colon separated output into
12 tab separated output for use in Galaxy.
13 """
14 import sys
15 import os
16 import subprocess
17
18 #The Galaxy auto-install via tool_dependencies.xml will set this environment variable
19 effectiveT3_dir = os.environ.get("EFFECTIVET3", "/opt/EffectiveT3/")
20 effectiveT3_jar = os.path.join(effectiveT3_dir, "TTSS_GUI-1.0.1.jar")
21
22 if "-v" in sys.argv or "--version" in sys.argv:
23 #TODO - Get version of the JAR file dynamically?
24 print "Wrapper v0.0.11, TTSS_GUI-1.0.1.jar"
25 sys.exit(0)
26
27 def stop_err(msg, error_level=1):
28 """Print error message to stdout and quit with given error level."""
29 sys.stderr.write("%s\n" % msg)
30 sys.exit(error_level)
31
32 if len(sys.argv) != 5:
33 stop_err("Require four arguments: model, threshold, input protein FASTA file & output tabular file")
34
35 model, threshold, fasta_file, tabular_file = sys.argv[1:]
36
37 if not os.path.isfile(fasta_file):
38 stop_err("Input FASTA file not found: %s" % fasta_file)
39
40 if threshold not in ["selective", "sensitive"] \
41 and not threshold.startswith("cutoff="):
42 stop_err("Threshold should be selective, sensitive, or cutoff=..., not %r" % threshold)
43
44 def clean_tabular(raw_handle, out_handle):
45 """Clean up Effective T3 output to make it tabular."""
46 count = 0
47 positive = 0
48 errors = 0
49 for line in raw_handle:
50 if not line or line.startswith("#") \
51 or line.startswith("Id; Description; Score;"):
52 continue
53 assert line.count(";") >= 3, repr(line)
54 #Normally there will just be three semi-colons, however the
55 #original FASTA file's ID or description might have had
56 #semi-colons in it as well, hence the following hackery:
57 try:
58 id_descr, score, effective = line.rstrip("\r\n").rsplit(";",2)
59 #Cope when there was no FASTA description
60 if "; " not in id_descr and id_descr.endswith(";"):
61 id = id_descr[:-1]
62 descr = ""
63 else:
64 id, descr = id_descr.split("; ",1)
65 except ValueError:
66 stop_err("Problem parsing line:\n%s\n" % line)
67 parts = [s.strip() for s in [id, descr, score, effective]]
68 out_handle.write("\t".join(parts) + "\n")
69 count += 1
70 if float(score) < 0:
71 errors += 1
72 if effective.lower() == "true":
73 positive += 1
74 return count, positive, errors
75
76 def run(cmd):
77 #Avoid using shell=True when we call subprocess to ensure if the Python
78 #script is killed, so too is the child process.
79 try:
80 child = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
81 except Exception, err:
82 stop_err("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
83 #Use .communicate as can get deadlocks with .wait(),
84 stdout, stderr = child.communicate()
85 return_code = child.returncode
86 if return_code:
87 if stderr and stdout:
88 stop_err("Return code %i from command:\n%s\n\n%s\n\n%s" % (return_code, err, stdout, stderr))
89 else:
90 stop_err("Return code %i from command:\n%s\n%s" % (return_code, err, stderr))
91
92 if not os.path.isdir(effectiveT3_dir):
93 stop_err("Effective T3 folder not found: %r" % effectiveT3_dir)
94
95 if not os.path.isfile(effectiveT3_jar):
96 stop_err("Effective T3 JAR file not found: %r" % effectiveT3_jar)
97
98 if not os.path.isdir(os.path.join(effectiveT3_dir, "module")):
99 stop_err("Effective T3 module folder not found: %r" % os.path.join(effectiveT3_dir, "module"))
100
101 effectiveT3_model = os.path.join(effectiveT3_dir, "module", model)
102 if not os.path.isfile(effectiveT3_model):
103 sys.stderr.write("Contents of %r is %s\n"
104 % (os.path.join(effectiveT3_dir, "module"),
105 ", ".join(repr(p) for p in os.listdir(os.path.join(effectiveT3_dir, "module")))))
106 sys.stderr.write("Main JAR was found: %r\n" % effectiveT3_jar)
107 stop_err("Effective T3 model JAR file not found: %r" % effectiveT3_model)
108
109 #We will have write access whereever the output should be,
110 temp_file = os.path.abspath(tabular_file + ".tmp")
111
112 #Use absolute paths since will change current directory...
113 tabular_file = os.path.abspath(tabular_file)
114 fasta_file = os.path.abspath(fasta_file)
115
116 cmd = ["java", "-jar", effectiveT3_jar,
117 "-f", fasta_file,
118 "-m", model,
119 "-t", threshold,
120 "-o", temp_file,
121 "-q"]
122
123 try:
124 #Must run from directory above the module subfolder:
125 os.chdir(effectiveT3_dir)
126 except:
127 stop_err("Could not change to Effective T3 folder: %s" % effectiveT3_dir)
128
129 run(cmd)
130
131 if not os.path.isfile(temp_file):
132 stop_err("ERROR - No output file from Effective T3")
133
134 out_handle = open(tabular_file, "w")
135 out_handle.write("#ID\tDescription\tScore\tEffective\n")
136 data_handle = open(temp_file)
137 count, positive, errors = clean_tabular(data_handle, out_handle)
138 data_handle.close()
139 out_handle.close()
140
141 os.remove(temp_file)
142
143 if errors:
144 print "%i sequences, %i positive, %i errors" \
145 % (count, positive, errors)
146 else:
147 print "%i/%i sequences positive" % (positive, count)
148
149 if count and count==errors:
150 #Galaxy will still allow them to see the output file
151 stop_err("All your sequences gave an error code")