The prior probabilities influence the posterior distribution.

<p>Running the sampler with no data so that the posterior probabilities are determined by the prior probabilities only (<b>a</b>), with flat priors (<b>b</b>), or with different values of the shape parameter for the gamma distributions of genome length and number of contigs (<b>c–e</b>) all cause shifts in the posterior distribution of edge frequencies. The baseline for the change in frequency is Chain 1 from <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0099497#pone-0099497-g003" target="_blank">Figure 3</a> with , and all chains here were initialized with the same random seed as Chain 1.