def voight_painting(h):
"""Paint haplotypes, assigning a unique integer to each shared haplotype
prefix.
Parameters
----------
h : array_like, int, shape (n_variants, n_haplotypes)
Haplotype array.
Returns
-------
painting : ndarray, int, shape (n_variants, n_haplotypes)
Painting array.
indices : ndarray, int, shape (n_hapotypes,)
Haplotype indices after sorting by prefix.
"""
# check inputs
# N.B., ensure int8 so we can use cython optimisation
h = HaplotypeArray(np.asarray(h), copy=False)
if h.max() > 1:
raise NotImplementedError('only biallelic variants are supported')
if h.min() < 0:
raise NotImplementedError('missing calls are not supported')
# sort by prefix
indices = h.prefix_argsort()
h = np.take(h, indices, axis=1)
# paint
painting = paint_shared_prefixes(np.asarray(h))
return painting, indices
def haplotype_diversity(h):
"""Estimate haplotype diversity.
Parameters
----------
h : array_like, int, shape (n_variants, n_haplotypes)
Haplotype array.
Returns
-------
hd : float
Haplotype diversity.
"""
# check inputs
h = HaplotypeArray(h, copy=False)
# number of haplotypes
n = h.n_haplotypes
# compute haplotype frequencies
f = h.distinct_frequencies()
# estimate haplotype diversity
hd = (1 - np.sum(f ** 2)) * n / (n - 1)
return hd
def test_mean_pairwise_diversity(self):
# start with simplest case, two haplotypes, one pairwise comparison
h = HaplotypeArray([[0, 0],
[1, 1],
[0, 1],
[1, 2],
[0, -1],
[-1, -1]])
ac = h.count_alleles()
expect = [0, 0, 1, 1, -1, -1]
actual = allel.stats.mean_pairwise_difference(ac, fill=-1)
aeq(expect, actual)
# four haplotypes, 6 pairwise comparison
h = HaplotypeArray([[0, 0, 0, 0],
[0, 0, 0, 1],
[0, 0, 1, 1],
[0, 1, 1, 1],
[1, 1, 1, 1],
[0, 0, 1, 2],
[0, 1, 1, 2],
[0, 1, -1, -1],
[-1, -1, -1, -1]])
ac = h.count_alleles()
expect = [0, 3/6, 4/6, 3/6, 0, 5/6, 5/6, 1, -1]
actual = allel.stats.mean_pairwise_difference(ac, fill=-1)
assert_array_close(expect, actual)
def haplotype_diversity(h):
"""Estimate haplotype diversity.
Parameters
----------
h : array_like, int, shape (n_variants, n_haplotypes)
Haplotype array.
Returns
-------
hd : float
Haplotype diversity.
"""
# check inputs
h = HaplotypeArray(h, copy=False)
# number of haplotypes
n = h.n_haplotypes
# compute haplotype frequencies
f = h.distinct_frequencies()
# estimate haplotype diversity
hd = (1 - np.sum(f**2)) * n / (n - 1)
return hd
def plot_haplotype_frequencies(h,
palette='Paired',
singleton_color='w',
ax=None):
"""Plot haplotype frequencies.
Parameters
----------
h : array_like, int, shape (n_variants, n_haplotypes)
Haplotype array.
palette : string, optional
A Seaborn palette name.
singleton_color : string, optional
Color to paint singleton haplotypes.
ax : axes, optional
The axes on which to draw. If not provided, a new figure will be
created.
Returns
-------
ax : axes
"""
import matplotlib.pyplot as plt
import seaborn as sns
# check inputs
h = HaplotypeArray(h, copy=False)
# setup figure
if ax is None:
width = plt.rcParams['figure.figsize'][0]
height = width / 10
fig, ax = plt.subplots(figsize=(width, height))
sns.despine(ax=ax, left=True)
# count distinct haplotypes
hc = h.distinct_counts()
# setup palette
n_colors = np.count_nonzero(hc > 1)
palette = sns.color_palette(palette, n_colors)
# paint frequencies
x1 = 0
for i, c in enumerate(hc):
x2 = x1 + c
if c > 1:
color = palette[i]
else:
color = singleton_color
ax.axvspan(x1, x2, color=color)
x1 = x2
# tidy up
ax.set_xlim(0, h.shape[1])
ax.set_yticks([])
return ax
def plot_haplotype_frequencies(h, palette='Paired', singleton_color='w',
ax=None):
"""Plot haplotype frequencies.
