def test_generate_windows(self):
assert self.generate_wins(10, 10, 3) == [(0, 10), (10, 20), (20, 30)]
assert self.generate_wins(10, None, 3) == [(0, 10), (10, 20), (20, 30)]
assert self.generate_wins(10, 5, 3) == [(0, 10), (5, 15), (10, 20)]
res = [(0, 10), (10, 20), (20, 30)]
assert list(generate_windows(size=10, step=10, end=30)) == res
def windows(self):
chrom_lengths = self._get_chrom_lengths()
snp_queue = self._snp_queue
for chrom, chrom_length in chrom_lengths.items():
wins = generate_windows(start=0,
size=self.win_size,
step=self.win_step,
end=chrom_length + 1)
snp_queue.empty()
for win in wins:
snp_queue.pop(win.start)
if snp_queue.queue:
new_strech_start = snp_queue.queue[-1].pos + 1
else:
new_strech_start = win.start
new_snps = self._reader.fetch_snvs(chrom, new_strech_start,
win.end)
snp_queue.extend(new_snps)
if len(snp_queue.queue) >= self.min_num_snps:
yield {
'chrom': chrom,
'start': win.start,
'end': win.end,
'snps': snp_queue.queue[:]
}
def test_generate_windows(self):
assert self.generate_wins(10, 10, 3) == [(0, 10), (10, 20), (20, 30)]
assert self.generate_wins(10, None, 3) == [(0, 10), (10, 20), (20, 30)]
assert self.generate_wins(10, 5, 3) == [(0, 10), (5, 15), (10, 20)]
res = [(0, 10), (10, 20), (20, 30)]
assert list(generate_windows(size=10, step=10, end=30)) == res
def calculate_coverage_distrib_in_region(self, region=None):
if region is None:
if self.window == 1:
regions = None
else:
regions = [(ref, 0, le_ - 1)
for ref, le_ in self._ref_lens.items()]
else:
regions = [region]
if self.window == 1:
if regions is None:
region = None
else:
region = regions[0]
return self._calculate_complete_coverage_distrib(region)
counts = {}
for region in regions:
chrom, start, end = region
for start, end in generate_windows(self.window,
start=0,
end=self._ref_lens[chrom],
step=1):
counts_in_win = self._calculate_coverages_in_win(
chrom, start, end)
for sample, cnts_in_win in counts_in_win.items():
if sample not in counts:
counts[sample] = IntCounter()
counts[sample][int(round(cnts_in_win))] += 1
return counts
def calculate_coverage_distrib_in_region(self, region=None):
if region is None:
if self.window == 1:
regions = None
else:
regions = [(ref, 0, le_ - 1) for ref, le_ in self._ref_lens.items()]
else:
regions = [region]
if self.window == 1:
if regions is None:
region = None
else:
region = regions[0]
return self._calculate_complete_coverage_distrib(region)
counts = {}
for region in regions:
chrom, start, end = region
for start, end in generate_windows(self.window, start=0,
end=self._ref_lens[chrom],
step=1):
counts_in_win = self._calculate_coverages_in_win(chrom, start,
end)
for sample, cnts_in_win in counts_in_win.items():
if sample not in counts:
counts[sample] = IntCounter()
counts[sample][int(round(cnts_in_win))] += 1
return counts
def generate_wins(self, size, step, number):
windows = generate_windows(size=size, step=step)
wins = []
for index, win in enumerate(windows):
if index >= number:
break
wins.append(win)
return wins
def generate_wins(self, size, step, number):
windows = generate_windows(size=size, step=step)
wins = []
for index, win in enumerate(windows):
if index >= number:
break
wins.append(win)
return wins
def windows(self):
chrom_lengths = self._get_chrom_lengths()
snp_queue = self._snp_queue
for chrom, chrom_length in chrom_lengths.items():
wins = generate_windows(start=0,
size=self.win_size, step=self.win_step,
end=chrom_length + 1)
snp_queue.empty()
for win in wins:
snp_queue.pop(win.start)
if snp_queue.queue:
new_strech_start = snp_queue.queue[-1].pos + 1
else:
new_strech_start = win.start
new_snps = self._reader.fetch_snvs(chrom, new_strech_start,
win.end)
snp_queue.extend(new_snps)
if len(snp_queue.queue) >= self.min_num_snps:
yield {'chrom': chrom, 'start': win.start, 'end': win.end,
'snps': snp_queue.queue[:]}
