annotate tools/effectiveT3/effectiveT3.py @ 9:b863c4bec840 draft

Uploaded v0.0.11, take 10, debug output if module folder missing
author peterjc
date Mon, 13 May 2013 05:14:33 -0400
parents 4a0aa59062f7
children 17fbf8b8a554
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
1 #!/usr/bin/env python
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
2 """Wrapper for EffectiveT3 v1.0.1 for use in Galaxy.
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
3
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
4 This script takes exactly five command line arguments:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
5 * model name (e.g. TTSS_STD-1.0.1.jar)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
6 * threshold (selective or sensitive)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
7 * an input protein FASTA filename
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
8 * output tabular filename
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
9
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
10 It then calls the standalone Effective T3 v1.0.1 program (not the
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
11 webservice), and reformats the semi-colon separated output into
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
12 tab separated output for use in Galaxy.
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
13 """
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
14 import sys
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
15 import os
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
16 import subprocess
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
17
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
18 #The Galaxy auto-install via tool_dependencies.xml will set this environment variable
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
19 effectiveT3_dir = os.environ.get("EFFECTIVET3", "/opt/EffectiveT3/")
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
20 effectiveT3_jar = os.path.join(effectiveT3_dir, "TTSS_GUI-1.0.1.jar")
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
21
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
22 if "-v" in sys.argv or "--version" in sys.argv:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
23 #TODO - Get version of the JAR file dynamically?
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
24 print "Wrapper v0.0.11, TTSS_GUI-1.0.1.jar"
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
25 sys.exit(0)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
26
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
27 def stop_err(msg, error_level=1):
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
28 """Print error message to stdout and quit with given error level."""
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
29 sys.stderr.write("%s\n" % msg)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
30 sys.exit(error_level)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
31
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
32 if len(sys.argv) != 5:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
33 stop_err("Require four arguments: model, threshold, input protein FASTA file & output tabular file")
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
34
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
35 model, threshold, fasta_file, tabular_file = sys.argv[1:]
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
36
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
37 if not os.path.isfile(fasta_file):
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
38 stop_err("Input FASTA file not found: %s" % fasta_file)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
39
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
40 if threshold not in ["selective", "sensitive"] \
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
41 and not threshold.startswith("cutoff="):
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
42 stop_err("Threshold should be selective, sensitive, or cutoff=..., not %r" % threshold)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
43
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
44 def clean_tabular(raw_handle, out_handle):
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
45 """Clean up Effective T3 output to make it tabular."""
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
46 count = 0
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
47 positive = 0
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
48 errors = 0
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
49 for line in raw_handle:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
50 if not line or line.startswith("#") \
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
51 or line.startswith("Id; Description; Score;"):
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
52 continue
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
53 assert line.count(";") >= 3, repr(line)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
54 #Normally there will just be three semi-colons, however the
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
55 #original FASTA file's ID or description might have had
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
56 #semi-colons in it as well, hence the following hackery:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
57 try:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
58 id_descr, score, effective = line.rstrip("\r\n").rsplit(";",2)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
59 #Cope when there was no FASTA description
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
60 if "; " not in id_descr and id_descr.endswith(";"):
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
61 id = id_descr[:-1]
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
62 descr = ""
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
63 else:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
64 id, descr = id_descr.split("; ",1)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
65 except ValueError:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
66 stop_err("Problem parsing line:\n%s\n" % line)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
67 parts = [s.strip() for s in [id, descr, score, effective]]
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
68 out_handle.write("\t".join(parts) + "\n")
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
69 count += 1
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
70 if float(score) < 0:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
71 errors += 1
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
72 if effective.lower() == "true":
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
73 positive += 1
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
74 return count, positive, errors
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
75
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
76 def run(cmd):
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
77 #Avoid using shell=True when we call subprocess to ensure if the Python
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
78 #script is killed, so too is the child process.
