Mercurial > repos > peterjc > effectivet3

comparison tools/effectiveT3/effectiveT3.py @ 8:60a9b3f760cc draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
v0.0.17 Used cached URL, python style updates
author peterjc
date Wed, 01 Feb 2017 09:22:21 -0500
parents 5f85301d50bf
children 512530020360
comparison
equal deleted inserted replaced
7:5f85301d50bf 8:60a9b3f760cc
14 import sys 14 import sys
15 import os 15 import os
16 import subprocess 16 import subprocess
17 17
18 # The Galaxy auto-install via tool_dependencies.xml will set this environment variable 18 # The Galaxy auto-install via tool_dependencies.xml will set this environment variable
19 effectiveT3_dir = os.environ.get("EFFECTIVET3", "/opt/EffectiveT3/") 19 effective_t3_dir = os.environ.get("EFFECTIVET3", "/opt/EffectiveT3/")
20 effectiveT3_jar = os.path.join(effectiveT3_dir, "TTSS_GUI-1.0.1.jar") 20 effective_t3_jar = os.path.join(effective_t3_dir, "TTSS_GUI-1.0.1.jar")
21 21
22 if "-v" in sys.argv or "--version" in sys.argv: 22 if "-v" in sys.argv or "--version" in sys.argv:
23 # TODO - Get version of the JAR file dynamically? 23 # TODO - Get version of the JAR file dynamically?
24 print("Wrapper v0.0.16, TTSS_GUI-1.0.1.jar") 24 print("Wrapper v0.0.17, TTSS_GUI-1.0.1.jar")
25 sys.exit(0) 25 sys.exit(0)
26 26
27 def sys_exit(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: 27 if len(sys.argv) != 5:
33 sys_exit("Require four arguments: model, threshold, input protein FASTA file & output tabular file") 28 sys.exit("Require four arguments: model, threshold, input protein FASTA file & output tabular file")
34 29
35 model, threshold, fasta_file, tabular_file = sys.argv[1:] 30 model, threshold, fasta_file, tabular_file = sys.argv[1:]
36 31
37 if not os.path.isfile(fasta_file): 32 if not os.path.isfile(fasta_file):
38 sys_exit("Input FASTA file not found: %s" % fasta_file) 33 sys.exit("Input FASTA file not found: %s" % fasta_file)
39 34
40 if threshold not in ["selective", "sensitive"] \ 35 if threshold not in ["selective", "sensitive"] \
41 and not threshold.startswith("cutoff="): 36 and not threshold.startswith("cutoff="):
42 sys_exit("Threshold should be selective, sensitive, or cutoff=..., not %r" % threshold) 37 sys.exit("Threshold should be selective, sensitive, or cutoff=..., not %r" % threshold)
38
43 39
44 def clean_tabular(raw_handle, out_handle): 40 def clean_tabular(raw_handle, out_handle):
45 """Clean up Effective T3 output to make it tabular.""" 41 """Clean up Effective T3 output to make it tabular."""