Parameters
----------
h : array_like, int, shape (n_variants, n_haplotypes)
Haplotype array.
palette : string, optional
A Seaborn palette name.
singleton_color : string, optional
Color to paint singleton haplotypes.
ax : axes, optional
The axes on which to draw. If not provided, a new figure will be
created.
Returns
-------
ax : axes
"""
import matplotlib.pyplot as plt
import seaborn as sns
# check inputs
h = HaplotypeArray(h, copy=False)
# setup figure
if ax is None:
width = plt.rcParams['figure.figsize'][0]
height = width / 10
fig, ax = plt.subplots(figsize=(width, height))
sns.despine(ax=ax, left=True)
# count distinct haplotypes
hc = h.distinct_counts()
# setup palette
n_colors = np.count_nonzero(hc > 1)
palette = sns.color_palette(palette, n_colors)
# paint frequencies
x1 = 0
for i, c in enumerate(hc):
x2 = x1 + c
if c > 1:
color = palette[i]
else:
color = singleton_color
ax.axvspan(x1, x2, color=color)
x1 = x2
# tidy up
ax.set_xlim(0, h.shape[1])
ax.set_yticks([])
return ax
def test_slice_types(self):
h = HaplotypeArray(haplotype_data, dtype='i1')
# row slice
s = h[1:]
assert_is_instance(s, HaplotypeArray)
# col slice
s = h[:, 1:]
assert_is_instance(s, HaplotypeArray)
# row index
s = h[0]
assert_is_instance(s, np.ndarray)
assert_not_is_instance(s, HaplotypeArray)
# col index
s = h[:, 0]
assert_is_instance(s, np.ndarray)
assert_not_is_instance(s, HaplotypeArray)
# item
s = h[0, 0]
assert_is_instance(s, np.int8)
assert_not_is_instance(s, HaplotypeArray)
def garud_h(h):
"""Compute the H1, H12, H123 and H2/H1 statistics for detecting signatures
of soft sweeps, as defined in Garud et al. (2015).
Parameters
----------
h : array_like, int, shape (n_variants, n_haplotypes)
Haplotype array.
Returns
-------
h1 : float
H1 statistic (sum of squares of haplotype frequencies).
h12 : float
H12 statistic (sum of squares of haplotype frequencies, combining
the two most common haplotypes into a single frequency).
h123 : float
H123 statistic (sum of squares of haplotype frequencies, combining
the three most common haplotypes into a single frequency).
h2_h1 : float
H2/H1 statistic, indicating the "softness" of a sweep.
"""
# check inputs
h = HaplotypeArray(h, copy=False)
# compute haplotype frequencies
f = h.distinct_frequencies()
# compute H1
h1 = np.sum(f ** 2)
# compute H12
h12 = np.sum(f[:2]) ** 2 + np.sum(f[2:] ** 2)
# compute H123
h123 = np.sum(f[:3]) ** 2 + np.sum(f[3:] ** 2)
# compute H2/H1
h2 = h1 - f[0] ** 2
h2_h1 = h2 / h1
return h1, h12, h123, h2_h1
def ehh_decay(h, truncate=False):
"""Compute the decay of extended haplotype homozygosity (EHH)
moving away from the first variant.
Parameters
----------
h : array_like, int, shape (n_variants, n_haplotypes)
Haplotype array.
truncate : bool, optional
If True, the return array will exclude trailing zeros.
Returns
-------
ehh : ndarray, float, shape (n_variants, )
EHH at successive variants from the first variant.
