Parsimonious model for analyzing parent-of-origin effects related to beef traits in dual-purpose Simmental.

Abstract

Genomic imprinting occurs when allelic effects depend on their parental origin. These parent-of-origin effects (POE) occur because of epigenetic DNA modifications during gametogenesis according to the sex of an animal. Animal breeding programs give little consideration to imprinting, although its relationship to important traits has been shown in different agricultural species. To incorporate imprinting, a previously proposed model (imprinting model) contains the genetic effects of the sire and dam, and it provides an estimate of the variance component due to POE, which is referred to as imprinting variance. Large volumes of data are sometimes available for commercial populations, so the dimension of mixed-model equations can become very large or even excessively large when estimating imprinting variances and other genetic parameters. To address this issue, we replaced the genetic effect as dam with the effect of the maternal grandsire in the imprinting model. When combined with appropriate weightings of the observations, this replacement yields an imprinting model with a parsimonious number of genetic effects for male parents and ancestors of slaughter animals, and it enables the inclusion of large volumes of data. In addition, we derived an equivalent model to facilitate the direct estimation of POE and their prediction error variances. We applied the parsimonious model to 1,366,160 fattening bulls as well as a pedigree of 2,637,761 ancestors to investigate the relevance of POE for beef performance in dual-purpose Simmental. We analyzed the killing-out percentage, net BW gain, carcass muscularity, and fat score as slaughter traits. The parsimonious model was applied as both linear and generalized linear versions with a logit-link function. The proportions of the total genetic variance attributable to POE ranged between 8.6% and 17.1%. For 3 of the 4 traits, the maternal gamete accounted for a greater proportion of the imprinting variance. The effects of POE and their reliabilities were estimated for up to 27,567 bulls and all traits, where the reliabilities ranged between 0.38 and 0.99. Thus, our new parsimonious model is appropriate for estimating the imprinting variance using large pedigree data sets. Our results highlight the need to consider POE in genetic evaluations.