This method models the covariance in allele frequencies between populations on a landscape as a decreasing function of their pairwise geographic and ecological distance. Allele frequencies are modeled as a spatial Gaussian process with a parametric covariance function. The parameters of this covariance function, as well as the spatially smoothed allele frequencies, are estimated in a custom Markov chain Monte Carlo.
The two inference functions are MCMC and MCMC_BB, which call the
Markov chain Monte Carlo algorithms on the standard and overdispersion (Beta-Binomial)
models, respectively. To evaluate MCMC performance, there are a number of MCMC diagnosis
and visualization functions, which variously show the trace, plots, marginal and joint
marginal densities, and parameter acceptance rates. To evaluate model adequacy, there is
a posterior predictive sample function (posterior.predictive.sample), and an
accompanying function to plot its output and visually assess the model's ability to
describe the user's data.
Gideon Bradburd
Maintainer: Gideon Bradburd <gbradburd@ucdavis.edu>
Bradburd, G.S., Ralph, P.L., and Coop, G.M. Disentangling the effects of geographic and ecological isolation on genetic differentiation. Evolution 2013.