"""
from allel.opt.stats import pairwise_shared_prefix_lengths_int8
# check inputs
# N.B., ensure int8 so we can use cython optimisation
h = HaplotypeArray(np.asarray(h, dtype="i1"), copy=False)
if h.max() > 1:
raise NotImplementedError("only biallelic variants are supported")
if h.min() < 0:
raise NotImplementedError("missing calls are not supported")
# initialise
n_variants = h.n_variants # number of rows, i.e., variants
n_haplotypes = h.n_haplotypes # number of columns, i.e., haplotypes
n_pairs = (n_haplotypes * (n_haplotypes - 1)) // 2
# compute the shared prefix length between all pairs of haplotypes
spl = pairwise_shared_prefix_lengths_int8(h)
# compute EHH by counting the number of shared prefixes extending beyond
# each variant
minlength = None if truncate else n_variants + 1
b = np.bincount(spl, minlength=minlength)
c = np.cumsum(b[::-1])[:-1]
ehh = (c / n_pairs)[::-1]
return ehh
def ehh_decay(h, truncate=False):
"""Compute the decay of extended haplotype homozygosity (EHH)
moving away from the first variant.
Parameters
----------
h : array_like, int, shape (n_variants, n_haplotypes)
Haplotype array.
truncate : bool, optional
If True, the return array will exclude trailing zeros.
Returns
-------
ehh : ndarray, float, shape (n_variants, )
EHH at successive variants from the first variant.
"""
from allel.opt.stats import pairwise_shared_prefix_lengths_int8
# check inputs
# N.B., ensure int8 so we can use cython optimisation
h = HaplotypeArray(np.asarray(h, dtype='i1'), copy=False)
if h.max() > 1:
raise NotImplementedError('only biallelic variants are supported')
if h.min() < 0:
raise NotImplementedError('missing calls are not supported')
# initialise
n_variants = h.n_variants # number of rows, i.e., variants
n_haplotypes = h.n_haplotypes # number of columns, i.e., haplotypes
n_pairs = (n_haplotypes * (n_haplotypes - 1)) // 2
# compute the shared prefix length between all pairs of haplotypes
spl = pairwise_shared_prefix_lengths_int8(h)
# compute EHH by counting the number of shared prefixes extending beyond
# each variant
minlength = None if truncate else n_variants + 1
b = np.bincount(spl, minlength=minlength)
c = np.cumsum(b[::-1])[:-1]
ehh = (c / n_pairs)[::-1]
return ehh
def garud_h(h):
"""Compute the H1, H12, H123 and H2/H1 statistics for detecting signatures
of soft sweeps, as defined in Garud et al. (2015).
Parameters
----------
h : array_like, int, shape (n_variants, n_haplotypes)
Haplotype array.
Returns
-------
h1 : float
H1 statistic (sum of squares of haplotype frequencies).
h12 : float
H12 statistic (sum of squares of haplotype frequencies, combining
the two most common haplotypes into a single frequency).
h123 : float
H123 statistic (sum of squares of haplotype frequencies, combining
the three most common haplotypes into a single frequency).
h2_h1 : float
H2/H1 statistic, indicating the "softness" of a sweep.
