def variant_context(
reference_fasta,
contig,
inclusive_start,
inclusive_end,
alt,
context_length):
"""
Retrieve the surronding reference region from a variant.
SNVs are canonicalized so the reference base is a pyrmidine (C/T). For
indels the reverse complement will still be taken if the first base of
the reference is not a pyrmidine, but since the reference will also be
reversed, that doesn't guarantee it will start with a pyrmidine.
Parameters
----------
reference_fasta : FastaReference
reference sequence from pyfaidx package
contig : str
Chromosome of the variant
inclusive_start : int
start of the variant in 1-based inclusive coordinates
inclusive_end : int
end of the variant in 1-based inclusive coordinates
alt : string
alt sequence
context_length : int
number of bases on either side of the variant to return
Returns
---------
A tuple of (5', mutation, 3') where
5' - bases immediately 5 prime to the mutation
3' - bases immediately 3 prime to the mutation
mutation - the ref sequence followed by a > character followed by the
the alt sequence
"""
# Move from 1-base coorindates to 0-base coordinates
start = int(inclusive_start) - 1
end = int(inclusive_end)
full_sequence = reference_fasta[contig]
left = str(full_sequence[start - context_length:start].seq).upper()
middle = str(full_sequence[start: end].seq).upper()
right = str(full_sequence[end: end + context_length].seq).upper()
# Complement and reverse the context if necessary so the ref base is a
# pyrmidine (C/T)
if middle[0] in ('A', 'G'):
context_5prime = pyfaidx.complement(right)[::-1]
context_3prime = pyfaidx.complement(left)[::-1]
context_mutation = "%s>%s" % (
pyfaidx.complement(middle)[::-1], pyfaidx.complement(alt)[::-1])
else:
context_5prime = left
context_3prime = right
context_mutation = "%s>%s" % (middle, alt)
return (context_5prime, context_mutation, context_3prime)
def complement_seq(seq):
return pyfaidx.complement(seq)
def calc_nuc_counts(pos_signal_bedtool, neg_signal_bedtool, fasta_filename,
revcomp_strand, min_counts,
offset_min, offset_max, region_size,
ignore_chroms, only_chroms, verbose):
''' main routine for calculating nuc_counts '''
if verbose:
msg = ">> analyzing sequences ...\n"
msg += ">> ignore:%s only:%s\n" % \
(str(ignore_chroms), str(only_chroms))
msg += ">> offset range: %d to %d\n" % (offset_min, offset_max)
msg += ">> region size: %d\n" % (region_size)
msg += ">> revcomp strand: %s\n" % str(revcomp_strand)
print >>sys.stderr, msg
seq_fasta = Fasta(fasta_filename, as_raw = True)
nuc_counts = defaultdict(Counter)
bedtools = (pos_signal_bedtool, neg_signal_bedtool)
strands = ('+', '-')
# total number of sites examined
total_sites = 0
for bedtool, strand in izip(bedtools, strands):
for row in bedtool:
# skip data based on specified chromosomes
if row.chrom in ignore_chroms:
continue
if only_chroms and row.chrom not in only_chroms:
continue
# skip data if counts are too low
if row.count < min_counts: continue
# sites in bedgraph examined - must come after all checks
# above
total_sites += 1
for offset in range(offset_min, offset_max + 1):
# upstream offsets are negative values
if strand == '+':
start = row.start + offset
elif strand == '-':
start = row.start - offset
if region_size == 1:
# half open at the position of interest
end = start + region_size
else:
# make sure that the 3' most position in a region
# is the base of interest
if strand == '+':
end = start + 1 # include position with + 1
start = end - region_size
else:
# negative strand
end = start + region_size
nucs = seq_fasta[row.chrom][start:end]