46 count = 0 42 count = 0
47 positive = 0 43 positive = 0
48 errors = 0 44 errors = 0
49 for line in raw_handle: 45 for line in raw_handle:
50 if not line or line.startswith("#") \ 46 if not line or line.startswith("#") \
51 or line.startswith("Id; Description; Score;"): 47 or line.startswith("Id; Description; Score;"):
52 continue 48 continue
53 assert line.count(";") >= 3, repr(line) 49 assert line.count(";") >= 3, repr(line)
54 # Normally there will just be three semi-colons, however the 50 # Normally there will just be three semi-colons, however the
55 # original FASTA file's ID or description might have had 51 # original FASTA file's ID or description might have had
56 # semi-colons in it as well, hence the following hackery: 52 # semi-colons in it as well, hence the following hackery:
57 try: 53 try:
58 id_descr, score, effective = line.rstrip("\r\n").rsplit(";",2) 54 id_descr, score, effective = line.rstrip("\r\n").rsplit(";", 2)
59 # Cope when there was no FASTA description 55 # Cope when there was no FASTA description
60 if "; " not in id_descr and id_descr.endswith(";"): 56 if "; " not in id_descr and id_descr.endswith(";"):
61 id = id_descr[:-1] 57 id = id_descr[:-1]
62 descr = "" 58 descr = ""
63 else: 59 else:
64 id, descr = id_descr.split("; ",1) 60 id, descr = id_descr.split("; ", 1)
65 except ValueError: 61 except ValueError:
66 sys_exit("Problem parsing line:\n%s\n" % line) 62 sys.exit("Problem parsing line:\n%s\n" % line)
67 parts = [s.strip() for s in [id, descr, score, effective]] 63 parts = [s.strip() for s in [id, descr, score, effective]]
68 out_handle.write("\t".join(parts) + "\n") 64 out_handle.write("\t".join(parts) + "\n")
69 count += 1 65 count += 1
70 if float(score) < 0: 66 if float(score) < 0:
71 errors += 1 67 errors += 1
72 if effective.lower() == "true": 68 if effective.lower() == "true":
73 positive += 1 69 positive += 1
74 return count, positive, errors 70 return count, positive, errors
71
75 72
76 def run(cmd): 73 def run(cmd):
77 # Avoid using shell=True when we call subprocess to ensure if the Python 74 # Avoid using shell=True when we call subprocess to ensure if the Python
78 # script is killed, so too is the child process. 75 # script is killed, so too is the child process.
79 try: 76 try:
80 child = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE) 77 child = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
81 except Exception, err: 78 except Exception, err:
82 sys_exit("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err)) 79 sys.exit("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
83 # Use .communicate as can get deadlocks with .wait(), 80 # Use .communicate as can get deadlocks with .wait(),
84 stdout, stderr = child.communicate() 81 stdout, stderr = child.communicate()
85 return_code = child.returncode 82 return_code = child.returncode
86 if return_code or stderr.startswith("Exception in thread"): 83 if return_code or stderr.startswith("Exception in thread"):
87 cmd_str= " ".join(cmd) # doesn't quote spaces etc 84 cmd_str = " ".join(cmd) # doesn't quote spaces etc
88 if stderr and stdout: 85 if stderr and stdout:
89 sys_exit("Return code %i from command:\n%s\n\n%s\n\n%s" % (return_code, cmd_str, stdout, stderr)) 86 sys.exit("Return code %i from command:\n%s\n\n%s\n\n%s" % (return_code, cmd_str, stdout, stderr))
90 else: 87 else:
91 sys_exit("Return code %i from command:\n%s\n%s" % (return_code, cmd_str, stderr)) 88 sys.exit("Return code %i from command:\n%s\n%s" % (return_code, cmd_str, stderr))
92 89
93 90
94 if not os.path.isdir(effectiveT3_dir): 91 if not os.path.isdir(effective_t3_dir):
95 sys_exit("Effective T3 folder not found: %r" % effectiveT3_dir) 92 sys.exit("Effective T3 folder not found: %r" % effective_t3_dir)
96 93
97 if not os.path.isfile(effectiveT3_jar): 94 if not os.path.isfile(effective_t3_jar):
98 sys_exit("Effective T3 JAR file not found: %r" % effectiveT3_jar) 95 sys.exit("Effective T3 JAR file not found: %r" % effective_t3_jar)