"""
# check inputs
h = HaplotypeArray(h, copy=False)
# compute haplotype frequencies
f = h.distinct_frequencies()
# compute H1
h1 = np.sum(f**2)
# compute H12
h12 = np.sum(f[:2])**2 + np.sum(f[2:]**2)
# compute H123
h123 = np.sum(f[:3])**2 + np.sum(f[3:]**2)
# compute H2/H1
h2 = h1 - f[0]**2
h2_h1 = h2 / h1
return h1, h12, h123, h2_h1
def test_windowed_diversity(self):
# four haplotypes, 6 pairwise comparison
h = HaplotypeArray([[0, 0, 0, 0],
[0, 0, 0, 1],
[0, 0, 1, 1],
[0, 1, 1, 1],
[1, 1, 1, 1],
[0, 0, 1, 2],
[0, 1, 1, 2],
[0, 1, -1, -1],
[-1, -1, -1, -1]])
ac = h.count_alleles()
# mean pairwise diversity
# expect = [0, 3/6, 4/6, 3/6, 0, 5/6, 5/6, 1, -1]
pos = SortedIndex([2, 4, 7, 14, 15, 18, 19, 25, 27])
expect = [(7/6)/10, (13/6)/10, 1/11]
actual, _, _, _ = allel.stats.windowed_diversity(pos, ac, size=10,
start=1,
stop=31)
assert_array_close(expect, actual)
def test_mean_pairwise_divergence(self):
# simplest case, two haplotypes in each population
h = HaplotypeArray([[0, 0, 0, 0],
[0, 0, 0, 1],
[0, 0, 1, 1],
[0, 1, 1, 1],
[1, 1, 1, 1],
[0, 0, 1, 2],
[0, 1, 1, 2],
[0, 1, -1, -1],
[-1, -1, -1, -1]])
h1 = h.take([0, 1], axis=1)
h2 = h.take([2, 3], axis=1)
ac1 = h1.count_alleles()
ac2 = h2.count_alleles()
expect = [0/4, 2/4, 4/4, 2/4, 0/4, 4/4, 3/4, -1, -1]
actual = allel.stats.mean_pairwise_difference_between(ac1, ac2,
fill=-1)
aeq(expect, actual)
def test_constructor(self):
# missing data arg
with self.assertRaises(TypeError):
# noinspection PyArgumentList
HaplotypeArray()
# data has wrong dtype
data = 'foo bar'
with self.assertRaises(TypeError):
HaplotypeArray(data)
# data has wrong dtype
data = [4., 5., 3.7]
with self.assertRaises(TypeError):
HaplotypeArray(data)
# data has wrong dimensions
data = [1, 2, 3]
with self.assertRaises(TypeError):
HaplotypeArray(data)
# data has wrong dimensions
data = diploid_genotype_data # use GenotypeArray instead
with self.assertRaises(TypeError):
HaplotypeArray(data)
# haploid data (typed)
h = HaplotypeArray(haplotype_data, dtype='i1')
aeq(haplotype_data, h)
eq(np.int8, h.dtype)
def test_windowed_divergence(self):
# simplest case, two haplotypes in each population
h = HaplotypeArray([[0, 0, 0, 0],
[0, 0, 0, 1],
[0, 0, 1, 1],
[0, 1, 1, 1],
[1, 1, 1, 1],
[0, 0, 1, 2],
[0, 1, 1, 2],
[0, 1, -1, -1],
[-1, -1, -1, -1]])
h1 = h.take([0, 1], axis=1)
h2 = h.take([2, 3], axis=1)
ac1 = h1.count_alleles()
ac2 = h2.count_alleles()
# mean pairwise divergence
# expect = [0/4, 2/4, 4/4, 2/4, 0/4, 4/4, 3/4, -1, -1]
pos = SortedIndex([2, 4, 7, 14, 15, 18, 19, 25, 27])
expect = [(6/4)/10, (9/4)/10, 0/11]
actual, _, _, _ = allel.stats.windowed_divergence(
pos, ac1, ac2, size=10, start=1, stop=31
)
assert_array_close(expect, actual)
def voight_painting(h):
"""Paint haplotypes, assigning a unique integer to each shared haplotype
prefix.
Parameters
----------
h : array_like, int, shape (n_variants, n_haplotypes)
Haplotype array.
Returns
-------
painting : ndarray, int, shape (n_variants, n_haplotypes)
Painting array.
indices : ndarray, int, shape (n_hapotypes,)
Haplotype indices after sorting by prefix.
"""
from allel.opt.stats import paint_shared_prefixes_int8
# check inputs
# N.B., ensure int8 so we can use cython optimisation
h = HaplotypeArray(np.asarray(h, dtype="i1"), copy=False)
if h.max() > 1:
raise NotImplementedError("only biallelic variants are supported")
if h.min() < 0:
raise NotImplementedError("missing calls are not supported")
# sort by prefix
indices = h.prefix_argsort()
h = np.take(h, indices, axis=1)
# paint
painting = paint_shared_prefixes_int8(h)
return painting, indices
def paint_transmission(parent_haplotypes, progeny_haplotypes):
"""Paint haplotypes inherited from a single diploid parent according to
their allelic inheritance.
