Configuring the coalescent

Configuring the coalescent#

An introduction of how to configure the Coalescent can be found in the Quickstart guide. There are also separate sections focussing on configuring the demography and how to obtain more complex moments using rewards. Here we focus on additional configuration options.

Multiple-merger coalescents#

Supported coalescent models are StandardCoalescent, BetaCoalescent and DiracCoalescent, and can be specified when constructing the Coalescent distribution.

import phasegen as pg
from matplotlib import pyplot as plt

# compare 2-SFS of Kingman and Beta coalescents
kingman = pg.Coalescent(n=10, model=pg.StandardCoalescent())
beta = pg.Coalescent(n=10, model=pg.BetaCoalescent(alpha=1.5))

_, axs = plt.subplots(ncols=2, figsize=(8, 4), subplot_kw={"projection": "3d"})

# beta coalescent shows positive correlations between disparate minor allele frequencies
kingman.sfs.cov.plot_surface(ax=axs[0], show=False, title='Kingman')
beta.sfs.cov.plot_surface(ax=axs[1], title='Beta');
../../_images/e099798ffe285570ad14b9512f21e68c72eada7e34db8ad5ff2c5be08f512d99.png

Recombination#

Recombination is currently restricted to 2 loci and not compatible with multiple-merger coalescents and summary statistics based on the SFS. If you think an extension to this would be useful, please open an issue on GitHub.

coal = pg.Coalescent(n=8, loci=pg.LocusConfig(recombination_rate=0.1, n=2))

# time to ultimate MRCA
coal.tree_height.mean

1.8352605047108952

# marginal tree height of first locus
coal.tree_height.loci[0].mean

1.7499999999999993

coal = pg.Coalescent(n=8, loci=pg.LocusConfig(recombination_rate=0.1, n=2, n_unlinked=8))

# time to ultimate MRCA starting with 8 unlinked lineages
coal.tree_height.mean

1.938321304233787
Contents