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
79 try:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
80 child = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
81 except Exception, err:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
82 stop_err("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
83 #Use .communicate as can get deadlocks with .wait(),
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
84 stdout, stderr = child.communicate()
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
85 return_code = child.returncode
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
86 if return_code:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
87 if stderr and stdout:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
88 stop_err("Return code %i from command:\n%s\n\n%s\n\n%s" % (return_code, err, stdout, stderr))
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
89 else:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
90 stop_err("Return code %i from command:\n%s\n%s" % (return_code, err, stderr))
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
91
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
92 if not os.path.isdir(effectiveT3_dir):
8
4a0aa59062f7 Uploaded v0.0.11, take 9, debug output if can't find model
peterjc
parents: 2
diff changeset
93 stop_err("Effective T3 folder not found: %r" % effectiveT3_dir)
2
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
94
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
95 if not os.path.isfile(effectiveT3_jar):
8
4a0aa59062f7 Uploaded v0.0.11, take 9, debug output if can't find model
peterjc
parents: 2
diff changeset
96 stop_err("Effective T3 JAR file not found: %r" % effectiveT3_jar)
2
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
97
9
b863c4bec840 Uploaded v0.0.11, take 10, debug output if module folder missing
peterjc
parents: 8
diff changeset
98 if not os.path.isdir(os.path.join(effectiveT3_dir, "module")):
b863c4bec840 Uploaded v0.0.11, take 10, debug output if module folder missing
peterjc
parents: 8
diff changeset
99 stop_err("Effective T3 module folder not found: %r" % os.path.join(effectiveT3_dir, "module"))
b863c4bec840 Uploaded v0.0.11, take 10, debug output if module folder missing
peterjc
parents: 8
diff changeset
100
2
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
101 effectiveT3_model = os.path.join(effectiveT3_dir, "module", model)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
102 if not os.path.isfile(effectiveT3_model):
8
4a0aa59062f7 Uploaded v0.0.11, take 9, debug output if can't find model
peterjc
parents: 2
diff changeset
103 sys.stderr.write("Contents of %r is %s\n"
4a0aa59062f7 Uploaded v0.0.11, take 9, debug output if can't find model
peterjc
parents: 2
diff changeset
104 % (os.path.join(effectiveT3_dir, "module"),
4a0aa59062f7 Uploaded v0.0.11, take 9, debug output if can't find model
peterjc
parents: 2
diff changeset
105 ", ".join(repr(p) for p in os.listdir(os.path.join(effectiveT3_dir, "module")))))
4a0aa59062f7 Uploaded v0.0.11, take 9, debug output if can't find model
peterjc
parents: 2
diff changeset
106 sys.stderr.write("Main JAR was found: %r\n" % effectiveT3_jar)
4a0aa59062f7 Uploaded v0.0.11, take 9, debug output if can't find model
peterjc
parents: 2
diff changeset
107 stop_err("Effective T3 model JAR file not found: %r" % effectiveT3_model)
2
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
108
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
109 #We will have write access whereever the output should be,
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
110 temp_file = os.path.abspath(tabular_file + ".tmp")
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
111
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
112 #Use absolute paths since will change current directory...
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
113 tabular_file = os.path.abspath(tabular_file)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
114 fasta_file = os.path.abspath(fasta_file)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
115
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
116 cmd = ["java", "-jar", effectiveT3_jar,
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
117 "-f", fasta_file,
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
118 "-m", model,
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
119 "-t", threshold,
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
120 "-o", temp_file,
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
121 "-q"]
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
122
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
123 try:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
124 #Must run from directory above the module subfolder:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
125 os.chdir(effectiveT3_dir)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
126 except:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
127 stop_err("Could not change to Effective T3 folder: %s" % effectiveT3_dir)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
128
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
129 run(cmd)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
130
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
131 if not os.path.isfile(temp_file):
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
132 stop_err("ERROR - No output file from Effective T3")
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
133
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
134 out_handle = open(tabular_file, "w")
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
135 out_handle.write("#ID\tDescription\tScore\tEffective\n")
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
136 data_handle = open(temp_file)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
137 count, positive, errors = clean_tabular(data_handle, out_handle)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
138 data_handle.close()
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
139 out_handle.close()
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
140
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
141 os.remove(temp_file)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
142
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
143 if errors:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
144 print "%i sequences, %i positive, %i errors" \
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
145 % (count, positive, errors)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
146 else:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
147 print "%i/%i sequences positive" % (positive, count)
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
148
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
149 if count and count==errors:
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
150 #Galaxy will still allow them to see the output file
66e9d4c44ca2 Uploaded v0.0.11, auto installation
peterjc
parents:
diff changeset
151 stop_err("All your sequences gave an error code")