def mergeFile( inFile, outFile="" ):
if outFile == "":
outFile = "%s.merged" % ( inFile )
if os.path.exists( outFile ):
os.remove( outFile )
tmpFile = "%s.sorted" % ( inFile )
AlignUtils.sortAlignFile( inFile, tmpFile )
tmpF = open( tmpFile, "r" )
dQrySbj2Aligns = {}
prevPairQrySbj = ""
while True:
line = tmpF.readline()
if line == "":
break
iAlign = Align()
iAlign.setFromString( line )
pairQrySbj = "%s_%s" % ( iAlign.getQueryName(), iAlign.getSubjectName() )
if not dQrySbj2Aligns.has_key( pairQrySbj ):
if prevPairQrySbj != "":
lMerged = AlignUtils.mergeList( dQrySbj2Aligns[ prevPairQrySbj ] )
AlignUtils.writeListInFile( lMerged, outFile, "a" )
del dQrySbj2Aligns[ prevPairQrySbj ]
prevPairQrySbj = pairQrySbj
else:
prevPairQrySbj = pairQrySbj
dQrySbj2Aligns[ pairQrySbj ] = []
dQrySbj2Aligns[ pairQrySbj ].append( iAlign )
lMerged = []
if len(dQrySbj2Aligns.keys()) > 0:
lMerged = AlignUtils.mergeList( dQrySbj2Aligns[ prevPairQrySbj ] )
AlignUtils.writeListInFile( lMerged, outFile, "a" )
tmpF.close()
os.remove( tmpFile )
def getAlignListFromFile( inFile ):
lAlignInstances = []
inFileHandler = open( inFile, "r" )
while True:
line = inFileHandler.readline()
if line == "":
break
a = Align()
a.setFromString( line )
lAlignInstances.append( a )
inFileHandler.close()
return lAlignInstances
def getScoreListFromFile( inFile ):
lScores = []
inFileHandler = open( inFile, "r" )
iAlign = Align()
while True:
line = inFileHandler.readline()
if line == "":
break
iAlign.reset()
iAlign.setFromString( line )
lScores.append( iAlign.score )
inFileHandler.close()
return lScores
def convertAlignFileIntoMapFileWithSubjectsOnQueries( alignFile, mapFile ):
alignFileHandler = open( alignFile, "r" )
mapFileHandler = open( mapFile, "w" )
iAlign = Align()
while True:
line = alignFileHandler.readline()
if line == "":
break
iAlign.setFromString( line )
iMapQ = iAlign.getSubjectAsMapOfQuery()
iMapQ.write( mapFileHandler )
alignFileHandler.close()
mapFileHandler.close()
def convertAlignFileIntoPathFile( alignFile, pathFile ):
alignFileHandler = open( alignFile, "r" )
pathFileHandler = open( pathFile, "w" )
iAlign = Align()
countAlign = 0
while True:
line = alignFileHandler.readline()
if line == "":
break
countAlign += 1
iAlign.setFromString( line, "\t" )
pathFileHandler.write( "%i\t%s\n" % ( countAlign, iAlign.toString() ) )
alignFileHandler.close()
pathFileHandler.close()
def getAlignInstance(self):
iAlign = Align()
lAttributes = []
lAttributes.append( self.range_query.seqname )
lAttributes.append( self.range_query.start )
lAttributes.append( self.range_query.end )
lAttributes.append( self.range_subject.seqname )
lAttributes.append( self.range_subject.start )
lAttributes.append( self.range_subject.end )
lAttributes.append( self.e_value )
lAttributes.append( self.score )
lAttributes.append( self.identity )
iAlign.setFromTuple( lAttributes )
return iAlign
def convertAlignFileIntoMapFileWithQueriesAndSubjects( alignFile, mapFile ):
alignFileHandler = open( alignFile, "r" )
mapFileHandler = open( mapFile, "w" )
iAlign = Align()
while True:
line = alignFileHandler.readline()
if line == "":
break
iAlign.setFromString( line )
iMapQ, iMapS = iAlign.getMapsOfQueryAndSubject()
iMapQ.write( mapFileHandler )
iMapS.write( mapFileHandler )
alignFileHandler.close()
mapFileHandler.close()
def filterRedundantMatches( inFile, outFile ):
"""
When a pairwise alignment is launched ~ all-by-all (ie one batch against all chunks),
one filters the redundant matches. For instance we keep 'chunk3-1-100-chunk7-11-110-...'
and we discards 'chunk7-11-110-chunk3-1-100-...'.
Also we keep 'chunk5-1-100-chunk5-11-110-...' and we discards
'chunk5-11-110-chunk5-1-100-...'.
For this of course the results need to be sorted by query, on plus strand,
and in ascending coordinates (always the case with Blaster).
"""
inFileHandler = open( inFile, "r" )
outFileHandler = open( outFile, "w" )
iAlign = Align()
countMatches = 0
tick = 100000
while True:
line = inFileHandler.readline()
if line == "":
break
countMatches += 1
iAlign.setFromString( line )
if "chunk" not in iAlign.range_query.seqname \
or "chunk" not in iAlign.range_subject.seqname:
print "ERROR: 'chunk' not in seqname"
sys.exit(1)
if int(iAlign.range_query.seqname.split("chunk")[1]) < int(iAlign.range_subject.seqname.split("chunk")[1]):
iAlign.write( outFileHandler )
elif int(iAlign.range_query.seqname.split("chunk")[1]) == int(iAlign.range_subject.seqname.split("chunk")[1]):
if iAlign.range_query.getMin() < iAlign.range_subject.getMin():
iAlign.write( outFileHandler )
if countMatches % tick == 0: # need to free buffer frequently as file can be big
outFileHandler.flush()
os.fsync( outFileHandler.fileno() )
inFileHandler.close()
outFileHandler.close()
def retrieveInitialSequenceHeadersForAlignFile( self, dNew2Init ):
inFileHandler = open( self._inFile, "r" )
outFileHandler = open( self._outFile, "w" )
a = Align()
while True:
line = inFileHandler.readline()
if line == "":
break
a.setFromTuple( line.split("\t") )
nameToBeReplaced = a.range_query.seqname
if dNew2Init.has_key( nameToBeReplaced ):
a.range_query.seqname = dNew2Init[ nameToBeReplaced ]
nameToBeReplaced = a.range_subject.seqname
if dNew2Init.has_key( nameToBeReplaced ):
a.range_subject.seqname = dNew2Init[ nameToBeReplaced ]
a.write( outFileHandler )
inFileHandler.close()
outFileHandler.close()
def updateScoresInFile( inFile, outFile ):
inHandler = open( inFile, "r" )
outHandler = open( outFile, "w" )
iAlign = Align()
while True:
line = inHandler.readline()
if line == "":
break
iAlign.reset()
iAlign.setFromString( line, "\t" )
iAlign.updateScore()
iAlign.write( outHandler )
inHandler.close()
outHandler.close()
def toString(self):
string = "%i" % ( self.id )
string += "\t%s" % (Align.toString(self))
return string
def reset(self):
self.id = -1
Align.reset(self)
def setFromTuple(self, tuple):
self.id = int(tuple[0])
Align.setFromTuple(self, tuple[1:])
def __eq__(self, o):
if self.id != o.id:
return False
else:
return Align.__eq__(self, o)
def __init__( self, id=-1, range_q=Range(), range_s=Range(), e_value=0, score=0, identity=0 ):
self.id = int( id )
Align.__init__( self, range_q, range_s, e_value, score, identity )
