annotate tools/effectiveT3/effectiveT3.py @ 5:1ea715da1879 draft

Uploaded v0.0.13, embed citation, relax test for floating point differences
author peterjc
date Tue, 25 Nov 2014 08:28:24 -0500
parents b0b927299aee
children 0f6eb4a75000
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
1 #!/usr/bin/env python
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
2 """Wrapper for EffectiveT3 v1.0.1 for use in Galaxy.
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
4 This script takes exactly five command line arguments:
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
5 * model name (e.g. TTSS_STD-1.0.1.jar)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
6 * threshold (selective or sensitive)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
7 * an input protein FASTA filename
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
8 * output tabular filename
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
9
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
10 It then calls the standalone Effective T3 v1.0.1 program (not the
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
11 webservice), and reformats the semi-colon separated output into
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
12 tab separated output for use in Galaxy.
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
13 """
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
14 import sys
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
15 import os
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
16 import subprocess
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
17
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
18 # The Galaxy auto-install via tool_dependencies.xml will set this environment variable
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
19 effectiveT3_dir = os.environ.get("EFFECTIVET3", "/opt/EffectiveT3/")
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
20 effectiveT3_jar = os.path.join(effectiveT3_dir, "TTSS_GUI-1.0.1.jar")
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
21
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
22 if "-v" in sys.argv or "--version" in sys.argv:
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
23 # TODO - Get version of the JAR file dynamically?
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
24 print("Wrapper v0.0.13, TTSS_GUI-1.0.1.jar")
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
25 sys.exit(0)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
26
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
27 def stop_err(msg, error_level=1):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
28 """Print error message to stdout and quit with given error level."""
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
29 sys.stderr.write("%s\n" % msg)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
30 sys.exit(error_level)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
31
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
32 if len(sys.argv) != 5:
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
33 stop_err("Require four arguments: model, threshold, input protein FASTA file & output tabular file")
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
34
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
35 model, threshold, fasta_file, tabular_file = sys.argv[1:]
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
36
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
37 if not os.path.isfile(fasta_file):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
38 stop_err("Input FASTA file not found: %s" % fasta_file)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
39
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
40 if threshold not in ["selective", "sensitive"] \
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
41 and not threshold.startswith("cutoff="):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
42 stop_err("Threshold should be selective, sensitive, or cutoff=..., not %r" % threshold)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
43
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
44 def clean_tabular(raw_handle, out_handle):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
45 """Clean up Effective T3 output to make it tabular."""
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
46 count = 0
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
47 positive = 0
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
48 errors = 0
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
49 for line in raw_handle:
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
50 if not line or line.startswith("#") \
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
51 or line.startswith("Id; Description; Score;"):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
52 continue
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
53 assert line.count(";") >= 3, repr(line)
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
54 # Normally there will just be three semi-colons, however the
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
55 # original FASTA file's ID or description might have had
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
56 # semi-colons in it as well, hence the following hackery:
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
57 try:
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
58 id_descr, score, effective = line.rstrip("\r\n").rsplit(";",2)
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
59 # Cope when there was no FASTA description
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
60 if "; " not in id_descr and id_descr.endswith(";"):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
61 id = id_descr[:-1]
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
62 descr = ""
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
63 else:
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
64 id, descr = id_descr.split("; ",1)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
65 except ValueError:
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
66 stop_err("Problem parsing line:\n%s\n" % line)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
67 parts = [s.strip() for s in [id, descr, score, effective]]
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
68 out_handle.write("\t".join(parts) + "\n")
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
69 count += 1
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
70 if float(score) < 0:
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
71 errors += 1
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
72 if effective.lower() == "true":
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
73 positive += 1
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
74 return count, positive, errors
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
75
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
76 def run(cmd):
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
77 # Avoid using shell=True when we call subprocess to ensure if the Python
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
78 # script is killed, so too is the child process.
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
79 try:
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
80 child = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
81 except Exception, err:
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
82 stop_err("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
83 # Use .communicate as can get deadlocks with .wait(),
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
84 stdout, stderr = child.communicate()
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
85 return_code = child.returncode
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
86 if return_code:
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
87 cmd_str= " ".join(cmd) # doesn't quote spaces etc
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
88 if stderr and stdout:
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
89 stop_err("Return code %i from command:\n%s\n\n%s\n\n%s" % (return_code, cmd_str, stdout, stderr))
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
90 else:
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
91 stop_err("Return code %i from command:\n%s\n%s" % (return_code, cmd_str, stderr))
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
92
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
93 if not os.path.isdir(effectiveT3_dir):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
94 stop_err("Effective T3 folder not found: %r" % effectiveT3_dir)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
95
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
96 if not os.path.isfile(effectiveT3_jar):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
97 stop_err("Effective T3 JAR file not found: %r" % effectiveT3_jar)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
98
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
99 if not os.path.isdir(os.path.join(effectiveT3_dir, "module")):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
100 stop_err("Effective T3 module folder not found: %r" % os.path.join(effectiveT3_dir, "module"))
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
101
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
102 effectiveT3_model = os.path.join(effectiveT3_dir, "module", model)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
103 if not os.path.isfile(effectiveT3_model):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
104 sys.stderr.write("Contents of %r is %s\n"
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
105 % (os.path.join(effectiveT3_dir, "module"),
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
106 ", ".join(repr(p) for p in os.listdir(os.path.join(effectiveT3_dir, "module")))))
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
107 sys.stderr.write("Main JAR was found: %r\n" % effectiveT3_jar)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
108 stop_err("Effective T3 model JAR file not found: %r" % effectiveT3_model)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
109
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
110 # We will have write access whereever the output should be,
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
111 temp_file = os.path.abspath(tabular_file + ".tmp")
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
112
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
113 # Use absolute paths since will change current directory...
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
114 tabular_file = os.path.abspath(tabular_file)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
115 fasta_file = os.path.abspath(fasta_file)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
116
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
117 cmd = ["java", "-jar", effectiveT3_jar,
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
118 "-f", fasta_file,
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
119 "-m", model,
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
120 "-t", threshold,
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
121 "-o", temp_file,
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
122 "-q"]
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
123
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
124 try:
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
125 # Must run from directory above the module subfolder:
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
126 os.chdir(effectiveT3_dir)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
127 except:
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
128 stop_err("Could not change to Effective T3 folder: %s" % effectiveT3_dir)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
129
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
130 run(cmd)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
131
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
132 if not os.path.isfile(temp_file):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
133 stop_err("ERROR - No output file from Effective T3")
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
134
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
135 out_handle = open(tabular_file, "w")
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
136 out_handle.write("#ID\tDescription\tScore\tEffective\n")
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
137 data_handle = open(temp_file)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
138 count, positive, errors = clean_tabular(data_handle, out_handle)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
139 data_handle.close()
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
140 out_handle.close()
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
141
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
142 os.remove(temp_file)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
143
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
144 if errors:
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
145 print("%i sequences, %i positive, %i errors"
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
146 % (count, positive, errors))
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
147 else:
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
148 print("%i/%i sequences positive" % (positive, count))
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
149
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
150 if count and count==errors:
5
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences
peterjc
parents: 3
diff changeset
151 # Galaxy will still allow them to see the output file
3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation.
peterjc
parents:
diff changeset
152 stop_err("All your sequences gave an error code")