def count_variants(self, pos2var):
if self.args.v:
print "processing %s..." % self.rnaseq_fn
dots = ProgressDots(self.args.dot_counter)
with open(self.rnaseq_fn) as rf:
reader = csv.reader(rf, delimiter="\t")
for line in reader:
if line[0].startswith("@"):
continue
chrom = line[2]
pos = int(line[3])
seq = line[9]
# print 'alignment: %s %d-%d (%d) %s' % (chrom, pos, pos+len(seq), len(seq), seq)
self.stats["n_alignments"] += 1
if self.args.progress:
dots.ping()
# crawl the aligned read, looking for a variant:
# This is O(m*n) on the number and length of the reads,
# but it's O(1) for programmer laziness.
variant = pos2var.variant_for(chrom, pos, len(seq))
if not variant:
continue
variant.n_alignments += 1
self.stats["n_variant_hits"] += 1
if variant.is_expressed_in_seq(seq, pos):
variant.n_mut += 1
else:
variant.n_wt += 1
def main(args):
if args.v: print args
print 'reading %s...' % args.variant_fn
pos2var=VariantPositions(args.variant_fn)
try: os.mkdir(args.output_dir)
except OSError: pass
for f in os.listdir(args.output_dir):
os.unlink(os.path.join(args.output_dir, f))
dots=ProgressDots(args.dot_counter)
stats={'n_reads':0,
'n_variants':0}
print 'reading %s...' % args.rnaseq_fn
with open(args.rnaseq_fn) as f:
for line in f:
dots.ping()
if line.startswith('@'):
continue
stats['n_reads']+=1
row=line.split('\t')
var=pos2var.variant_for(row[2], int(row[3]), len(row[9]))
if not var: continue
stats['n_variants']+=1
try:
var.reads.append(line)
except AttributeError:
var.reads=[line]
print
# write out all reads for each variant:
for var in pos2var.values():
try:
fn=os.path.join(args.output_dir, '%s.fastq' % var.symbol)
with open(fn, 'w') as var_f:
for line in var.reads:
var_f.write(line)
except AttributeError: # on var.reads
pass
print stats
return 0