def filter_spanning_reads(chimeras, reads,
anchor_min,
anchor_length,
anchor_mismatches,
library_type):
for i,r in enumerate(reads):
if r.is_unmapped:
continue
# make a discordant read object
# TODO: need to annotate reads elsewhere since they have already been sorted here
r.tags = r.tags + [("HI", 0),
("IH", 1),
("NH", 1),
(DISCORDANT_TAG_NAME, DiscordantTags.DISCORDANT_GENE),
(ORIENTATION_TAG_NAME, get_gene_orientation(r, library_type))]
dr = DiscordantRead.from_read(r)
dr.is_spanning = True
# check read alignment against chimeras
for c in chimeras:
if check_breakpoint_alignment(c, r,
anchor_min,
anchor_length,
anchor_mismatches):
# valid spanning read
yield c,dr
def extract_single_mapped_reads(
chimera_file, unmapped_bam_file, single_mapped_bam_file, unmapped_fastq_file, library_type, tmp_dir
):
# find all reads that need to be remapped to see if they span the
# breakpoint junction
fqfh = open(unmapped_fastq_file, "w")
# annotate mapped reads with sequence/quality of unmapped mate
bamfh = pysam.Samfile(unmapped_bam_file, "rb")
unsorted_single_mapped_bam_file = os.path.join(tmp_dir, "unsorted_single_mapped_reads.bam")
singlemap_bamfh = pysam.Samfile(unsorted_single_mapped_bam_file, "wb", template=bamfh)
# get list of 'gene' references in bam file to compare with
gene_tids = set([tid for tid, refname in enumerate(bamfh.references) if refname.startswith(config.GENE_REF_PREFIX)])
for pe_reads in parse_pe_reads(bamfh):
# find which of the original reads was unmapped
r1_unmapped = any(r.is_unmapped for r in pe_reads[0])
r2_unmapped = any(r.is_unmapped for r in pe_reads[1])
# if both reads unmapped, then remap to breakpoints
if r1_unmapped and r2_unmapped:
for readnum in (0, 1):
print >> fqfh, to_fastq(
pe_reads[readnum][0].qname, readnum, pe_reads[readnum][0].seq, pe_reads[readnum][0].qual
)
else:
# annotate the mapped reads with the seq/qual of the
# unmapped reads
mapped_readnum = 0 if r2_unmapped else 1
unmapped_readnum = 1 if r2_unmapped else 0
unmapped_seq = pe_reads[unmapped_readnum][0].seq
unmapped_qual = pe_reads[unmapped_readnum][0].qual
for r in pe_reads[mapped_readnum]:
# only consider gene mappings
if r.rname not in gene_tids:
continue
orientation = get_gene_orientation(r, library_type)
# TODO: may need to REVERSE read here to get original
r.tags = r.tags + [("R2", unmapped_seq), ("Q2", unmapped_qual), (ORIENTATION_TAG_NAME, orientation)]
singlemap_bamfh.write(r)
singlemap_bamfh.close()
fqfh.close()
# sort/index the annotated single-mapper unmapped reads by reference/position
logging.debug("Sorting single-mapped mates by reference")
single_mapped_bam_prefix = os.path.splitext(single_mapped_bam_file)[0]
pysam.sort("-m", str(int(1e9)), unsorted_single_mapped_bam_file, single_mapped_bam_prefix)
pysam.index(single_mapped_bam_file)
# remove unsorted file
if os.path.exists(unsorted_single_mapped_bam_file):
os.remove(unsorted_single_mapped_bam_file)
return config.JOB_SUCCESS