# 1. libs where the captured strand is sequenced
# are the correct polarity as-is (i.e. Excision-seq
# libs)
# 2. libs where the *copy* of the captured strand
# is sequenced should be revcomplemented (i.e.
# circularization-based libs)
if (strand == '+' and revcomp_strand) or \
(strand == '-' and not revcomp_strand):
nucs = complement(nucs[::-1])
nuc_counts[offset][nucs] += row.count
# remove nucs that are not len region_size
for offset, counts in nuc_counts.items():
for nuc, count in counts.items():
if len(nuc) != region_size:
counts.pop(nuc)
return total_sites, nuc_counts
def calc_nuc_counts(pos_signal_bedtool, neg_signal_bedtool, fasta_filename,
revcomp_strand, min_counts, offset_min, offset_max,
region_size):
''' main routine for calculating nuc_counts '''
#if verbose:
# msg = ">> analyzing sequences ...\n"
# msg += ">> ignore:%s only:%s\n" % \
# (str(ignore_chroms), str(only_chroms))
# msg += ">> offset range: %d to %d\n" % (offset_min, offset_max)
# msg += ">> region size: %d\n" % (region_size)
# msg += ">> revcomp strand: %s\n" % str(revcomp_strand)
# print >>sys.stderr, msg
seq_fasta = Fasta(fasta_filename, as_raw=True)
nuc_counts = defaultdict(Counter)
bedtools = (pos_signal_bedtool, neg_signal_bedtool)
strands = ('+', '-')
# total number of sites examined
total_sites = 0
for bedtool, strand in zip(bedtools, strands):
for row in bedtool:
# skip data based on specified chromosomes
# if row.chrom in ignore_chroms:
# continue
# if only_chroms and row.chrom not in only_chroms:
# continue
# skip data if counts are too low
if row.count < min_counts: continue
# sites in bedgraph examined - must come after all checks
# above
total_sites += 1
for offset in range(offset_min, offset_max + 1):
# upstream offsets are negative values
if strand == '+':
start = row.start + offset
elif strand == '-':
start = row.start - offset
if region_size == 1:
# half open at the position of interest
end = start + region_size
else:
# make sure that the 3' most position in a region
# is the base of interest
if strand == '+':
end = start + 1 # include position with + 1
start = end - region_size
else:
# negative strand
end = start + region_size
nucs = seq_fasta[row.chrom][start:end]
# 1. libs where the captured strand is sequenced
# are the correct polarity as-is (i.e. Excision-seq
# libs)
# 2. libs where the *copy* of the captured strand
# is sequenced should be revcomplemented (i.e.
# circularization-based libs)
if (strand == '+' and revcomp_strand) or \
(strand == '-' and not revcomp_strand):
nucs = complement(nucs[::-1])
nuc_counts[offset][nucs] += row.count
# remove nucs that are not len region_size
# for offset, counts in nuc_counts.items():
# for nuc, count in counts.items():
# if len(nuc) != region_size:
# counts.pop(nuc)
return total_sites, nuc_counts
def extract(self, interval, variants, anchor, fixed_len=True, use_strand=None, **kwargs):
"""
Args:
interval: pybedtools.Interval Region of interest from
which to query the sequence. 0-based
variants: List[cyvcf2.Variant]: variants overlapping the `interval`.
can also be indels. 1-based
anchor: absolution position w.r.t. the interval start. (0-based).
E.g. for an interval of `chr1:10-20` the anchor of 10 denotes
the point chr1:10 in the 0-based coordinate system.
fixed_len: if True, the return sequence will have the same length
as the `interval` (e.g. `interval.end - interval.start`)
use_strand (bool, optional): if True, the extracted sequence
is reverse complemented in case interval.strand == "-".
Overrides `self.use_strand`
Returns:
A single sequence (`str`) with all the variants applied.
