Fast genomic prediction of breeding values using parallel Markov chain Monte Carlo with convergence diagnosis

BMC Bioinformatics

Table 1 Genomic predictive accuracies obtained using FBM-BayesA, TBM-BayesA, FBM-BayesCπ, TBM-BayesCπ, and GBLUP in Scenario 1

Chains	FBM-BayesA	TBM-BayesA	FBM-BayesCπ	TBM-BayesCπ	GBLUP
1	0.5239	0.5231	0.6316	0.6317	0.6016
2	0.5227	0.5229	0.6301	0.6296
4	0.5230	0.5229	0.6304	0.6314
6	0.5230	0.5230	0.6310	0.6304
8	0.5231	0.5232	0.6313	0.6311
10	0.5232	0.5231	0.6309	0.6307
12	0.5228	0.5232	0.6315	0.6305
14	0.5230	0.5231	0.6306	0.6313
16	0.5230	0.5231	0.6307	0.6303
18	0.5230	0.5231	0.6310	0.6311

FBM-BayesA fixed burn-in, multiple-chain BayesA, TBM-BayesA tunable burn-in, multiple-chain BayesA, FBM-BayesCπ fixed burn-in, multiple-chain BayesCπ, TBM-BayesCπ tunable burn-in, multiple-chain BayesCπ, and Chains number of parallel MCMC running for each genomic prediction model. Simulation parameters are as follows: population size = 1000; number of QTL = 200; heritability = 0.1; number of chromosomes = 5; and number of markers per chromosome = 4000

ISSN: 1471-2105