def annotate_single_read(args, read, juncs, ex_ref, ref_trees, outbam=None, genes=''):
'''
Annotate insertions, deletions and soft-clips on a single read
'''
ci_file = args.contig_info_output
genes = get_overlapping_genes(read, ref_trees) if genes == '' else genes
fusion = any([op == constants.CIGAR['hard-clip'] and val >= MIN_CLIP for op, val in read.cigar])
if genes == '' and not fusion:
logging.info('No gene(s) intersecting read %s; skipping' % read.query_name)
return
# check for contig gaps or soft-clips
has_gaps = any([op in constants.GAPS and val >= MIN_GAP for op, val in read.cigar])
has_scs = any([op == constants.CIGAR['soft-clip'] and val >= MIN_CLIP for op, val in read.cigar])
is_spliced = any([op == constants.CIGAR['skipped'] for op, val in read.cigar])
# check junctions
tx_juncs = get_tx_juncs(read)
unknown_juncs = ['%s:%s-%s' % (c, s, e) not in juncs[0] for c, s, e in tx_juncs]
has_novel_juncs = any(unknown_juncs)
# check for novel blocks
chr_ref = get_chr_ref(read, ex_ref)
has_novel_blocks = any([bh.is_novel_block(block, chr_ref, MIN_CLIP) for block in read.get_blocks()])
if has_gaps or has_scs or has_novel_juncs or has_novel_blocks:
cv = CrypticVariant().from_read(read)
cv.genes = genes
if has_gaps:
annotate_gaps(cv, read, ci_file)
if has_scs:
annotate_softclips(cv, read, ci_file)
if has_novel_juncs:
novel_juncs = [list(x) for x in np.array(tx_juncs)[unknown_juncs]]
annotate_juncs(cv, read, juncs[1], novel_juncs, ci_file)
if has_novel_blocks:
annotate_blocks(cv, read, chr_ref, ci_file)
if outbam:
outbam.write(read)
else:
logging.info('Nothing to annotate for read %s (read matches reference)' % read.query_name)
def annotate_blocks(cv, read, chr_ref, ci_file):
'''
Annotate any sequence that is outside of exonic regions
'''
cv.parid = '.' # blocks don't have pairs
novel_blocks = [(idx, block) for idx, block in cv.blocks
if bh.is_novel_block(block, chr_ref, MIN_CLIP)]
for block_idx, block in novel_blocks:
cpos1 = sum([
v for c, v in read.cigar[:block_idx]
if c in constants.AFFECT_CONTIG
])
cpos2 = sum([
v for c, v in read.cigar[:block_idx + 1]
if c in constants.AFFECT_CONTIG
])
# whether sequence block is overlapping, or on left or right side of contig block
olapping = chr_ref[np.logical_and(block[0] < chr_ref.start,
block[1] > chr_ref.end)]
left = chr_ref[np.logical_and(block[1] > chr_ref.start,
block[1] <= chr_ref.end)]
right = chr_ref[np.logical_and(block[0] >= chr_ref.start,
block[0] < chr_ref.end)]
if len(left) > 0 and len(right) > 0:
# retained intron
cv.cvtype = 'RI'
qseq, rseq = bh.get_block_sequence(read, block_idx)
seq_right_pos = block[1] - min(left.start)
seq_left_pos = max(right.end) - block[0]
cv.pos = block[0] + seq_left_pos + 1
cv.ref = rseq[seq_left_pos:(-seq_right_pos)]
cv.alt = ']' + qseq[seq_left_pos:(-seq_right_pos)] + '['
cv.cpos = cpos1 + seq_left_pos
cv.vsize, cv.cvsize = abs(len(cv.alt) - 2 -
len(cv.ref)), len(cv.alt) - 2
cv.vid = get_next_id(read.query_name)
elif len(olapping) > 0:
# annotate left side
cv.cvtype = 'EE'
cv = annotate_block_left(cv, read, cpos2, olapping, block,
block_idx)
cv.vid = get_next_id(read.query_name)
print(cv.vcf_output())
CrypticVariant.write_contig_info(ci_file, cv)
# annotate right side
cv = annotate_block_right(cv, read, cpos1, olapping, block,
block_idx)
cv.vid = get_next_id(read.query_name)
elif len(left) > 0:
# annotate left side
cv.cvtype = 'EE'
cv = annotate_block_left(cv, read, cpos2, left, block, block_idx)
cv.vid = get_next_id(read.query_name)
elif len(right) > 0:
# annotate right side
cv.cvtype = 'EE'
cv = annotate_block_right(cv, read, cpos1, right, block, block_idx)
cv.vid = get_next_id(read.query_name)
else:
# block does not cross any annotation
qseq, rseq = bh.get_block_sequence(read, block_idx)
cv.ref, cv.alt = rseq, '[' + qseq + ']'
cv.pos, cv.cvtype = block[0] + 1, 'NE'
cv.cpos = cpos1
cv.vid = get_next_id(read.query_name)
cv.vsize, cv.cvsize = abs(len(cv.alt) - 2 -
len(cv.ref)), len(cv.alt) - 2
print(cv.vcf_output())
CrypticVariant.write_contig_info(ci_file, cv)
def test_is_novel_block(coord, expected):
assert bh.is_novel_block(coord, chr_ref, MIN_CLIP) == expected