Comparison of analyses of the QTLMAS XIV common dataset. I: genomic selection

BMC Proceedings

Table 1 Bayesian models developed for genomic selection.

Feature Model	BayesA	BayesB	BayesC (=SSVS stochastic search variable selection)	BayesCpi
Probability for a locus to be a QTL	1	1-p	1-p	1-p
QTL-specific effect variance (variance heterogeneity)	Yes	Yes	No	No
Modelling of no-QTL	Not aplicable	Null variance	Tiny variance	Null variance
Estimated parameter				p(uniform prior)
Hyperparameters (assumed known)	df¹, S²	df, S, p	df, S, p	df, S
Use Metropolis-Hastings sampler?	No	Yes	No	No

¹df=degrees of freedom; ²S=scale parameter, the two parameters of scaled inverted Chi-square distribution (df, S) used as a priori distribution for QTL effect variance

ISSN: 1753-6561