Effect of genomic selection and genotyping strategy on estimation of variance components in animal models using different relationship matrices

Abstract

Background: The traditional way to estimate variance components (VC) is based on the animal model using a pedigree-based relationship matrix (A) (A-AM). After genomic selection was introduced into breeding programs, it was anticipated that VC estimates from A-AM would be biased because the effect of selection based on genomic information is not captured. The single-step method (H-AM), which uses an H matrix as (co)variance matrix, can be used as an alternative to estimate VC. Here, we compared VC estimates from A-AM and HAM and investigated the effect of genomic selection, genotyping strategy and genotyping proportion on the estimation of VC from the two methods, by analyzing a dataset from a commercial broiler line and a simulated dataset that mimicked the broiler population. Results: VC estimates from HAM were severely overestimated with a high proportion of selective genotyping, and overestimation increased as proportion of genotyping increased in the analysis of both commercial and simulated data. This bias in HAM estimates arises when selective genotyping is used to construct the H-matrix, regardless of whether selective genotyping is applied or not in the selection process. For simulated populations under genomic selection, estimates of genetic variance from A-AM were also significantly overestimated when the effect of genomic selection was strong. Our results suggest that VC estimates from HAM under random genotyping have the expected values. Predicted breeding values from HAM were inflated when VC estimates were biased, and inflation differed between genotyped and ungenotyped animals, which can lead to suboptimal selection decisions. Conclusions: We conclude that VC estimates from HAM are biased with selective genotyping, but are close to expected values with random genotyping.VC estimates from A-AM in populations under genomic selection are also biased but to a much lesser degree. Therefore, we recommend the use of HAM with random genotyping to estimate VC for populations under genomic selection. Our results indicate that it is still possible to use selective genotyping in selection, but then VC estimation should avoid the use of genotypes from one side only of the distribution of phenotypes. Hence, a dual genotyping strategy may be needed to address both selection and VC estimation.

Phenotypically Selective Genotyping Realizes More Genetic Gains in a Rainbow Trout Breeding Program in the Presence of Genotype-by-Environment Interactions

“…In another case of selective genotyping of top animals, when both non-genotyped and genotyped individuals are the selection candidates, it could be unfair to genotyped individuals to compare breeding values. The use of ssGBLUP to estimate variance components and predict EBV leads to severely inflated EBV for non-genotyped individuals in this situation (Wang et al, 2020). In addition, bias may lead to estimates of genetic trends that are higher or lower than the true rate of genetic gain.…”

“…However, when genotyping is selective, it is difficult to obtain unbiased estimates of the variance components. Wang et al (2020) showed that the use of ssGBLUP led to an overestimation of variance components when selective genotyping of top animals was used. This stands in contrast to our study, in which variance components were underestimated when the GBLUP model was used in scenarios that involved selective genotyping of top fish.…”

“…In many breeding programs, genotyping is limited to phenotypically superior animals, referred to as selective genotyping of top animals. Such selective genotyping leads to biased predictions of genomic breeding values (GEBV) when genomic-based best linear unbiased prediction (GBLUP) is used (Gowane et al, 2019;Wang et al, 2020). For example, Wang et al (2020) showed that the use of a combined matrix (Christensen and Lund, 2010;Aguilar et al, 2011) of pedigree and genomic relationships in a single-step GBLUP (ssGBLUP) prediction resulted in the upward bias of GEBV and overestimation of variance components when only a proportion of top individuals were genotyped.…”

“…According to Gowane et al (2019), genotyping of phenotypically contrasting animals (selective genotyping of top and bottom animals) for selection candidates is superior to selective genotyping of top animals. The undesirable consequences of selectively genotyping top animals have been addressed extensively (VanRaden et al, 2009;Patry and Ducrocq, 2011;Vitezica et al, 2011;Boligon et al, 2012;Jiménez-Montero et al, 2012;Chu et al, 2019;Gowane et al, 2019;Wang et al, 2020), but the superiority of this selective strategy for breeding programs has been shown only by Howard et al (2018). In addition, some simulation studies (Patry and Ducrocq, 2011;Vitezica et al, 2011;Boligon et al, 2012;Gowane et al, 2019) have used the correlation between true breeding values (TBV) and GEBV to compare different genotyping strategies.…”

“…Comparisons of genotyping strategies have focused on species other than rainbow trout (Boligon et al, 2012;Howard et al, 2018;Gowane et al, 2019;Wang et al, 2020). In two studies (Boligon et al, 2012;Gowane et al, 2019), different genotyping strategies based on phenotypes were used with individuals in the reference population, but the animals in the validation population did not have phenotypes, and thus, random genotyping was used with these validation individuals.…”