def write_output(reader, outf, logger=None):
logger = logger or NullLogger()
seq_count = 0
name_serializer = SeqNameSerializer()
for r in reader:
fastq_records = build_fastq_records(r['label'],
r['mask'],
name_serializer,
logger=logger)
seq_count += len(fastq_records)
for r in fastq_records:
outf.write("%s\n" % "\n".join(r))
return seq_count
def get_extracted_seqs(fn):
with open(fn) as f:
reader = csv.reader(f, delimiter="\t")
data = {}
serializer = SeqNameSerializer()
for r in reader:
try:
label, _, _, alleles = serializer.deserialize(r[3])
seq = r[-1].upper()
except IndexError:
raise ValueError("%r: bad input format" % fn)
data[label] = (seq, alleles)
return data
def __init__(self,
ref_tag,
outf,
outfmt=DEFAULT_OUTPUT_FORMAT,
flank_size=DEFAULT_FLANK_SIZE,
logger=None):
self.logger = logger or NullLogger()
self.ref_tag = ref_tag
self.outf = outf
self.outfmt = outfmt
self.flank_size = flank_size
self.current_id = None
self.current_hits = []
self.serializer = SeqNameSerializer()
def write_output(logger, args):
serializer = SeqNameSerializer()
index = None
fields = MARKER_DEF_FIELDS + ("status", "extended_mask")
try:
index = shelve.open(args.index_file, "r")
logger.info("getting extracted sequences")
extracted_seqs = get_extracted_seqs(args.input_file)
if args.align_file:
logger.info("getting sorting order from %r" % (args.align_file))
idx_map = get_sort_idx(args.align_file)
max_idx = max(idx_map.itervalues())
fields += ("marker_indx", )
with nested(open(args.orig_file), open(args.output_file,
'w')) as (f, outf):
outf.write("\t".join(fields) + "\n")
reader = csv.DictReader(f, delimiter="\t")
logger.info("looking up against %r" % args.index_file)
for i, r in enumerate(reader):
label = r['label']
old_rs_label = r['rs_label']
mask = r['mask']
try:
seq, alleles = extracted_seqs[label]
except KeyError:
rs_label = extended_mask = 'None'
status = Status.NO_INFO
else:
extended_mask = build_mask(seq, alleles)
key = build_index_key(seq)
tags = index.get(key, [])
n_matches = len(tags)
if n_matches != 1:
logger.warning("%r maps to %d tags: %r" %
(label, n_matches, tags))
rs_label = 'None'
status = Status.NO_MATCH if n_matches == 0 else Status.MULTI_MATCH
else:
rs_label, _, _, _ = serializer.deserialize(tags[0])
if old_rs_label == "None":
status = Status.ADDED
else:
status = (Status.CONFIRMED if rs_label
== old_rs_label else Status.REPLACED)
if rs_label == 'None':
rs_label = label
out_r = [
label, rs_label, mask, r['allele_a'], r['allele_b'],
status, extended_mask
]
if args.align_file:
try:
idx = idx_map[label]
except KeyError:
max_idx += 1
idx = max_idx
out_r.append(str(idx))
outf.write("%s\n" % "\t".join(out_r))
logger.info("processed %d records overall" % (i + 1))
finally:
if index:
index.close()
