def alignIndels( all_alleles, colcounts, extend_by = 0 ):
'''align all indel-regions.'''
aa = alignlib_lite.py_makeAlignatorDPFull( alignlib_lite.py_ALIGNMENT_LOCAL, 0, 0 )
alignator = alignlib_lite.py_makeMultipleAlignatorSimple( aa)
ids = all_alleles.keys()
for x,c in enumerate(colcounts):
if c <= 1: continue
sequences = alignlib_lite.py_StringVector()
for sid in ids:
for allele in all_alleles[sid]:
sequences.append( allele[x] )
mali = alignlib_lite.py_makeMultAlignment()
alignator.align( mali, sequences )
realigned = []
for line in str(alignlib_lite.py_MultAlignmentFormatPlain( mali, sequences )).split("\n")[:-1]:
data = line[:-1].split("\t")
realigned.append( data[1] )
assert len(realigned) == len(sequences)
l = max( [len(r) for r in realigned] )
i = 0
for sid in ids:
for allele in all_alleles[sid]:
if realigned[i]: allele[x] = realigned[i]
else: allele[x] = "-" * l
i += 1
colcounts[x] = l
def convertAlignlib2Mali( mali, identifiers = None, seqs = None ):
"""convert a multiple alignment into an alignlib_lite.py_multiple alignment object."""
m = Mali()
if not identifiers:
identifiers = [ "%i" % x for x in range( mali.getNumSequences() ) ]
if seqs == None:
# old style MultipleAlignment
for x in range(mali.getNumSequences()):
a = mali.getRow( x )
m.addSequence( identifiers[x], a.getFrom(), a.getTo(), a.getString() )
else:
import alignlib_lite
output = alignlib_lite.py_MultAlignmentFormatPlain( mali, seqs )
for x in range(mali.getNumSequences()):
a = output.mData[ x ]
m.addSequence( identifiers[x], a.getFrom(), a.getTo(), a.getString() )
return m
def convertAlignlib2Mali(mali, identifiers=None, seqs=None):
"""convert a multiple alignment into an alignlib_lite.py_multiple alignment object."""
m = Mali()
if not identifiers:
identifiers = ["%i" % x for x in range(mali.getNumSequences())]
if seqs == None:
# old style MultipleAlignment
for x in range(mali.getNumSequences()):
a = mali.getRow(x)
m.addSequence(identifiers[x], a.getFrom(), a.getTo(),
a.getString())
else:
import alignlib_lite
output = alignlib_lite.py_MultAlignmentFormatPlain(mali, seqs)
for x in range(mali.getNumSequences()):
a = output.mData[x]
m.addSequence(identifiers[x], a.getFrom(), a.getTo(),
a.getString())
return m