|
18
|
1 #!/usr/bin/env python
|
|
|
2
|
|
|
3 import user, os, sys, getopt, exceptions, logging, ConfigParser
|
|
|
4
|
|
|
5 if not os.environ.has_key( "REPET_PATH" ):
|
|
|
6 print "*** Error: no environment variable REPET_PATH"
|
|
|
7 sys.exit(1)
|
|
|
8 sys.path.append( os.environ["REPET_PATH"] )
|
|
|
9
|
|
|
10 import pyRepet.sql.RepetJobMySQL
|
|
|
11 import pyRepet.launcher.Launcher
|
|
|
12
|
|
|
13 #-----------------------------------------------------------------------------
|
|
|
14
|
|
|
15 def help():
|
|
|
16
|
|
|
17 """
|
|
|
18 Give the list of the command-line options.
|
|
|
19 """
|
|
|
20
|
|
|
21 print
|
|
|
22 print "usage:",sys.argv[0]," [ options ]"
|
|
|
23 print "options:"
|
|
|
24 print " -h: this help"
|
|
|
25 print " -g: name of the group identifier (same for all the jobs)"
|
|
|
26 print " -q: name of the query directory"
|
|
|
27 print " -S: suffix in the query directory (default='*.fa')"
|
|
|
28 print " -Q: name of the queue (on the cluster)"
|
|
|
29 print " -d: absolute path to the temporary directory"
|
|
|
30 print " -C: configuration file from TEdenovo or TEannot pipeline"
|
|
|
31 print " -t: job table name (default=jobs)"
|
|
|
32 print " -p: absolute path to project directory (if jobs management via files)"
|
|
|
33 print " -c: clean (remove job launch files and job stdout)"
|
|
|
34 print " -v: verbose (default=0/1/2)"
|
|
|
35 print
|
|
|
36
|
|
|
37 #-----------------------------------------------------------------------------
|
|
|
38
|
|
|
39 def main():
|
|
|
40
|
|
|
41 """
|
|
|
42 This program takes a directory as input and launches MAP on each file in it.
|
|
|
43 """
|
|
|
44
|
|
|
45 groupid = ""
|
|
|
46 queryDir = ""
|
|
|
47 patternSuffix = "*.fa"
|
|
|
48 queue = ""
|
|
|
49 tmpDir = ""
|
|
|
50 configFileName = ""
|
|
|
51 jobTable = "jobs"
|
|
|
52 projectDir = ""
|
|
|
53 clean = False
|
|
|
54 verbose = 0
|
|
|
55
|
|
|
56 try:
|
|
|
57 opts, args = getopt.getopt(sys.argv[1:],"hg:q:S:Q:d:C:t:p:cv:")
|
|
|
58 except getopt.GetoptError, err:
|
|
|
59 print str(err)
|
|
|
60 help()
|
|
|
61 sys.exit(1)
|
|
|
62 for o,a in opts:
|
|
|
63 if o == "-h":
|
|
|
64 help()
|
|
|
65 sys.exit(0)
|
|
|
66 elif o == "-g":
|
|
|
67 groupid = a
|
|
|
68 elif o == "-q":
|
|
|
69 queryDir = a
|
|
|
70 elif o == "-S":
|
|
|
71 patternSuffix = a
|
|
|
72 elif o == "-Q":
|
|
|
73 queue = a
|
|
|
74 elif o == "-d":
|
|
|
75 tmpDir = a
|
|
|
76 elif o == "-C":
|
|
|
77 configFileName = a
|
|
|
78 elif o == "-t":
|
|
|
79 jobTable = a
|
|
|
80 elif o == "-p":
|
|
|
81 projectDir = a
|
|
|
82 elif o == "-c":
|
|
|
83 clean = True
|
|
|
84 elif o == "-v":
|
|
|
85 verbose = int(a)
|
|
|
86
|
|
|
87 if groupid == "" or queryDir == "":
|
|
|
88 print "*** Error: missing compulsory options"
|
|
|
89 help()
|
|
|
90 sys.exit(1)
|
|
|
91
|
|
|
92 if os.environ["REPET_JOBS"] == "files" and projectDir == "":
