def compute_probabilities(self, c, err, q1, q2, q, g, *_):
b_o = self.get_b_o(q1, q2, q)
threshold_o = self.get_hist_threshold(b_o, self.threshold)
# read to kmer coverage
c = self.correct_c(c)
# lambda for kmers with s errors
l_s = self._get_lambda_s(c, err)
pl_s = [(s * 0.5) for s in l_s]
# expected probability of kmers with s errors and coverage >= 1
# noinspection PyTypeChecker
n_os = [None] + [[
self.comb[s] * (1.0 - exp(o * -l_s[s]))
for s in range(self.max_error)
] for o in range(1, threshold_o)]
sum_n_os = [None] + [
fix_zero(sum(n_os[o][t] for t in range(self.max_error)))
for o in range(1, threshold_o)
]
pn_os = [None] + [[
self.comb[s] * (1.0 - exp(o * -pl_s[s]))
for s in range(self.max_error)
] for o in range(1, threshold_o)]
sum_pn_os = [None] + [
fix_zero(sum(pn_os[o][t] for t in range(self.max_error)))
for o in range(1, threshold_o)
]
# portion of kmers wit1h s errors
# noinspection PyTypeChecker
a_os = [None] + [[
n_os[o][s] / (sum_n_os[o] if sum_n_os[o] != 0 else 1)
for s in range(self.max_error)
] for o in range(1, threshold_o)]
pa_os = [None] + [[
pn_os[o][s] / (sum_pn_os[o] if sum_pn_os[o] != 0 else 1)
for s in range(self.max_error)
] for o in range(1, threshold_o)]
# probability that unique kmer has coverage j (j > 0)
p_j = {
j: (1 - g) * sum(
b_o(o) * sum(a_os[o][s] * tr_poisson(o * l_s[s], j)
for s in range(self.max_error))
for o in range(1, threshold_o)) + g * sum(
b_o(o) * sum(pa_os[o][s] * tr_poisson(o * pl_s[s], j)
for s in range(self.max_error))
for o in range(1, threshold_o))
for j in self.hist
}
return p_j
def compute_probabilities(self, c, err, q1, q2, q, *_):
b_o = self.get_b_o(q1, q2, q)
threshold_o = self.get_hist_threshold(b_o, self.threshold)
# read to kmer coverage
c = self.correct_c(c)
# lambda for kmers with s errors
l_s = self._get_lambda_s(c, err)
# expected probability of kmers with s errors and coverage >= 1
# noinspection PyTypeChecker
n_os = [None] + [
[self.comb[s] * (1.0 - exp(o * -l_s[s])) for s in range(self.max_error)]
for o in range(1, threshold_o)
]
sum_n_os = [None] + [
fix_zero(sum(n_os[o][t] for t in range(self.max_error))) for o in range(1, threshold_o)
]
# portion of kmers wit1h s errors
# noinspection PyTypeChecker
a_os = [None] + [
[n_os[o][s] / (sum_n_os[o] if sum_n_os[o] != 0 else 1) for s in range(self.max_error)]
for o in range(1, threshold_o)
]
# probability that unique kmer has coverage j (j > 0)
p_j = {
j: sum(
b_o(o) * sum(
a_os[o][s] * tr_poisson(o * l_s[s], j) for s in range(self.max_error)
) for o in range(1, threshold_o)
) for j in self.hist
}
return p_j
def compute_polymorphism_rate(self, c, err, g, genome_size):
ck = self.correct_c(c) / 2.0
# lambda for kmers with s errors --- lambda pre k-mery s 's' chybami
l_s = self._get_lambda_s(ck, err)
# expected probability of kmers with s errors and coverage >= 1 --- ocakavana pravdepodobnost kmerov s 's' chybami a pokrytim aspon 1
n_s = [
self.comb[s] * (1.0 - exp(-l_s[s])) for s in range(self.max_error)
]
sum_n_s = fix_zero(sum(
n_s[t]