99 96
100 if not os.path.isdir(os.path.join(effectiveT3_dir, "module")): 97 if not os.path.isdir(os.path.join(effective_t3_dir, "module")):
101 sys_exit("Effective T3 module folder not found: %r" % os.path.join(effectiveT3_dir, "module")) 98 sys.exit("Effective T3 module folder not found: %r" % os.path.join(effective_t3_dir, "module"))
102 99
103 effectiveT3_model = os.path.join(effectiveT3_dir, "module", model) 100 effective_t3_model = os.path.join(effective_t3_dir, "module", model)
104 if not os.path.isfile(effectiveT3_model): 101 if not os.path.isfile(effective_t3_model):
105 sys.stderr.write("Contents of %r is %s\n" 102 sys.stderr.write("Contents of %r is %s\n"
106 % (os.path.join(effectiveT3_dir, "module"), 103 % (os.path.join(effective_t3_dir, "module"),
107 ", ".join(repr(p) for p in os.listdir(os.path.join(effectiveT3_dir, "module"))))) 104 ", ".join(repr(p) for p in os.listdir(os.path.join(effective_t3_dir, "module")))))
108 sys.stderr.write("Main JAR was found: %r\n" % effectiveT3_jar) 105 sys.stderr.write("Main JAR was found: %r\n" % effective_t3_jar)
109 sys_exit("Effective T3 model JAR file not found: %r" % effectiveT3_model) 106 sys.exit("Effective T3 model JAR file not found: %r" % effective_t3_model)
110 107
111 # We will have write access whereever the output should be, 108 # We will have write access whereever the output should be,
112 temp_file = os.path.abspath(tabular_file + ".tmp") 109 temp_file = os.path.abspath(tabular_file + ".tmp")
113 110
114 # Use absolute paths since will change current directory... 111 # Use absolute paths since will change current directory...
115 tabular_file = os.path.abspath(tabular_file) 112 tabular_file = os.path.abspath(tabular_file)
116 fasta_file = os.path.abspath(fasta_file) 113 fasta_file = os.path.abspath(fasta_file)
117 114
118 cmd = ["java", "-jar", effectiveT3_jar, 115 cmd = ["java", "-jar", effective_t3_jar,
119 "-f", fasta_file, 116 "-f", fasta_file,
120 "-m", model, 117 "-m", model,
121 "-t", threshold, 118 "-t", threshold,
122 "-o", temp_file, 119 "-o", temp_file,
123 "-q"] 120 "-q"]
124 121
125 try: 122 try:
126 # Must run from directory above the module subfolder: 123 # Must run from directory above the module subfolder:
127 os.chdir(effectiveT3_dir) 124 os.chdir(effective_t3_dir)
128 except: 125 except Exception:
129 sys_exit("Could not change to Effective T3 folder: %s" % effectiveT3_dir) 126 sys.exit("Could not change to Effective T3 folder: %s" % effective_t3_dir)
130 127
131 run(cmd) 128 run(cmd)
132 129
133 if not os.path.isfile(temp_file): 130 if not os.path.isfile(temp_file):
134 sys_exit("ERROR - No output file from Effective T3") 131 sys.exit("ERROR - No output file from Effective T3")
135 132
136 out_handle = open(tabular_file, "w") 133 out_handle = open(tabular_file, "w")
137 out_handle.write("#ID\tDescription\tScore\tEffective\n") 134 out_handle.write("#ID\tDescription\tScore\tEffective\n")
138 data_handle = open(temp_file) 135 data_handle = open(temp_file)
139 count, positive, errors = clean_tabular(data_handle, out_handle) 136 count, positive, errors = clean_tabular(data_handle, out_handle)
141 out_handle.close() 138 out_handle.close()
142 139
143 os.remove(temp_file) 140 os.remove(temp_file)
144 141
145 if errors: 142 if errors:
146 print("%i sequences, %i positive, %i errors" 143 print("%i sequences, %i positive, %i errors"
147 % (count, positive, errors)) 144 % (count, positive, errors))
148 else: 145 else:
149 print("%i/%i sequences positive" % (positive, count)) 146 print("%i/%i sequences positive" % (positive, count))
150 147
151 if count and count==errors: 148 if count and count == errors:
152 # Galaxy will still allow them to see the output file 149 # Galaxy will still allow them to see the output file
153 sys_exit("All your sequences gave an error code") 150 sys.exit("All your sequences gave an error code")