def dbShuffle( inData, outData, verbose=0 ):
if CheckerUtils.isExecutableInUserPath("esl-shuffle"):
prg = "esl-shuffle"
else : prg = "shuffle"
genericCmd = prg + " -d INPUT > OUTPUT"
if os.path.isfile( inData ):
if verbose > 0:
print "shuffle input file '%s'" % inData
cmd = genericCmd.replace("INPUT",inData).replace("OUTPUT",outData)
print cmd
returnStatus = os.system( cmd )
if returnStatus != 0:
sys.stderr.write( "ERROR: 'shuffle' returned '%i'\n" % returnStatus )
sys.exit(1)
elif os.path.isdir( inData ):
if verbose > 0:
print "shuffle files in input directory '%s'" % inData
if os.path.exists( outData ):
shutil.rmtree( outData )
os.mkdir( outData )
lInputFiles = glob.glob( "%s/*.fa" %( inData ) )
nbFastaFiles = 0
for inputFile in lInputFiles:
nbFastaFiles += 1
if verbose > 1:
print "%3i / %3i" % ( nbFastaFiles, len(lInputFiles) )
fastaBaseName = os.path.basename( inputFile )
prefix, extension = os.path.splitext( fastaBaseName )
cmd = genericCmd.replace("INPUT",inputFile).replace("OUTPUT","%s/%s_shuffle.fa"%(outData,prefix))
returnStatus = os.system( cmd )
if returnStatus != 0:
sys.stderr.write( "ERROR: 'shuffle' returned '%i'\n" % returnStatus )
sys.exit(1)
def launchShuffle( self, outFileName="", run="yes", verbose="no" ):
if CheckerUtils.isExecutableInUserPath("esl-shuffle"):
prg = "esl-shuffle"
else : prg = "shuffle"
cmd = prg
cmd += " -d " + self.inFileName
if outFileName != "":
cmd += " > " + outFileName
if verbose == "yes":
print "launch: %s" % ( cmd ); sys.stdout.flush()
if run == "yes":
return self.launch( prg, cmd, verbose )
elif run == "no":
return cmd
def mergeCoordsInFile( inFile, outFile ):
if not sys.modules.has_key( "pyRepetUnit.commons.checker.CheckerUtils" ):
msg = "WARNING: can't find module 'CheckerUtils'"
sys.stderr.write( "%s\n" % msg )
elif not CheckerUtils.isExecutableInUserPath( "mapOp" ):
msg = "WARNING: can't find executable 'mapOp'"
sys.stderr.write( "%s\n" % msg )
else:
cmd = "mapOp"
cmd += " -q %s" % ( inFile )
cmd += " -m"
cmd += " 2>&1 > /dev/null"
returnStatus = os.system( cmd )
if returnStatus != 0:
print "ERROR: mapOp returned %i" % ( returnStatus )
sys.exit(1)
os.rename( "%s.merge" % inFile,
outFile )
def compareInputSequencesWithMummer( self, nbInSeq ):
"""
Launch MUmmer on two single-sequence fasta files to find all maximal matches regardless of their uniqueness and record stdout.
Only N(N-1)/2 comparisons are made.
@param nbInSeq: number of input sequences
@type nbInSeq: integer
"""
if self._verbose > 0:
print "aligning input sequences..."
sys.stdout.flush()
if not CheckerUtils.isExecutableInUserPath( "mummer" ):
msg = "ERROR: 'mummer' is not in your PATH"
sys.stderr.write( "%s\n" % ( msg ) )
sys.exit(1)
lInFiles = glob.glob( "batch_*.fa" )
for i in range( 1, nbInSeq+1 ):
for j in range( i+1, nbInSeq+1 ):
if self._verbose > 1:
print "launch MUmmer on %i versus %i" % ( i, j )
sys.stdout.flush()
prg = "mummer"
cmd = prg
cmd += " -maxmatch"
cmd += " -l %i" % ( self._minMatchLength )
cmd += " -b"
cmd += " -F"
cmd += " batch_%s.fa" % ( str(j).zfill( len( str( len(lInFiles) ) ) ) )
cmd += " batch_%s.fa" % ( str(i).zfill( len( str( len(lInFiles) ) ) ) )
cmd += " > mummer_%i_vs_%i.txt" % ( i, j )
if self._verbose < 3:
cmd += " 2> /dev/null"
returnStatus = os.system( cmd )
if returnStatus != 0:
msg = "ERROR: '%s' returned '%i'" % ( prg, returnStatus )
sys.stderr.write( "%s\n" % ( msg ) )
sys.exit(1)
def _checkProgramName( self ):
if not CheckerUtils.isExecutableInUserPath( self.getProgramName() ):
msg = "ERROR: '%s' not in your PATH" % ( self.getProgramName() )
sys.stderr.write( "%s\n" %( msg ) )
sys.exit(1)
