We observe that the two methods using a marker-based relationship matrix perform better than the method using the pedigree based relationship matrix, but as expected the one-step method performs the best.
The first method (ped) computes traditional EBVs using the pedigree based relationship matrix (without using markers).
For comparison, the prediction using the pedigree based relationship matrix (ped method) and the genomic prediction using (10) based on genotyped animals (two-step) are also shown.
Allele frequencies were calculated using the pedigree base population, that is, the 1020 individuals in the pedigree with unknown parents.
Narrow sense heritability is based on a ratio of the between and within family variance and is therefore reduced (and was similarly reduced in the pedigree based estimate using only the 1,000 genotyped individuals (not shown)).
Similarly, the pedigree based BLUP methods were also accurate for closely related animals, however when the pedigree based BLUP methods were used to predict unrelated animals, the accuracy was close to zero.
For the gap in kinship between the dogs genotyped on only one of the arrays, we used their pedigree based kinship.
BLUP: conventional BLUP using a pedigree based relationship matrix; G-BLUP: Genome-enabled Best Linear Unbiased Prediction (G-BLUP); RidgeP: Regressionession BLUP; RRPCA: Ridge Regression with PCA reduction; BayesSSVS: Bayesian Stochastic Search Variable Selection BayesCC.
Description: BLUP: conventional BLUP using a pedigree based relationship matrix; G-BLUP: Genome-enabled Best Linear Unbiased Prediction (G-BLUP); RidgeP: Regressionession BLUP; RRPCA: Ridge Regression with PCA reduction; BayesSSVS: Bayesian Stochastic Search Variable Selection BayesCC.
We expect using IBD estimates obtained from such methods, rather than solely using pedigree based estimates of relatedness, will provide gains similar to what we obtained here.
