def align_to_germline(self, alignment, avg_len=None, avg_mut=None):
if avg_len is not None and avg_mut is not None:
alignment.v_gene = self.v_germlines.get_ties(
alignment.v_gene, avg_len, avg_mut)
alignment.j_gene = self.j_germlines.get_ties(
alignment.j_gene, avg_len, avg_mut)
# Set the germline to the V gene up to the CDR3
germ = get_common_seq(
[self.v_germlines[v] for v in alignment.v_gene], cutoff=False
)
alignment.germline = germ[:CDR3_OFFSET]
# If we need to pad the sequence, do so, otherwise trim the sequence to
# the germline length
if alignment.seq_offset >= 0:
alignment.sequence.pad(alignment.seq_offset)
else:
alignment.sequence.remove_prefix(-alignment.seq_offset)
alignment.j_anchor_pos += alignment.seq_offset
# Add germline gaps to sequence before CDR3 and update anchor positions
for i, c in enumerate(alignment.germline):
if c == '-':
alignment.sequence.add_gap(i)
alignment.j_anchor_pos += 1
if i < alignment.seq_start:
alignment.seq_offset += 1
j_germ = get_common_seq(
[self.j_germlines[j] for j in alignment.j_gene], right=True
)
# Calculate the length of the CDR3
alignment.cdr3_num_nts = (
alignment.j_anchor_pos + self.j_germlines.anchor_len -
self.j_germlines.upstream_of_cdr3 - alignment.cdr3_start
)
v_end = alignment.seq_start + alignment.num_gaps + alignment.v_length
v_germ = germ[CDR3_OFFSET:v_end]
alignment.germline_cdr3 = ''.join((
v_germ,
'-' * (alignment.cdr3_num_nts -
len(v_germ) -
len(alignment.germline_cdr3)),
alignment.germline_cdr3
))
if alignment.cdr3_num_nts < 3:
raise AlignmentException('CDR3 has no AAs')
alignment.j_anchor_pos += alignment.cdr3_num_nts
# Fill germline CDR3 with gaps
alignment.germline += '-' * alignment.cdr3_num_nts
alignment.germline += j_germ[-self.j_germlines.upstream_of_cdr3:]
# If the sequence is longer than the germline, trim it
if len(alignment.sequence) > len(alignment.germline):
alignment.sequence.trim_right(len(alignment.germline))
elif len(alignment.sequence) < len(alignment.germline):
alignment.sequence.pad_right(len(alignment.germline) -
len(alignment.sequence))
def align_to_germline(self, alignment, avg_len=None, avg_mut=None):
if avg_len is not None and avg_mut is not None:
alignment.v_gene = self.v_germlines.get_ties(
alignment.v_gene, avg_len, avg_mut)
alignment.j_gene = self.j_germlines.get_ties(
alignment.j_gene, avg_len, avg_mut)
# Set the germline to the V gene up to the CDR3
germ = get_common_seq(
[self.v_germlines[v] for v in alignment.v_gene], cutoff=False
)
alignment.germline = germ[:CDR3_OFFSET]
# If we need to pad the sequence, do so, otherwise trim the sequence to
# the germline length
if alignment.seq_offset >= 0:
alignment.sequence.pad(alignment.seq_offset)
else:
alignment.sequence.remove_prefix(-alignment.seq_offset)
alignment.j_anchor_pos += alignment.seq_offset
# Add germline gaps to sequence before CDR3 and update anchor positions
for i, c in enumerate(alignment.germline):
if c == '-':
alignment.sequence.add_gap(i)
alignment.j_anchor_pos += 1
if i < alignment.seq_start:
alignment.seq_offset += 1
j_germ = get_common_seq(
[self.j_germlines[j] for j in alignment.j_gene], right=True
)
# Calculate the length of the CDR3
alignment.cdr3_num_nts = (
alignment.j_anchor_pos + self.j_germlines.anchor_len -
self.j_germlines.upstream_of_cdr3 - alignment.cdr3_start
)
v_end = alignment.seq_start + alignment.num_gaps + alignment.v_length
v_germ = germ[CDR3_OFFSET:v_end]
alignment.germline_cdr3 = ''.join((
v_germ,
'-' * (alignment.cdr3_num_nts -
len(v_germ) -
len(alignment.germline_cdr3)),
alignment.germline_cdr3
))
if alignment.cdr3_num_nts < 3:
raise AlignmentException('CDR3 has no AAs')
alignment.j_anchor_pos += alignment.cdr3_num_nts
# Fill germline CDR3 with gaps
alignment.germline += '-' * alignment.cdr3_num_nts
