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

Uploaded
author m-zytnicki
date Mon, 29 Apr 2013 03:20:15 -0400
parents
children
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/commons/launcher/BlatProgramLauncher.py	Mon Apr 29 03:20:15 2013 -0400
@@ -0,0 +1,192 @@
+#!/usr/bin/env python
+
+##@file
+# Launch Blat (pairwise alignment).
+#
+# options:
+#      -h: this help
+#      -i: name of the input file (queries, format='fasta')
+#      -s: name of the subject file (format='fasta')
+#      -p: parameters for 'blat' (default='-d 2')
+#      -o: name of the output file (format='align', default=inFile+'.align')
+#      -c: clean
+#      -v: verbosity level (default=0/1)
+
+
+import os
+import sys
+
+from commons.pyRepetUnit.components.AbstractProgramLauncher import  AbstractProgramLauncher
+
+
+class BlatProgramLauncher( AbstractProgramLauncher ):
+    """
+    Launch Blat (pairwise alignment).
+    """
+    
+    def __init__( self ):
+        """
+        Constructor.
+        """
+        AbstractProgramLauncher.__init__( self )
+        self._prgName = "blat"
+        self._formatInFile = "fasta"
+        self._sbjFile = ""
+        self._prgParam = ""
+        self._allByAll = False
+        
+        
+    def getHelpAsString( self ):
+        string = AbstractProgramLauncher.getHelpAsString(self)
+        string += "\nspecific options:"
+        string += "\n     -s: name of the subject file (database, format='fasta')"
+        string += "\n     -p: parameters for '%s'" % ( self.getProgramName() )
+        string += "\n     -A: same sequences (all-by-all)"
+        string += "\n     -o: name of the output file (format='align', default=inFile+'.align')"
+        return string
+    
+    def getCmdLineOptions(self):
+        return AbstractProgramLauncher.getCmdLineOptions(self) + "s:p:Ao:"
+
+    def setAttributesFromCmdLine( self, o, a = "" ):
+        AbstractProgramLauncher.setAttributesFromCmdLine(self, o, a)
+        if o == "-s":
+            self.setSubjectFile( a )
+        elif o == "-p":
+            self.setProgramParameters( a )
+        elif o == "-A":
+            self.setAllByAll()
+        elif o == "-o":
+            self.setOutputFile( a )
+            
+            
+    def setSubjectFile( self, arg ):
+        self._sbjFile = arg
+        
+        
+    def getSubjectFile( self ):
+        return self._sbjFile
+    
+    
+    def setAllByAll( self ):
+        self._allByAll = True
+        
+        
+    def getAllByAll( self ):
+        return self._allByAll
+    
+    
+    def check( self ):
+        """
+        Check the specific attributes before running the program.
+        """
+        AbstractProgramLauncher.check(self)
+        if self._sbjFile == "":
+            string = "ERROR: missing subject file (-s)"
+            print string
+            print self.getHelpAsString()
+            sys.exit(1)
+        if self.getOutputFile() == "":
+            self.setOutputFile( "%s.align" % ( self.getInputFile() ) )
+            
+            
+    def setWrapperCommandLine( self ):
+        """
+        Set the command-line of the wrapper.
+        Required for BlatClusterLauncher.
+        """
+        self._wrpCmdLine = self.getWrapperName()
+        self._wrpCmdLine += " -i %s" % ( self.getInputFile() )
+        self._wrpCmdLine += " -s %s" % ( self.getSubjectFile() )
+        if self.getProgramParameters() != "":
+            self._wrpCmdLine += " -p '%s'" % ( self.getProgramParameters() )
+        if self.getAllByAll():
+            self._wrpCmdLine += " -A"
+        if self.getOutputFile() == "":
+            self.setOutputFile( "%s.align" % ( self.getInputFile() ) )
+        self._wrpCmdLine += " -o %s" % ( self.getOutputFile() )
+        if self.getClean():
+            self._wrpCmdLine += " -c"
+        self._wrpCmdLine += " -v %i" % ( self.getVerbosityLevel() )
+        
+        
+    def setProgramCommandLine( self ):
+        """
+        Set the command-line of the program.
+        """
+        self._prgCmdLine = self.getProgramName()
+        self._prgCmdLine += " %s" % ( self.getSubjectFile() )
+        self._prgCmdLine += " %s" % ( self.getInputFile() )
+        if self.getProgramParameters() != "":
+            self._prgCmdLine += " %s" % ( self.getProgramParameters() )
+        self._prgCmdLine += " -out=blast8"
+        self._prgCmdLine += " %s.blast" % ( self.getInputFile() )
+        
+        
+    def setListFilesToKeep( self ):
+        """
+        Set the list of files to keep.
+        """
+        if self.getOutputFile() == "":
+            self.setOutputFile( "%s.align" % ( self.getInputFile() ) )
+        self.appendFileToKeep( self.getOutputFile() )
+        
+        
+    def setListFilesToRemove( self ):
+        """
+        Set the list of files to remove.
+        """
+        self.appendFileToRemove( "%s.blast" % ( self.getInputFile() ) )
+        
+        
+    def convertBlastIntoAlign( self ):
+        """
+        Convert a 'blast' file into the 'align' format.
+        """
+        cmd = os.environ["REPET_PATH"] + "/bin/blast2align.py"
+        cmd += " -i %s.blast" % ( self.getInputFile() )
+        cmd += " -o %s" % ( self.getOutputFile() )
+        exitStatus = os.system( cmd )
+        if exitStatus != 0:
+            string = "ERROR while converting 'blast' file into 'align' format"
+            print string
+            sys.exit(1)
+            
+            
+    def setSummary( self ):
+        self._summary = "input file: %s" % ( self.getInputFile() )
+        self._summary += "\nsubject file: %s" % ( self.getSubjectFile() )
+        self._summary += "\nparameters: %s" % ( self.getProgramParameters() )
+        if self.getAllByAll():
+            self._summary += "\nall-by-all"
+        if self.getOutputFile() == "":
+            self.setOutputFile( "%s.align" % ( self.getInputFile() ) )
+        self._summary += "\noutput file: %s" % ( self.getOutputFile() )
+        
+        
+    def run( self ):
+        """
+        Run the program.
+        """
+        self.start()
+        
+        self.setProgramCommandLine()
+        cmd = self.getProgramCommandLine()
+        if self.getVerbosityLevel() > 0:
+            print "LAUNCH: %s" % ( cmd )
+            sys.stdout.flush()
+        exitStatus = os.system( cmd )
+        if exitStatus != 0:
+            string = "ERROR: program '%s' returned exit status '%i'" % ( self.getProgramName(), exitStatus )
+            print string
+            sys.exit(1)
+            
+        self.convertBlastIntoAlign()
+        
+        self.end()
+        
+        
+if __name__ == "__main__":
+    i = BlatProgramLauncher()
+    i.checkAttributesFromCmdLine()
+    i.run()