Parallel Computing for Mixed Model Implementation of Genomic Prediction and Variance Component Estimation of Additive and Dominance Effects

Abstract Text:

We developed the GVCBLUP package using shared memory (SM) and Message Passing Interface (MPI) parallel computing for genomic prediction and variance component estimation using mixed model methods. The GREML_CE and GREML_QM programs in the package offer complementary computing advantages and identical results of GBLUP and GREML along with heritability estimates using a combination of EM-REML and AI-REML algorithms. GREML_CE was designed for large numbers of SNP markers and GREML_QM for large numbers of individuals. For the SM version, GREML_CE could analyze 50,000 individuals with 400K SNP markers and GREML_QM could analyze 100,000 individuals with 50K SNP markers. For the MPI version, GREML_CE was tested for 50,000 individuals with 1 million SNP markers and 100,000 individuals with 41K SNP markers. 

Keywords:

GBLUP, genomic selection, variance component, heritability, BLUP