Projects per year
The aim of this research was to develop and to test different procedures that integrate estimates of single nucleotide polymorphism (SNP) effects and associated measures of precision from a foreign SNP Best Linear Unbiased Prediction (SNPBLUP), into a domestic SNPBLUP when exchange of genotypes or phenotypes is prohibited for whatever reason. In addition to the foreign estimates of SNP effects, procedures were developed assuming the availability of associated: 1) prediction error (co)variance (PEC) matrix; 2) PEC matrix separately for each chromosome; 3) prediction error variances (PEV) only; 4) PEV, allele frequencies, and linkage disequilibrium (LD) of foreign training set; and 5) as 4) but with LD measured on foreign selection candidates. We tested these approaches with a simulation of two historically related populations for a single trait. We confirmed that integrating foreign estimates of SNP effects and the associated PEC matrix led to the same direct genomic values for selection candidates as the joint SNPBLUP using datasets from both populations. Integrating foreign estimates and PEV only led to biased and inaccurate predictions. Procedures based on partial PEC matrices or on LD information gave almost as accurate and unbiased predictions as the joint SNPBLUP. Therefore, accurate integration of foreign estimates of SNP effects into a domestic SNPBLUP seems possible, even if only PEV and some population statistics are available.