Bayesian methods for genomic association of chromosomic regions considering the additive-dominance model

Abstract

Bayesian approaches applied in association studies select regions of single-nucleotide polymorphisms, indicating genes with important effects. The Bayesian methods differ in terms of the distribution assumed for the marker effects. Here, we used the window posterior probability of association to detect potential regions. The present study evaluated the efficiency of these methods in identifying regions located close to genes. Data were simulated in six scenarios. Considering the lack of dominance, BayesA was more efficient in the scenario with three QTLs. For scenarios with 10 or 100 QTLs, BayesCπ and BayesDπ were more efficient according to the false positive rate and detection power. Considering the presence of dominance, all methods were similar in the scenario with three QTLs, except in terms of accuracy. BayesDπ was superior in the scenario with 10 QTLs, while BRR was more efficient in the scenario with 100 QTLs.