Prediction accuracy for a disease with population or case-control samples when true proportion of variance explained by the set of SNPs on the liability scale is 0.5, <i>τ</i> =  N/M is 1 for different disease prevalences.

<p>Exp1: Expected value from <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0071494#pone.0071494.e004" target="_blank">equation (2</a>) or <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0071494#pone.0071494.e019" target="_blank">equation (6</a>) of Daetwyler et al. (2008).</p><p>Exp2: Expected value from <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0071494#pone.0071494.e038" target="_blank">equation (9</a>) of Daetwyler et al (2008).</p><p>Exp3: Expected value from <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0071494#pone.0071494.e008" target="_blank">equation (3</a>).</p><p>Est: Average of estimates from 100 replicates.</p><p>se: Empirical standard error over 100 replicates.</p><p>Proportion of cases in case-control study is P = 0.5.