def get_beta_diversity(self):
# FIXME: you can't call this unless you call microbiome_sequence_alignment() first
sequence_set = list(set(self.alignment))
self.beta_diversity = 0
number_sequence = len(sequence_set)
if number_sequence > 1:
for i in range(1, number_sequence):
fre_i = self.alignment.count(sequence_set[i]) / float(self.number_of_host_in_population)
for j in range(i):
fre_j = self.alignment.count(sequence_set[j]) / float(self.number_of_host_in_population)
self.beta_diversity += fre_i * fre_j * different_element(
sequence_set[i], sequence_set[j]) / self.number_of_environmental_species
# NOTE: Can factor out fre_i, it's faster but less clear
# temp_div += fre_j * different_element(
# sequence_set[i], sequence_set[j]) / self.number_of_environmental_species
# self.beta_diversity += (fre_i * temp_div)
# self.beta_diversity = self.beta_diversity * self.number_of_host_in_population / (self.number_of_host_in_population - 1) * 2
self.beta_diversity *= self.beta_diversity_coef
# return self.beta_diversity
# else:
return self.beta_diversity
def watterson_tajima(self): # http://en.wikipedia.org/wiki/Tajima%27s_D
a_1 = 0
a_2 = 0
for i in range(1, self.number_of_host_in_population):
a_1 += 1 / float(i)
a_2 += 1 / float(i ** 2)
self.theta = self.number_of_segregating_site / a_1
K = 0
N = 0
b_1 = (self.number_of_host_in_population + 1) / float(3 * self.number_of_host_in_population - 3)
b_2 = float(2) / 9 * (self.number_of_host_in_population ** 2 + self.number_of_host_in_population + 3) / (self.number_of_host_in_population ** 2 - self.number_of_host_in_population)
c_1 = b_1 - 1 / a_1
c_2 = b_2 - (self.number_of_host_in_population + 2) / (a_1 * self.number_of_host_in_population) + a_2 / (a_1 ** 2)
e_1 = c_1 / a_1
e_2 = c_2 / (a_1 ** 2 + a_2)
for i in range(1, self.number_of_host_in_population):
for j in range(i):
K += different_element(self.alignment[i], self.alignment[j])
N += 1
k = K / float(N)
if self.theta != 0:
self.D = (k - self.theta) / (e_1 * self.number_of_segregating_site + e_2 * self.number_of_segregating_site * (self.number_of_segregating_site - 1)) ** 0.5
else:
self.D = None
