Genome size and selection strength in prokaryotes.

<p>(<b>A</b>) Mean number of genes <i>x</i> is plotted against inferred selection strength d<i>N</i>/d<i>S</i> where each point represents one prokaryotic cluster (ATGC). Error bars represent genome sizes distributions widths and indicate one standard deviation. (<b>B</b>) Mean number of genes is plotted against extracted effective population size <i>N</i>_<i>e</i>. A representative global trend curve of mean genome size as predicted by the model (see Eq (<a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0195571#pone.0195571.e007" target="_blank">7</a>)), where all model parameters are assumed to be global <b><i>θ</i></b> = {<i>s</i>,<i>r</i>′,<i>λ</i>} is indicated by a red line. The approach implemented in the hard fitting methodology, where Eq (<a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0195571#pone.0195571.e007" target="_blank">7</a>) is used in order to set latent variable value such that model distributions are centered around observed genome sizes, is illustrated in a dashed orange line.