for t in range(self.max_error))) # suma tychto pravdepodobnosti
# portion of kmers with s errors --- pomer kmerov s 's' chybami
a_s = [n_s[s] / sum_n_s for s in range(self.max_error)]
# probability that unique kmer has coverage j (j > 0) --- pravdepodobnost, ze dany kmer ma pokrytie j (j > 0)
p_j = {
j:
sum(a_s[s] * tr_poisson(l_s[s], j) for s in range(self.max_error))
for j in range(1,
max(self.orig_hist.keys()) // 2)
}
x_pd = g * self.total_distinct_kmers
sum_pp = sum([j * p_j[j] for j in p_j])
p_kmers = (x_pd * sum_pp) / (2 * genome_size * self.correct_c(c) / 2.0)
p = 1 - (1 - p_kmers)**(1.0 / self.k)
return p
def compute_probabilities(self, c, err, g, *_):
# read to kmer coverage
ck = self.correct_c(c)
# lambda for non-polymorphic kmers with s errors
l_s = self._get_lambda_s(ck, err)
# lambda for polymorfic k-mers
pl_s = [(s * 0.5) for s in l_s]
# expected probability of non-polymorphic kmers with s errors and coverage >= 1
n_s = [
self.comb[s] * (1.0 - exp(-l_s[s])) for s in range(self.max_error)
]
sum_n_s = fix_zero(sum(n_s[t] for t in range(self.max_error)))
# expected probability of polymorphic kmers with s errors and coverage >= 1
pn_s = [
self.comb[s] * (1.0 - exp(-pl_s[s])) for s in range(self.max_error)
]
sum_pn_s = fix_zero(sum(
pn_s[t]
for t in range(self.max_error))) # suma tychto pravdepodobnosti
# portion of non-polymorphic kmers with s errors
a_s = [n_s[s] / sum_n_s for s in range(self.max_error)]
# portion of non-polymorphic kmers with s errors
pa_s = [pn_s[s] / sum_pn_s for s in range(self.max_error)]
# probability that unique kmer has coverage j (j > 0)
p_j = {
j: (1 - g) * sum(a_s[s] * tr_poisson(l_s[s], j)
for s in range(self.max_error)) +
g * sum(pa_s[s] * tr_poisson(pl_s[s], j)
for s in range(self.max_error))
for j in self.hist
}
return p_j
def compute_probabilities(self, c, err, *_):
# read to kmer coverage
ck = self.correct_c(c)
# lambda for kmers with s errors
l_s = self._get_lambda_s(ck, err)
# expected probability of kmers with s errors and coverage >= 1
n_s = [self.comb[s] * (1.0 - exp(-l_s[s])) for s in range(self.max_error)]
sum_n_s = fix_zero(sum(n_s[t] for t in range(self.max_error)))
# portion of kmers with s errors
a_s = [n_s[s] / sum_n_s for s in range(self.max_error)]
# probability that unique kmer has coverage j (j > 0)
p_j = {
j: sum(
a_s[s] * tr_poisson(l_s[s], j) for s in range(self.max_error)
)
for j in self.hist
}
return p_j
def compute_probabilities(self, c, err, *_):
# read to kmer coverage
ck = self.correct_c(c)
# lambda for kmers with s errors
l_s = self._get_lambda_s(ck, err)
# expected probability of kmers with s errors and coverage >= 1
n_s = [
self.comb[s] * (1.0 - exp(-l_s[s])) for s in range(self.max_error)
]
sum_n_s = fix_zero(sum(n_s[t] for t in range(self.max_error)))
# portion of kmers with s errors
a_s = [n_s[s] / sum_n_s for s in range(self.max_error)]
# probability that unique kmer has coverage j (j > 0)
p_j = {
j:
sum(a_s[s] * tr_poisson(l_s[s], j) for s in range(self.max_error))
for j in self.hist
}
return p_j