def analysis(self, chrom, start, stop, win_len=200, sift_len=20):
global pos_gc, win_gc
chrom = str(chrom)
tmp_start = start = int(start)
stop = int(stop)
flank_stop = min(
int(stop) + win_len + 1,
self.reference.get_reference_length(chrom))
try:
rmtk = list(self.rmtk.fetch(chrom, start, stop))
bases = self.reference.fetch(chrom, start, flank_stop)
except Exception as err:
raise ValueError(err)
self.contigs.add(chrom)
for pos in xrange(start, stop + 1):
flank_b = max(pos - win_len / 2, 0)
flank_e = min(flank_b + win_len + 1,
self.reference.get_reference_length(chrom))
base_gc = count_gc(
self.reference.fetch(chrom, flank_b, flank_e).upper())
pos_gc[chrom][pos] = base_gc
feback = list()
if len(rmtk):
for lines in rmtk:
rows = lines.strip().split("\t")
begin = int(rows[1])
end = int(rows[2])
if begin > start:
feback.append([start, begin - 1])
else:
begin = start
end = min(stop, end)
if begin > end:
continue
start = end + 1
else:
feback.append([start, stop])
for s, e in list(join_ranges(feback, offset=60)):
if e - s < sift_len:
continue
for win in xrange(s, e, sift_len):
offset = win - tmp_start
seq = bases[offset:offset + win_len].upper()
gc_radio = count_gc(seq)
if 0.1 < gc_radio < 0.9:
win_gc[(chrom, win, win + win_len - 1)] = gc_radio
else:
continue
def analysis(self, chrom, start, stop, win_len=200, sift_len=20):
global pos_gc, win_gc
chrom = str(chrom)
tmp_start = start = int(start)
stop = int(stop)
flank_stop = min(int(stop) + win_len + 1, self.reference.get_reference_length(chrom))
try:
rmtk = list(self.rmtk.fetch(chrom, start, stop))
bases = self.reference.fetch(chrom, start, flank_stop)
except Exception as err:
raise ValueError(err)
self.contigs.add(chrom)
for pos in xrange(start, stop + 1):
flank_b = max(pos - win_len / 2, 0)
flank_e = min(flank_b + win_len + 1, self.reference.get_reference_length(chrom))
base_gc = count_gc(self.reference.fetch(chrom, flank_b, flank_e).upper())
pos_gc[chrom][pos] = base_gc
feback = list()
if len(rmtk):
for lines in rmtk:
rows = lines.strip().split("\t")
begin = int(rows[1])
end = int(rows[2])
if begin > start:
feback.append([start, begin - 1])
else:
begin = start
end = min(stop, end)
if begin > end:
continue
start = end + 1
else:
feback.append([start, stop])
for s, e in list(join_ranges(feback, offset=60)):
if e - s < sift_len:
continue
for win in xrange(s, e, sift_len):
offset = win - tmp_start
seq = bases[offset : offset + win_len].upper()
gc_radio = count_gc(seq)
if 0.1 < gc_radio < 0.9:
win_gc[(chrom, win, win + win_len - 1)] = gc_radio
else:
continue
def annotation(self, sample, debug=False):
output = os.path.join(self.outdir, sample, "%s.cnv.anno.tsv" % sample)
f_out = smart_open(output, 'w')
titles = ["#Chrom", "Ploid", "Start", "Stop", "length", "copyNumber", "Mtype", "meanP", "Z-score", "GCpr",
"MutationName"]
dbtitle = self.DBAnno.dbtitle.split("\t")
titles.extend(dbtitle)
f_out.write("\t".join(titles) + '\n')
for chrom in self.contigs:
cnvs = os.path.join(self.outdir, sample, "%s.cnv" % chrom)
if not os.path.exists(cnvs):
continue
f_in = smart_open(cnvs)
for line in f_in:
rows = line.strip().split("\t")
try:
chrom = str(rows[0])
start = int(rows[2])
stop = int(rows[3])
mtype = str(rows[6])
except ValueError:
continue
z_s = self.z_score(chrom, start, stop, sample)
gcr = count_gc(self.DBAnno.refer.fetch(chrom, start, stop))
if not (0.3 <= gcr <= 0.7):
continue
variation = dict([("Chrom", chrom), ("Start", start), ("Stop", stop), ("Mtype", mtype)])
m_name = set()
for hgvs in self.HGVS.annobed(chrom, start, stop):
trans = str(hgvs.Transcript)
gene = str(hgvs.geneSym)
chgvs = str(hgvs.cHgvs)
protein = str(hgvs.Protein)
exons = str(hgvs.ExonRegions)
mess = ":".join(filter(lambda x: x != ".", [trans, protein, gene, chgvs, exons]))
m_name.add(mess)
dbinfo = self.DBAnno.dbanno(variation)
anno_message = [str(i) for i in rows]
anno_message.append(str(z_s))
anno_message.append(str(gcr))
anno_message.append("|".join(m_name))
for i in dbtitle:
if i in dbinfo:
anno_message.append("|".join(dbinfo[i]))
else:
anno_message.append(".")
f_out.write("\t".join(anno_message) + '\n')
f_in.close()
if not debug:
os.remove(cnvs)
f_out.close()
return output