def get_alignment_parasail(reference_genome, input_genome):
# the dna full matrix supports ambiguity codes, although "N"s are not given free mismatches as we might like
# the alignments appear good enough for our purpose however
result = parasail.nw_trace_striped_32(input_genome.sequence,
reference_genome.sequence, 10, 1,
parasail.dnafull)
traceback = result.traceback
return traceback.ref, traceback.comp, traceback.query
def build_envelope(len1, seq1, path1, len2, seq2, path2, padding=15):
# needleman-wunsch alignment with constant gap penalty.
aln = parasail.nw_trace_striped_32(seq2, seq1, 2, 2, parasail.dnafull)
# pair up positions
alignment = np.column_stack([
np.cumsum([x != '-' for x in aln.traceback.ref]) - 1,
np.cumsum([x != '-' for x in aln.traceback.query]) - 1
])
path_range1 = np.column_stack([path1, path1[1:] + [len1]])
path_range2 = np.column_stack([path2, path2[1:] + [len2]])
envelope = np.full((len1, 2), -1, dtype=int)
for idx1, idx2 in alignment.clip(0):
st_1, en_1 = path_range1[idx1]
st_2, en_2 = path_range2[idx2]
for idx in range(st_1, en_1):
if st_2 < envelope[idx, 0] or envelope[idx, 0] < 0:
envelope[idx, 0] = st_2
if en_2 > envelope[idx, 1] or envelope[idx, 1] < 0:
envelope[idx, 1] = en_2
# add a little padding to ensure some overlap
envelope[:, 0] = envelope[:, 0] - padding
envelope[:, 1] = envelope[:, 1] + padding
envelope = np.clip(envelope, 0, len2)
prev_end = 0
for i in range(envelope.shape[0]):
if envelope[i, 0] > envelope[i, 1]:
envelope[i, 0] = 0
if envelope[i, 0] > prev_end:
envelope[i, 0] = prev_end
prev_end = envelope[i, 1]
return envelope.astype(np.uint64)
def get_alignment(reference_genome, input_genome):
# the dna full matrix supports ambiguity codes, although "N"s are not given free mismatches as we might like
# the alignments appear good enough for our purpose however
result = parasail.nw_trace_striped_32(input_genome.sequence,
reference_genome.sequence, 10, 1,
parasail.dnafull)
traceback = result.traceback
columns = 120
position_map = list()
reference_index = 0
input_index = 0
for (ref, query) in zip(traceback.ref, traceback.query):
if ref != '-' and query != '-':
position_map.append((reference_index, input_index))
if ref != '-':
reference_index += 1
if query != '-':
input_index += 1
return position_map