"""
# Preprocessing
anchor = max(min(anchor, interval.end), interval.start)
variant_pairs = self._variant_to_sequence(variants)
# 1. Split variants overlapping with anchor
# and interval start end if not fixed_len
variant_pairs = self._split_overlapping(variant_pairs, anchor)
if not fixed_len:
variant_pairs = self._split_overlapping(
variant_pairs, interval.start, which='right')
variant_pairs = self._split_overlapping(
variant_pairs, interval.end, which='left')
variant_pairs = list(variant_pairs)
# 2. split the variants into upstream and downstream
# and sort the variants in each interval
upstream_variants = sorted(
filter(lambda x: x[0].start >= anchor, variant_pairs),
key=lambda x: x[0].start
)
downstream_variants = sorted(
filter(lambda x: x[0].start < anchor, variant_pairs),
key=lambda x: x[0].start,
reverse=True
)
# 3. Extend start and end position for deletions
if fixed_len:
istart, iend = self._updated_interval(
interval, upstream_variants, downstream_variants)
else:
istart, iend = interval.start, interval.end
# 4. Iterate from the anchor point outwards. At each
# register the interval from which to take the reference sequence
# as well as the interval for the variant
down_sb = self._downstream_builder(
downstream_variants, interval, anchor, istart)
up_sb = self._upstream_builder(
upstream_variants, interval, anchor, iend)
# 5. fetch the sequence and restore intervals in builder
seq = self._fetch(interval, istart, iend)
up_sb.restore(seq)
down_sb.restore(seq)
# 6. Concate sequences from the upstream and downstream splits. Concat
# upstream and downstream sequence. Cut to fix the length.
down_str = down_sb.concat()
up_str = up_sb.concat()
if fixed_len:
down_str, up_str = self._cut_to_fix_len(
down_str, up_str, interval, anchor)
seq = down_str + up_str
if use_strand is None:
use_strand = self.use_strand
if use_strand and interval.strand == '-':
# reverse-complement
seq = complement(seq)[::-1]
return seq
def variant_context(reference_fasta, contig, inclusive_start, inclusive_end,
alt, context_length):
"""
Retrieve the surronding reference region from a variant.
SNVs are canonicalized so the reference base is a pyrmidine (C/T). For
indels the reverse complement will still be taken if the first base of
the reference is not a pyrmidine, but since the reference will also be
reversed, that doesn't guarantee it will start with a pyrmidine.
Parameters
----------
reference_fasta : FastaReference
reference sequence from pyfaidx package
contig : str
Chromosome of the variant
inclusive_start : int
start of the variant in 1-based inclusive coordinates
inclusive_end : int
end of the variant in 1-based inclusive coordinates
alt : string
alt sequence
context_length : int
number of bases on either side of the variant to return
Returns
---------
A tuple of (5', mutation, 3') where
5' - bases immediately 5 prime to the mutation
3' - bases immediately 3 prime to the mutation
mutation - the ref sequence followed by a > character followed by the
the alt sequence
"""
# Move from 1-base coorindates to 0-base coordinates
start = int(inclusive_start) - 1
end = int(inclusive_end)
full_sequence = reference_fasta[contig]
left = str(full_sequence[start - context_length:start].seq).upper()
middle = str(full_sequence[start:end].seq).upper()
right = str(full_sequence[end:end + context_length].seq).upper()
# Complement and reverse the context if necessary so the ref base is a
# pyrmidine (C/T)
if middle[0] in ('A', 'G'):
context_5prime = pyfaidx.complement(right)[::-1]
context_3prime = pyfaidx.complement(left)[::-1]
context_mutation = "%s>%s" % (pyfaidx.complement(middle)[::-1],
pyfaidx.complement(alt)[::-1])
else:
context_5prime = left
context_3prime = right
context_mutation = "%s>%s" % (middle, alt)
return (context_5prime, context_mutation, context_3prime)
def test_comp_empty():
assert complement('') == ''
def test_comp_valid():
assert complement(comp_valid).startswith("AACTTCTAAAnCG")
assert complement(complement(comp_valid)) == comp_valid
def test_comp_invalid():
complement(comp_invalid)
