Inferred average <i>h</i>*<i>s</i> for each of the 4096 discrete DFE/<i>h</i> models considered.

<p>Inferred average <i>h</i>*<i>s</i> is calculated as the sum over all the bins of the DFE of the expected value of s for the bin multiplied by the value of <i>h</i> for that bin multiplied by the proportion of mutations inferred to be in that bin of the DFE. Each point represents a particular model. Red points denote those models with a log-likelihood <4.72 units below the fully additive model. These models fit the data well. (<b>A)</b> All models. <b>(B)</b> Zooming in to the models with the highest log-likelihood.</p> <p>(PDF)</p>