def alignAcross(data, field_map, muscle_exec=default_muscle_exec):
"""
Multiple aligns sequence fields column wise
Arguments:
data : DbData object with Receptor objects to process.
field_map : a dictionary of {input sequence : output sequence) field names to multiple align.
muscle_exec : the MUSCLE executable.
Returns:
changeo.Multiprocessing.DbResult : object containing Receptor objects with multiple aligned sequence fields.
"""
# Define return object
result = DbResult(data.id, data.data)
result.results = data.data
result.valid = True
# Fail invalid groups
if result.id is None:
result.log = None
result.valid = False
return result
seq_fields = list(field_map.keys())
for f in seq_fields:
seq_list = [
SeqRecord(r.getSeq(f), id=r.sequence_id.replace(' ', '_'))
for r in data.data
]
seq_aln = runMuscle(seq_list, aligner_exec=muscle_exec)
if seq_aln is not None:
aln_map = {x.id: i for i, x in enumerate(seq_aln)}
for i, r in enumerate(result.results, start=1):
idx = aln_map[r.sequence_id.replace(' ', '_')]
seq = str(seq_aln[idx].seq)
r.annotations[field_map[f]] = seq
result.log['%s-%s' % (f, r.sequence_id)] = seq
else:
result.valid = False
#for r in result.results: print r.annotations
return result
def alignBlocks(data, seq_fields, muscle_exec=default_muscle_exec):
"""
Multiple aligns blocks of sequence fields together
Arguments:
data : a DbData object with IgRecords to process.
seq_fields : the sequence fields to multiple align.
muscle_exec : the MUSCLE executable.
Returns:
changeo.Multiprocessing.DbResult : object containing IgRecords with multiple aligned sequence fields.
"""
# Define return object
result = DbResult(data.id, data.data)
result.results = data.data
result.valid = True
# Fail invalid groups
if result.id is None:
result.log = None
result.valid = False
return result
seq_list = [SeqRecord(r.getSeqField(f), id='%s_%s' % (r.id, f)) for f in seq_fields \
for r in data.data]
seq_aln = runMuscle(seq_list, aligner_exec=muscle_exec)
if seq_aln is not None:
aln_map = {x.id: i for i, x in enumerate(seq_aln)}
for i, r in enumerate(result.results, start=1):
for f in seq_fields:
idx = aln_map['%s_%s' % (r.id, f)]
seq = str(seq_aln[idx].seq)
r.annotations['%s_ALIGN' % f] = seq
result.log['%s-%s' % (f, r.id)] = seq
else:
result.valid = False
#for r in result.results: print r.annotations
return result
def alignWithin(data, field_map, muscle_exec=default_muscle_exec):
"""
Multiple aligns sequence fields within a row
Arguments:
data : DbData object with Receptor objects to process.
field_map : a dictionary of {input sequence : output sequence) field names to multiple align.
muscle_exec : the MUSCLE executable.
Returns:
changeo.Multiprocessing.DbResult : object containing Receptor objects with multiple aligned sequence fields.
"""
# Define return object
result = DbResult(data.id, data.data)
result.results = data.data
result.valid = True
# Fail invalid groups
if result.id is None:
result.log = None
result.valid = False
return result
record = data.data
seq_fields = list(field_map.keys())
seq_list = [SeqRecord(record.getSeq(f), id=f) for f in seq_fields]
seq_aln = runMuscle(seq_list, aligner_exec=muscle_exec)
if seq_aln is not None:
aln_map = {x.id: i for i, x in enumerate(seq_aln)}
for f in seq_fields:
idx = aln_map[f]
seq = str(seq_aln[idx].seq)
record.annotations[field_map[f]] = seq
result.log[f] = seq
else:
result.valid = False
return result
def alignWithin(data, seq_fields, muscle_exec=default_muscle_exec):
"""
Multiple aligns sequence fields within a row
Arguments:
data : a DbData object with an IgRecords to process.
seq_fields : the sequence fields to multiple align.
muscle_exec : the MUSCLE executable.
Returns:
changeo.Multiprocessing.DbResult : object containing IgRecords with multiple aligned sequence fields.
"""
# Define return object
result = DbResult(data.id, data.data)
result.results = data.data
result.valid = True
# Fail invalid groups
if result.id is None:
result.log = None
result.valid = False
return result
record = data.data
seq_list = [SeqRecord(record.getSeqField(f), id=f) for f in seq_fields]
seq_aln = runMuscle(seq_list, aligner_exec=muscle_exec)
if seq_aln is not None:
aln_map = {x.id: i for i, x in enumerate(seq_aln)}
for f in seq_fields:
idx = aln_map[f]
seq = str(seq_aln[idx].seq)
record.annotations['%s_ALIGN' % f] = seq
result.log[f] = seq
else:
result.valid = False
return result