def get_hwe_likelihood(obs_hom_ref, obs_het, obs_hom_alt, aaf):
"""
Compute the likelihood of deviation from HWE using X^2,
as well as the inbreeding coefficient.
"""
# Bail out if aaf is undefined. This occurs
# when there are multiple alternate alleles
if aaf is None:
return (None, None)
# how many total genotypes?
sum = (float(obs_hom_ref) + float(obs_het) + float(obs_hom_alt))
# get the reference allele freq
raf = 1.0 - float(aaf)
#compute the expected number of each genotype based on p and q
exp_hom_ref = (raf**2)*sum
exp_het = (2.0*(raf*aaf))*sum
exp_hom_alt = (aaf**2)*sum
# get the X^2 statistcs for each genotype class.
x2_hom_ref = ((obs_hom_ref - exp_hom_ref)**2)/exp_hom_ref if exp_hom_ref > 0 else 0
x2_hom_alt = ((obs_hom_alt - exp_hom_alt)**2)/exp_hom_alt if exp_hom_alt > 0 else 0
x2_het = ((obs_het - exp_het)**2)/exp_het if exp_het > 0 else 0
x2_statistic = x2_hom_ref + x2_hom_alt + x2_het
# return the p-value (null hyp. is that the genotypes are in HWE)
# 1 degree of freedom b/c 3 genotypes, 2 alleles (3-2)
# estimate the inbreeding coefficient (F_hat):
# F_hat = 1 - O_hets / E_hets
inbreeding_coeff = (1.0 - (float(obs_het)/(float(exp_het)))) if obs_het > 0 else None
return stats.lchisqprob(x2_statistic, 1), inbreeding_coeff
def chi_squared_p_value(self):
df = self.number_of_bins() - 1
chisq = float(self.sum_chi_squared())
chi_squared_p_value = stats.lchisqprob(chisq, df)
return chi_squared_p_value * 100
def chi_squared_p_value(self): df = self.number_of_bins()-1 chisq = float(self.sum_chi_squared()) chi_squared_p_value = stats.lchisqprob(chisq,df) return chi_squared_p_value*100
