def calc_snv_read_pos_stats(sam, snvs, max_snps=None, max_pos=None):
"This implementation is using pysam pysam"
pileup_cols = sam.pileup()
read_5_pos_cnts_rg = {}
read_3_pos_cnts_rg = {}
read_5_pos_box_rg = {}
read_3_pos_box_rg = {}
for index, snv in enumerate(snvs):
if max_snps and index >= max_snps:
break
chrom = snv.chrom
ref_pos = snv.pos
snv_qual = snv.qual
snv_col = None
for col in pileup_cols:
ref_name = sam.getrname(col.reference_id)
if ref_name == chrom and col.reference_pos == ref_pos:
snv_col = col
break
if snv_col is None:
raise RuntimeError('No pileup found for snv {}:{}'.format(
chrom, ref_pos))
for pileup_read in snv_col.pileups:
try:
read_group = pileup_read.alignment.opt('RG')
except KeyError:
read_group = None
read_ref_coord = ReadRefCoord(pileup_read.alignment, sam)
read_pos = read_ref_coord.get_read_pos((chrom, ref_pos))
read_pos_end = read_ref_coord.get_read_pos_counting_from_end(
(chrom, ref_pos))
if read_group not in read_5_pos_cnts_rg:
read_5_pos_cnts_rg[read_group] = IntCounter()
read_3_pos_cnts_rg[read_group] = IntCounter()
read_5_pos_box_rg[read_group] = IntBoxplot()
read_3_pos_box_rg[read_group] = IntBoxplot()
read_5_pos_cnts = read_5_pos_cnts_rg[read_group]
read_3_pos_cnts = read_3_pos_cnts_rg[read_group]
read_5_pos_box = read_5_pos_box_rg[read_group]
read_3_pos_box = read_3_pos_box_rg[read_group]
if (read_pos is not None
and (not max_pos or read_pos + 1 <= max_pos)):
read_5_pos_cnts[read_pos + 1] += 1
read_5_pos_box.append(read_pos + 1, snv_qual)
if (read_pos_end is not None
and (not max_pos or abs(read_pos_end) <= max_pos)):
read_3_pos_cnts[abs(read_pos_end)] += 1
read_3_pos_box.append(abs(read_pos_end), snv_qual)
return {
'5_read_pos_counts': read_5_pos_cnts_rg,
'3_read_pos_counts': read_3_pos_cnts_rg,
'5_read_pos_boxplot': read_5_pos_box_rg,
'3_read_pos_boxplot': read_3_pos_box_rg
}
def __init__(self,
vcf_fpath,
gq_threshold=None,
dp_threshold=100,
min_calls_for_pop_stats=DEF_MIN_CALLS_FOR_POP_STATS,
remarkable_coverages=None,
window_size=WINDOWS_SIZE):
if remarkable_coverages is None:
remarkable_depths = REMARKABLE_DEPTHS
self.remarkable_depths = remarkable_depths
self._reader = VCFReader(
open(vcf_fpath), min_calls_for_pop_stats=min_calls_for_pop_stats)
self._random_reader = pyvcfReader(filename=vcf_fpath)
self.window_size = window_size
self._gq_threshold = 0 if gq_threshold is None else gq_threshold
self.dp_threshold = dp_threshold
self._gt_qual_depth_counter = {HOM: IntBoxplot(), HET: IntBoxplot()}
self._ac2d = _AlleleCounts2D()
self.sample_dp_coincidence = {1: IntCounter()}
for cov in remarkable_depths:
self.sample_dp_coincidence[cov] = IntCounter()
self.called_snvs = 0
self.called_gts = IntCounter()
# sample_counter
self._sample_counters = {}
for counter_name in SAMPLE_COUNTERS:
if counter_name not in self._sample_counters:
self._sample_counters[counter_name] = {}
for sample in self._reader.samples:
if counter_name in (GT_DEPTHS, GT_QUALS):
counters = {HOM: IntCounter(), HET: IntCounter()}
else:
counters = IntCounter()
self._sample_counters[counter_name][sample] = counters
self._snv_counters = {
MAFS: IntCounter(),
