def phase_parents_by_transmission(g, window_size):
"""Phase parent genotypes from a trio or cross, given progeny genotypes
already phased by Mendelian transmission.
Parameters
----------
g : GenotypeArray
Genotype array, with parents as first two columns and progeny as
remaining columns, where progeny genotypes are already phased.
window_size : int
Number of previous heterozygous sites to include when phasing each
parent. A number somewhere between 10 and 100 may be appropriate,
depending on levels of heterozygosity and quality of data.
Returns
-------
g : GenotypeArray
Genotype array with parents phased where possible.
"""
# setup
check_type(g, GenotypeArray)
check_dtype(g.values, 'i1')
check_ploidy(g.ploidy, 2)
if g.is_phased is None:
raise ValueError('genotype array must first have progeny phased by transmission')
check_min_samples(g.n_samples, 3)
# run the phasing
is_phased = g.is_phased.view('u1')
phase_parents_by_transmission_int8(g.values, is_phased, window_size)
# outputs
return g
def is_phased(self, is_phased):
if is_phased is not None:
is_phased = ensure_dask_array(is_phased, self.chunks[:-1])
check_shape(is_phased, self.shape[:-1])
check_dtype(is_phased, bool)
self._is_phased = is_phased
def mask(self, mask):
if mask is not None:
mask = ensure_dask_array(mask, self.chunks[:-1])
check_shape(mask, self.shape[:-1])
check_dtype(mask, bool)
self._mask = mask