Parameters
----------
parent_haplotypes : array_like, int, shape (n_variants, 2)
Both haplotypes from a single diploid parent.
progeny_haplotypes : array_like, int, shape (n_variants, n_progeny)
Haplotypes found in progeny of the given parent, inherited from the
given parent. I.e., haplotypes from gametes of the given parent.
Returns
-------
painting : ndarray, uint8, shape (n_variants, n_progeny)
An array of integers coded as follows: 1 = allele inherited from
first parental haplotype; 2 = allele inherited from second parental
haplotype; 3 = reference allele, also carried by both parental
haplotypes; 4 = non-reference allele, also carried by both parental
haplotypes; 5 = non-parental allele; 6 = either or both parental
alleles missing; 7 = missing allele; 0 = undetermined.
Examples
--------
>>> import allel
>>> haplotypes = allel.HaplotypeArray([
... [0, 0, 0, 1, 2, -1],
... [0, 1, 0, 1, 2, -1],
... [1, 0, 0, 1, 2, -1],
... [1, 1, 0, 1, 2, -1],
... [0, 2, 0, 1, 2, -1],
... [0, -1, 0, 1, 2, -1],
... [-1, 1, 0, 1, 2, -1],
... [-1, -1, 0, 1, 2, -1],
... ], dtype='i1')
>>> painting = allel.paint_transmission(haplotypes[:, :2],
... haplotypes[:, 2:])
>>> painting
array([[3, 5, 5, 7],
[1, 2, 5, 7],
[2, 1, 5, 7],
[5, 4, 5, 7],
[1, 5, 2, 7],
[6, 6, 6, 7],
[6, 6, 6, 7],
[6, 6, 6, 7]], dtype=uint8)
"""
# check inputs
parent_haplotypes = HaplotypeArray(parent_haplotypes)
progeny_haplotypes = HaplotypeArray(progeny_haplotypes)
if parent_haplotypes.n_haplotypes != 2:
raise ValueError('exactly two parental haplotypes should be provided')
# convenience variables
parent1 = parent_haplotypes[:, 0, np.newaxis]
parent2 = parent_haplotypes[:, 1, np.newaxis]
progeny_is_missing = progeny_haplotypes < 0
parent_is_missing = np.any(parent_haplotypes < 0, axis=1)
# need this for broadcasting, but also need to retain original for later
parent_is_missing_bc = parent_is_missing[:, np.newaxis]
parent_diplotype = GenotypeArray(parent_haplotypes[:, np.newaxis, :])
parent_is_hom_ref = parent_diplotype.is_hom_ref()
parent_is_het = parent_diplotype.is_het()
parent_is_hom_alt = parent_diplotype.is_hom_alt()
# identify allele calls where inheritance can be determined
is_callable = ~progeny_is_missing & ~parent_is_missing_bc
is_callable_seg = is_callable & parent_is_het
# main inheritance states
inherit_parent1 = is_callable_seg & (progeny_haplotypes == parent1)
inherit_parent2 = is_callable_seg & (progeny_haplotypes == parent2)
nonseg_ref = (is_callable & parent_is_hom_ref & (progeny_haplotypes == parent1))
nonseg_alt = (is_callable & parent_is_hom_alt & (progeny_haplotypes == parent1))
nonparental = (
is_callable & (progeny_haplotypes != parent1) & (progeny_haplotypes != parent2)
)
# record inheritance states
# N.B., order in which these are set matters
painting = np.zeros(progeny_haplotypes.shape, dtype='u1')
painting[inherit_parent1] = INHERIT_PARENT1
painting[inherit_parent2] = INHERIT_PARENT2
painting[nonseg_ref] = INHERIT_NONSEG_REF
painting[nonseg_alt] = INHERIT_NONSEG_ALT
painting[nonparental] = INHERIT_NONPARENTAL
painting[parent_is_missing] = INHERIT_PARENT_MISSING
painting[progeny_is_missing] = INHERIT_MISSING
return painting
def setup_instance(self, data):
return HaplotypeArray(data)
