def unifrac_mix(As, otusid, t):
#unifracc a partir de una matriz sparse pero luego utilizando slice de columna densos al comparar muestras
As.convert('csc')
m, n = As.shape
dist_vec, pos = readtree(t, otusid)
dist = numpy.zeros([n, n])
x = numpy.zeros(m, bool)
x[As._unpkd_ax[As._pkd_ax[0]:As._pkd_ax[1]]] = True
x = dev_vec(pos, x)
for i in range(n):
for j in range(n - 1, i, -1):
y = numpy.zeros(m, bool)
y[As._unpkd_ax[As._pkd_ax[j]:As._pkd_ax[j + 1]]] = True
y = dev_vec(pos, y)
dist[i, j] = dist[j, i] = calc_dist3(x, y, dist_vec)
x = y
return dist
def unifrac_mix_weighted2(As, otusid, t):
As.convert('csc')
m, n = As.shape
dist_vec, pos = readtree(t, otusid)
dist = numpy.zeros([n, n])
x = numpy.zeros(m)
x[As._unpkd_ax[As._pkd_ax[0]:As._pkd_ax[1]]] = As._values[As._pkd_ax[0]:As.
_pkd_ax[1]]
sx = sum(x)
x = dev_vec_weighted(dist_vec, pos, x)
for i in range(n):
for j in range(n - 1, i, -1):
y = numpy.zeros(m)
y[As._unpkd_ax[As._pkd_ax[j]:As._pkd_ax[j + 1]]] = As._values[
As._pkd_ax[j]:As._pkd_ax[j + 1]]
sy = sum(y)
y = dev_vec_weighted(dist_vec, pos, y)
dist[i, j] = dist[j,
i] = calc_dist_weighted(x, y, dist_vec, sx, sy)
x = y
sx = sy
return dist