alignment.germline += j_germ[-self.j_germlines.upstream_of_cdr3:]
# If the sequence is longer than the germline, trim it
if len(alignment.sequence) > len(alignment.germline):
alignment.sequence.trim_right(len(alignment.germline))
elif len(alignment.sequence) < len(alignment.germline):
alignment.sequence.pad_right(len(alignment.germline) -
len(alignment.sequence))
def all_ties(self, length, mutation, cutoff=True):
ties = {}
for name in self:
tie_name = tuple(sorted(self.get_ties([name], length, mutation)))
if tie_name not in ties:
ties[tie_name] = get_common_seq([self[n] for n in tie_name],
cutoff=cutoff)
return ties
def all_ties(self, length, mutation, cutoff=True):
ties = {}
for name in self:
tie_name = tuple(sorted(self.get_ties([name], length, mutation)))
if tie_name not in ties:
ties[tie_name] = get_common_seq(
[self[n] for n in tie_name], cutoff=cutoff
)
return ties
def create_alignment(seq, line, v_germlines, j_germlines):
alignment = VDJAlignment(seq)
alignment.v_gene = set([GeneName(g) for g in line['V_CALL'].split(',')])
alignment.j_gene = set([GeneName(g) for g in line['J_CALL'].split(',')])
alignment.cdr3_num_nts = int(line['JUNCTION_LENGTH'])
alignment.v_length = int(line['V_SEQ_LENGTH'])
alignment.seq_offset = re.match(r'[-]*', alignment.sequence.sequence).end()
# TODO: Calculate these
alignment.v_length = CDR3_OFFSET - seq[:CDR3_OFFSET].count('-')
alignment.j_length = j_germlines.upstream_of_cdr3
germ_v = [v_germlines[g] for g in alignment.v_gene if g in v_germlines]
germ_j = [j_germlines[g] for g in alignment.j_gene if g in j_germlines]
if len(germ_v) == 0 or len(germ_j) == 0:
raise AlignmentException('Missing germlines: V={} J={}'.format(
','.join([str(v) for v in alignment.v_gene]),
','.join([str(j) for j in alignment.j_gene])))
germ_v = get_common_seq(germ_v)
germ_j = get_common_seq(germ_j)
alignment.germline_cdr3 = ''.join((
germ_v,
'-' * (len(alignment.sequence) - len(germ_v) - len(germ_j)),
germ_j
))[CDR3_OFFSET:CDR3_OFFSET + alignment.cdr3_num_nts]
alignment.germline = ''.join([
germ_v[:CDR3_OFFSET],
'-' * alignment.cdr3_num_nts,
germ_j[-j_germlines.upstream_of_cdr3:]
])
alignment.sequence.pad_right(
len(alignment.germline) -
len(alignment.sequence.sequence)
)
if len(alignment.germline) != len(alignment.sequence.sequence):
raise AlignmentException('Sequence and germline differ in size')
return alignment
def create_alignment(seq, line, v_germlines, j_germlines):
alignment = VDJAlignment(seq)
alignment.v_gene = set([GeneName(g) for g in line['V_CALL'].split(',')])
alignment.j_gene = set([GeneName(g) for g in line['J_CALL'].split(',')])
alignment.cdr3_num_nts = int(line['JUNCTION_LENGTH'])
alignment.v_length = int(line['V_SEQ_LENGTH'])
alignment.seq_offset = re.match('[\-]*', alignment.sequence.sequence).end()
# TODO: Calculate these
alignment.v_length = CDR3_OFFSET - seq[:CDR3_OFFSET].count('-')
alignment.j_length = j_germlines.upstream_of_cdr3
germ_v = [v_germlines[g] for g in alignment.v_gene if g in v_germlines]
germ_j = [j_germlines[g] for g in alignment.j_gene if g in j_germlines]
if len(germ_v) == 0 or len(germ_j) == 0:
raise AlignmentException('Missing germlines: V={} J={}'.format(
','.join([str(v) for v in alignment.v_gene]),
','.join([str(j) for j in alignment.j_gene])))
germ_v = get_common_seq(germ_v)
germ_j = get_common_seq(germ_j)
alignment.germline_cdr3 = ''.join(
(germ_v, '-' * (len(alignment.sequence) - len(germ_v) - len(germ_j)),
germ_j))[CDR3_OFFSET:CDR3_OFFSET + alignment.cdr3_num_nts]
alignment.germline = ''.join([
germ_v[:CDR3_OFFSET], '-' * alignment.cdr3_num_nts,
germ_j[-j_germlines.upstream_of_cdr3:]
])
alignment.sequence.pad_right(
len(alignment.germline) - len(alignment.sequence.sequence))
if len(alignment.germline) != len(alignment.sequence.sequence):
