Ancestral Graph with Bias in Gene Conversion

Gene conversion is a mechanism by which a double-strand break in a DNA molecule is repaired using a homologous DNA molecule as a template. As a result, one gene is 'copied and pasted' onto the other gene. It was recently reported that the direction of gene conversion appears to be biased towards G and C nucleotides. In this paper a stochastic model of the dynamics of the bias in gene conversion is developed for a finite population of members in a multigene family. The dual process is the biased voter model, which generates an ancestral random graph for a given sample. An importance-sampling algorithm for computing the likelihood of the sample is also given.