annotate commons/launcher/BlatProgramLauncher.py @ 18:94ab73e8a190

Uploaded
author m-zytnicki
date Mon, 29 Apr 2013 03:20:15 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
18
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
1 #!/usr/bin/env python
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
2
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
3 ##@file
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
4 # Launch Blat (pairwise alignment).
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
5 #
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
6 # options:
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
7 # -h: this help
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
8 # -i: name of the input file (queries, format='fasta')
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
9 # -s: name of the subject file (format='fasta')
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
10 # -p: parameters for 'blat' (default='-d 2')
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
11 # -o: name of the output file (format='align', default=inFile+'.align')
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
12 # -c: clean
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
13 # -v: verbosity level (default=0/1)
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
14
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
15
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
16 import os
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
17 import sys
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
18
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
19 from commons.pyRepetUnit.components.AbstractProgramLauncher import AbstractProgramLauncher
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
20
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
21
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
22 class BlatProgramLauncher( AbstractProgramLauncher ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
23 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
24 Launch Blat (pairwise alignment).
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
25 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
26
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
27 def __init__( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
28 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
29 Constructor.
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
30 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
31 AbstractProgramLauncher.__init__( self )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
32 self._prgName = "blat"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
33 self._formatInFile = "fasta"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
34 self._sbjFile = ""
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
35 self._prgParam = ""
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
36 self._allByAll = False
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
37
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
38
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
39 def getHelpAsString( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
40 string = AbstractProgramLauncher.getHelpAsString(self)
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
41 string += "\nspecific options:"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
42 string += "\n -s: name of the subject file (database, format='fasta')"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
43 string += "\n -p: parameters for '%s'" % ( self.getProgramName() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
44 string += "\n -A: same sequences (all-by-all)"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
45 string += "\n -o: name of the output file (format='align', default=inFile+'.align')"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
46 return string
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
47
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
48 def getCmdLineOptions(self):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
49 return AbstractProgramLauncher.getCmdLineOptions(self) + "s:p:Ao:"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
50
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
51 def setAttributesFromCmdLine( self, o, a = "" ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
52 AbstractProgramLauncher.setAttributesFromCmdLine(self, o, a)
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
53 if o == "-s":
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
54 self.setSubjectFile( a )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
55 elif o == "-p":
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
56 self.setProgramParameters( a )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
57 elif o == "-A":
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
58 self.setAllByAll()
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
59 elif o == "-o":
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
60 self.setOutputFile( a )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
61
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
62
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
63 def setSubjectFile( self, arg ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
64 self._sbjFile = arg
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
65
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
66
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
67 def getSubjectFile( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
68 return self._sbjFile
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
69
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
70
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
71 def setAllByAll( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
72 self._allByAll = True
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
73
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
74
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
75 def getAllByAll( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
76 return self._allByAll
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
77
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
78
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
79 def check( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
80 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
81 Check the specific attributes before running the program.
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
82 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
83 AbstractProgramLauncher.check(self)
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
84 if self._sbjFile == "":
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
85 string = "ERROR: missing subject file (-s)"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
86 print string
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
87 print self.getHelpAsString()
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
88 sys.exit(1)
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
89 if self.getOutputFile() == "":
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
90 self.setOutputFile( "%s.align" % ( self.getInputFile() ) )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
91
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
92
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
93 def setWrapperCommandLine( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
94 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
95 Set the command-line of the wrapper.
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
96 Required for BlatClusterLauncher.
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
97 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
98 self._wrpCmdLine = self.getWrapperName()
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
99 self._wrpCmdLine += " -i %s" % ( self.getInputFile() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
100 self._wrpCmdLine += " -s %s" % ( self.getSubjectFile() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
101 if self.getProgramParameters() != "":
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
102 self._wrpCmdLine += " -p '%s'" % ( self.getProgramParameters() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
103 if self.getAllByAll():
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
104 self._wrpCmdLine += " -A"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
105 if self.getOutputFile() == "":
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
106 self.setOutputFile( "%s.align" % ( self.getInputFile() ) )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
107 self._wrpCmdLine += " -o %s" % ( self.getOutputFile() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
108 if self.getClean():
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
109 self._wrpCmdLine += " -c"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
110 self._wrpCmdLine += " -v %i" % ( self.getVerbosityLevel() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
111
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
112
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
113 def setProgramCommandLine( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
114 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
115 Set the command-line of the program.
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
116 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
117 self._prgCmdLine = self.getProgramName()
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
118 self._prgCmdLine += " %s" % ( self.getSubjectFile() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
119 self._prgCmdLine += " %s" % ( self.getInputFile() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
120 if self.getProgramParameters() != "":
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
121 self._prgCmdLine += " %s" % ( self.getProgramParameters() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
122 self._prgCmdLine += " -out=blast8"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
123 self._prgCmdLine += " %s.blast" % ( self.getInputFile() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
124
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
125
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
126 def setListFilesToKeep( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
127 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
128 Set the list of files to keep.
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
129 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
130 if self.getOutputFile() == "":
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
131 self.setOutputFile( "%s.align" % ( self.getInputFile() ) )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
132 self.appendFileToKeep( self.getOutputFile() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
133
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
134
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
135 def setListFilesToRemove( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
136 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
137 Set the list of files to remove.
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
138 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
139 self.appendFileToRemove( "%s.blast" % ( self.getInputFile() ) )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
140
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
141
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
142 def convertBlastIntoAlign( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
143 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
144 Convert a 'blast' file into the 'align' format.
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
145 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
146 cmd = os.environ["REPET_PATH"] + "/bin/blast2align.py"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
147 cmd += " -i %s.blast" % ( self.getInputFile() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
148 cmd += " -o %s" % ( self.getOutputFile() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
149 exitStatus = os.system( cmd )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
150 if exitStatus != 0:
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
151 string = "ERROR while converting 'blast' file into 'align' format"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
152 print string
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
153 sys.exit(1)
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
154
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
155
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
156 def setSummary( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
157 self._summary = "input file: %s" % ( self.getInputFile() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
158 self._summary += "\nsubject file: %s" % ( self.getSubjectFile() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
159 self._summary += "\nparameters: %s" % ( self.getProgramParameters() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
160 if self.getAllByAll():
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
161 self._summary += "\nall-by-all"
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
162 if self.getOutputFile() == "":
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
163 self.setOutputFile( "%s.align" % ( self.getInputFile() ) )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
164 self._summary += "\noutput file: %s" % ( self.getOutputFile() )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
165
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
166
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
167 def run( self ):
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
168 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
169 Run the program.
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
170 """
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
171 self.start()
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
172
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
173 self.setProgramCommandLine()
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
174 cmd = self.getProgramCommandLine()
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
175 if self.getVerbosityLevel() > 0:
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
176 print "LAUNCH: %s" % ( cmd )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
177 sys.stdout.flush()
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
178 exitStatus = os.system( cmd )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
179 if exitStatus != 0:
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
180 string = "ERROR: program '%s' returned exit status '%i'" % ( self.getProgramName(), exitStatus )
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
181 print string
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
182 sys.exit(1)
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
183
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
184 self.convertBlastIntoAlign()
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
185
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
186 self.end()
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
187
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
188
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
189 if __name__ == "__main__":
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
190 i = BlatProgramLauncher()
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
191 i.checkAttributesFromCmdLine()
94ab73e8a190 Uploaded
m-zytnicki
parents:
diff changeset
192 i.run()