def _loadFromFile(self, fileName, ignoreFirst=False, keepTermGaps=False, referenceFirst=False):
reader = SimpleFastaReader(fileName)
self.msa = MSA()
first = True
for entry in reader:
if ignoreFirst and first:
first = False
continue
aSeq = AlignedSequence( entry.sequence, aligned=True, keepTermGaps=keepTermGaps )
sStart = entry.getAnnotation('start')
if sStart: aSeq.alignedRange.addDelta( int(sStart) )
strand = entry.getAnnotation('strand')
if not strand: strand = '+'
isReference = False
if referenceFirst == True and first:
first = False
isReference = True
self.msa.add( aSeq, name=entry.name, strand=strand, isReference=isReference )
reader.close()
class MsaReader:
def __init__(self, fileName, ignoreFirst=False, iterate=False, keepTermGaps=False, referenceFirst=False):
self._fileName = fileName
self.ignoreFirst = ignoreFirst
if not iterate:
self._loadFromFile(self._fileName,ignoreFirst=ignoreFirst, keepTermGaps=keepTermGaps, referenceFirst=referenceFirst)
def _loadFromFile(self, fileName, ignoreFirst=False, keepTermGaps=False, referenceFirst=False):
reader = SimpleFastaReader(fileName)
self.msa = MSA()
first = True
for entry in reader:
if ignoreFirst and first:
first = False
continue
aSeq = AlignedSequence( entry.sequence, aligned=True, keepTermGaps=keepTermGaps )
sStart = entry.getAnnotation('start')
if sStart: aSeq.alignedRange.addDelta( int(sStart) )
strand = entry.getAnnotation('strand')
if not strand: strand = '+'
isReference = False
if referenceFirst == True and first:
first = False
isReference = True
self.msa.add( aSeq, name=entry.name, strand=strand, isReference=isReference )
reader.close()
def nameStrandSeqIter( self ):
reader = SimpleFastaReader( self._fileName )
first = True
for entry in reader:
if self.ignoreFirst and first:
first = False
continue
aSeq = AlignedSequence( entry.sequence, aligned=True )
sStart = entry.getAnnotation('start')
if sStart:
aSeq.alignedRange.addDelta( int(sStart) )
strand = entry.getAnnotation('strand')
if not strand:
strand = '+'
yield entry.name, strand, aSeq
reader.close()
def nameStrandCrSeqIter( self ):
reader = SimpleFastaReader( self._fileName )
first = True
for entry in reader:
if self.ignoreFirst and first:
first = False
continue
aSeq = AlignedSequence( entry.sequence, aligned=True )
sStart = entry.getAnnotation('start')
if sStart:
aSeq.alignedRange.addDelta( int(sStart) )
strand = entry.getAnnotation('strand')
if not strand:
strand = '+'
cr = entry.getAnnotation('cr')
yield entry.name, strand, cr, aSeq
reader.close()
def getMsa(self):
return self.msa