S2 Table -

<p><b>Table A.</b> Dinucleotide confusion matrix. Row headings are the model simulated while column headings are the predicted model. Bold-faced numbers indicate prediction accuracy for a given model, while underlined numbers mark the most common misclassification for a given model. Although false prediction rates are high for all four selection models, the primary source of confusion is distinguishing additive from dominant selection. Detection of neutral vs. non-neutral evolution is considerably more accurate. <b>Table B</b>. Trinucleotide confusion matrix. Notation is the same as that in <a href="http://www.plosgenetics.org/article/info:doi/10.1371/journal.pgen.1011524#pgen.1011524.t001" target="_blank">Table 1a</a>. Bold-faced numbers indicate prediction accuracy for a given model, while underlined numbers mark the most common misclassification for a given model. Although false prediction rates are high for all four selection models, the primary source of confusion is distinguishing additive from dominant selection. Detection of neutral vs. non-neutral evolution is considerably more accurate.</p> <p>(DOCX)</p>