MACS: IntCounter(),
MAFS_DP: IntCounter(),
SNV_QUALS: IntCounter(),
HET_IN_SNP: IntCounter(),
SNV_DENSITY: IntCounter(),
INBREED_F_IN_SNP: IntCounter(),
DEPTHS: IntCounter()
}
self._calculate()
def test_int_boxplot(self):
box = IntBoxplot()
box.append(1, 50)
box.append(1, 40)
box.append(1, 30)
box.append(1, 40)
box.append(2, 30)
box.append(2, 10)
box.append(2, 20)
box.append(2, 40)
fhand = NamedTemporaryFile(suffix='.png')
draw_int_boxplot(box, fhand=fhand)
def calc_snv_read_pos_stats2(sam, snvs, max_snps=None, max_pos=None):
"this implementation is using pysam fetch"
read_5_pos_cnts_rg = {}
read_3_pos_cnts_rg = {}
read_5_pos_box_rg = {}
read_3_pos_box_rg = {}
for index, snv in enumerate(snvs):
if max_snps and index >= max_snps:
break
chrom = snv.chrom
ref_pos = snv.pos
snv_qual = snv.qual
for alignment_read in sam.fetch(chrom, ref_pos, ref_pos + 1):
try:
read_group = alignment_read.opt('RG')
except KeyError:
read_group = None
read_ref_coord = ReadRefCoord(alignment_read, sam)
read_pos = read_ref_coord.get_read_pos((chrom, ref_pos))
read_pos_end = read_ref_coord.get_read_pos_counting_from_end(
(chrom, ref_pos))
if read_group not in read_5_pos_cnts_rg:
read_5_pos_cnts_rg[read_group] = IntCounter()
read_3_pos_cnts_rg[read_group] = IntCounter()
read_5_pos_box_rg[read_group] = IntBoxplot()
read_3_pos_box_rg[read_group] = IntBoxplot()
read_5_pos_cnts = read_5_pos_cnts_rg[read_group]
read_3_pos_cnts = read_3_pos_cnts_rg[read_group]
read_5_pos_box = read_5_pos_box_rg[read_group]
read_3_pos_box = read_3_pos_box_rg[read_group]
if (read_pos is not None
and (not max_pos or read_pos + 1 <= max_pos)):
read_5_pos_cnts[read_pos + 1] += 1
read_5_pos_box.append(read_pos + 1, snv_qual)
if (read_pos_end is not None
and (not max_pos or abs(read_pos_end) <= max_pos)):
read_3_pos_cnts[abs(read_pos_end)] += 1
read_3_pos_box.append(abs(read_pos_end), snv_qual)
return {
'5_read_pos_counts': read_5_pos_cnts_rg,
'3_read_pos_counts': read_3_pos_cnts_rg,
'5_read_pos_boxplot': read_5_pos_box_rg,
'3_read_pos_boxplot': read_3_pos_box_rg
}
def test_int_boxplot(self):
box = IntBoxplot()
box.append(1, 50)
box.append(1, 40)
box.append(1, 30)
box.append(1, 40)
box.append(2, 30)
box.append(2, 10)
box.append(2, 20)
box.append(2, 40)
fhand = NamedTemporaryFile(suffix='.png')
draw_int_boxplot(box, fhand=fhand)
def test_boxplot(self):
'It does a bloxplot for integers'
box = IntBoxplot()
box.append(1, 50)
box.append(1, 40)
box.append(1, 30)
box.append(1, 40)
box.append(2, 30)
box.append(2, 10)
box.append(2, 20)
box.append(2, 40)
box.append('no distrib', 40)
counts = box.aggregated_array
assert sum(counts.values()) == 9
plot = box.ascii_plot
assert '2:10.0,15.0,25.0,35.0,40.0 <-----[============|=======' in plot
def test_boxplot(self):
'It does a bloxplot for integers'
box = IntBoxplot()
box.append(1, 50)
box.append(1, 40)
box.append(1, 30)
box.append(1, 40)
box.append(2, 30)
box.append(2, 10)
box.append(2, 20)
box.append(2, 40)
box.append('no distrib', 40)
counts = box.aggregated_array
assert sum(counts.values()) == 9
plot = box.ascii_plot
assert '2:10.0,15.0,25.0,35.0,40.0 <-----[============|=======' in plot