raise AlignmentException('Sequence and germline differ in size')
return alignment
def align_to_germline(self, avg_len=None, avg_mut=None, trim_to=None):
if avg_len is not None and avg_mut is not None:
self._v = self.v_germlines.get_ties(self.v_gene, avg_len, avg_mut)
self._j = self.j_germlines.get_ties(self.j_gene, avg_len, avg_mut)
# Set the germline to the V gene up to the CDR3
self.germline = get_common_seq([self.v_germlines[v]
for v in self._v])[:CDR3_OFFSET]
# If we need to pad the sequence, do so, otherwise trim the sequence to
# the germline length
if self._pad_len >= 0:
self.sequence = 'N' * self._pad_len + str(self.sequence)
if self.quality is not None:
self.quality = (' ' * self._pad_len) + self.quality
else:
self.removed_prefix = self.sequence[:-self._pad_len]
self.sequence = str(self.sequence[-self._pad_len:])
if self.quality is not None:
self.removed_prefix_qual = self.quality[:-self._pad_len]
self.quality = self.quality[-self._pad_len:]
# Update the anchor positions after adding padding / trimming
self.j_anchor_pos += self._pad_len
# Add germline gaps to sequence before CDR3 and update anchor positions
for i, c in enumerate(self.germline):
if c == '-':
self.sequence = self.sequence[:i] + '-' + self.sequence[i:]
if self.quality is not None:
self.quality = self.quality[:i] + ' ' + self.quality[i:]
self.j_anchor_pos += 1
j_germ = get_common_seq(
map(reversed, [self.j_germlines[j] for j in self.j_gene]))
j_germ = ''.join(reversed(j_germ))
# Calculate the length of the CDR3
self._cdr3_len = (self.j_anchor_pos + self.j_germlines.anchor_len -
self.j_germlines.upstream_of_cdr3 - self.cdr3_start)
if self._cdr3_len < 3:
raise AlignmentException('CDR3 has no AAs'.format(self._cdr3_len))
self.j_anchor_pos += self._cdr3_len
# Fill germline CDR3 with gaps
self.germline += '-' * self._cdr3_len
self.germline += j_germ[-self.j_germlines.upstream_of_cdr3:]
# If the sequence is longer than the germline, trim it
if len(self.sequence) > len(self.germline):
self.sequence = self.sequence[:len(self.germline)]
if self.quality is not None:
self.quality = self.quality[:len(self.germline)]
elif len(self.sequence) < len(self.germline):
self.sequence += 'N' * (len(self.germline) - len(self.sequence))
if self.quality is not None:
self.quality += ' ' * (len(self.germline) - len(self.quality))
if trim_to is not None:
old_padding = max(self._pad_len, 0)
new_prefix = ''.join(
[c if c == '-' else 'N' for c in self.sequence[:trim_to]])
self.sequence = new_prefix + self.sequence[trim_to:]
v_start = re.match('[N\-]*', self.sequence).span()[1]
self._pad_len = self.sequence[:v_start].count('N')
self.v_length -= self._pad_len - old_padding
# Get the pre-CDR3 germline
pre_cdr3_germ = self.germline[:self.cdr3_start]
pre_cdr3_seq = self.sequence[:self.cdr3_start]
# If there is padding, get rid of it in the sequence and align the
# germline
if self._pad_len > 0:
pre_cdr3_germ = pre_cdr3_germ[self._pad_len:]
pre_cdr3_seq = pre_cdr3_seq[self._pad_len:]
# Calculate the pre-CDR3 length and distance
self.pre_cdr3_length = len(pre_cdr3_seq)
self.pre_cdr3_match = self.pre_cdr3_length - dnautils.hamming(
str(pre_cdr3_seq), str(pre_cdr3_germ))
# Get the length of J after the CDR3
self.post_cdr3_length = self.j_germlines.upstream_of_cdr3
# Get the sequence and germline sequences after CDR3
post_j = j_germ[-self.post_cdr3_length:]
post_s = self.sequence[-self.post_cdr3_length:]
# Calculate their match count
self.post_cdr3_match = self.post_cdr3_length - dnautils.hamming(
post_j, post_s)
self.v_match = self.v_length - dnautils.hamming(
self.germline[:self.cdr3_start], self.sequence[:self.cdr3_start])
self.j_match = self.j_length - dnautils.hamming(
self.germline[-len(j_germ):], self.sequence[-len(j_germ):])