def create_het_distribution(
seqs, distrib_fhand=None, plot_fhand=None, summary_fhand=None, group_kind=None, groups=None, ploidy=2
):
"""It creates the distribution of the heterozygosity
(not takes in account ref allele)"""
title = "heterozygosity"
if groups and group_kind:
title = "heterozygosity (%s: %s)" % (group_kind, ",".join(groups))
hets = CachedArray("f")
for seq in seqs:
for snv in seq.get_features("snv"):
if not group_kind and "heterozygosity" in snv.qualifiers:
het = snv.qualifiers["heterozygosity"]
else:
het = calculate_heterozygosity(snv, ploidy, group_kind=group_kind, groups=groups)
if het is not None:
hets.append(het)
if list(hets):
create_distribution(
hets,
labels={"title": title},
distrib_fhand=distrib_fhand,
bins=None,
plot_fhand=plot_fhand,
range_=None,
summary_fhand=summary_fhand,
calculate_freqs=False,
remove_outliers=False,
)
def calculate_hets_group(seqs, groups=None, group_kind=None, ploidy=2):
'It calculates the snv heterozygosity of a given group'
het_profile = {}
for seq in seqs:
for snv in seq.get_features('snv'):
het = calculate_heterozygosity(snv, ploidy,
group_kind=group_kind,
groups=groups)
if het is not None:
location = snv.location.start.position
seq_name = seq.name
if seq_name not in het_profile:
het_profile[seq_name] = []
het_profile[seq_name].append((location, het))
return het_profile