|
|
|
93 print "*** Error: missing compulsory options for jobs management via files"
|
|
|
94 help()
|
|
|
95 sys.exit(1)
|
|
|
96
|
|
|
97 if verbose > 0:
|
|
|
98 print "\nbeginning of %s" % (sys.argv[0].split("/")[-1])
|
|
|
99 sys.stdout.flush()
|
|
|
100
|
|
|
101 #--------------------------------------------------------------------------
|
|
|
102
|
|
|
103 # create the 'log' file
|
|
|
104
|
|
|
105 logFileName = "%s_pid%s.log" % ( groupid, os.getpid() )
|
|
|
106 handler = logging.FileHandler( logFileName )
|
|
|
107 formatter = logging.Formatter( "%(asctime)s %(levelname)s: %(message)s" )
|
|
|
108 handler.setFormatter( formatter )
|
|
|
109 logging.getLogger('').addHandler( handler )
|
|
|
110 logging.getLogger('').setLevel( logging.DEBUG )
|
|
|
111 logging.info( "started" )
|
|
|
112
|
|
|
113
|
|
|
114 # open a connection to the MySQL table
|
|
|
115
|
|
|
116 if configFileName != "":
|
|
|
117 if not os.path.exists( configFileName ):
|
|
|
118 print "*** Error: configuration file '%s' doesn't exist" % ( configFileName )
|
|
|
119 sys.exit(1)
|
|
|
120 config = ConfigParser.ConfigParser()
|
|
|
121 config.readfp( open(configFileName) )
|
|
|
122 host = config.get("repet_env","repet_host")
|
|
|
123 user = config.get("repet_env","repet_user")
|
|
|
124 passwd = config.get("repet_env","repet_pw")
|
|
|
125 dbname = config.get("repet_env","repet_db")
|
|
|
126 else:
|
|
|
127 host = os.environ["REPET_HOST"]
|
|
|
128 user = os.environ["REPET_USER"]
|
|
|
129 passwd = os.environ["REPET_PW"]
|
|
|
130 dbname = os.environ["REPET_DB"]
|
|
|
131
|
|
|
132 if os.environ["REPET_JOBS"] == "files":
|
|
|
133 jobdb = pyRepet.sql.RepetJobMySQL.RepetJob( dbname = projectDir + "/" + os.environ["REPET_DB"] )
|
|
|
134 elif os.environ["REPET_JOBS"] == "MySQL":
|
|
|
135 jobdb = pyRepet.sql.RepetJobMySQL.RepetJob( user, host, passwd, dbname )
|
|
|
136 else:
|
|
|
137 print "*** Error: REPET_JOBS is '%s'" % ( os.environ["REPET_JOBS"] )
|
|
|
138 sys.exit(1)
|
|
|
139
|
|
|
140
|
|
|
141 currentDir = os.getcwd()
|
|
|
142 if tmpDir == "":
|
|
|
143 tmpDir = currentDir
|
|
|
144
|
|
|
145 # launch PhyML on each fasta file in queryDir
|
|
|
146 cL = pyRepet.launcher.Launcher.PhyMlLauncher( jobdb=jobdb, query=queryDir, cdir=currentDir, tmpdir=tmpDir, job_table=jobTable, queue=queue, groupid=groupid, acro="PhyML" )
|
|
|
147 cL.run( patternSuffix )
|
|
|
148
|
|
|
149 # clean
|
|
|
150 if clean == True:
|
|
|
151 cL.clean()
|
|
|
152
|
|
|
153
|
|
|
154 logging.info( "finished" )
|
|
|
155
|
|
|
156 if verbose > 0:
|
|
|
157 print "%s finished successfully\n" % (sys.argv[0].split("/")[-1])
|
|
|
158 sys.stdout.flush()
|
|
|
159
|
|
|
160 return 0
|
|
|
161
|
|
|
162 #----------------------------------------------------------------------------
|
|
|
163
|
|
|
164 if __name__ == '__main__':
|
|
|
165 main()
|
