Framework for stochastic simulations of additive and dominance genetic variation: Case study in layer chickens

Description

Here we present the framework for stochastic simulations of additive and dominance genetic variation that reflect variation found in real-world datasets. Stochastic simulations are a cost-effective method for testing any novelty within a breeding programme in silico before actual experimental validation and eventual deployment in the real-world. Accordingly, to be informative as a hypothesis-generation and decision-making tool, the simulations should reflect real-world variation as close as possible. While additive genetic variation is a staple to many breeding methods and underlying simulations, dominance genetic variation is typically neglected, oversimplified, or simulated in an ad hoc manner. However, dominance variation is a vital genetic component of many breeding programmes, especially in the terms of inbreeding depression and heterosis. Here we showcase a framework for calibrating stochastic simulation of additive and dominance variation to reflect variation in a real-world dataset. To this end, we used 24,492 SNP markers and egg production phenotypes from 2,408 commercial layer chickens to estimate additive and dominance genetic variances and inbreeding depression. We fitted both the full genomic directional dominance model and reduced dominance model (without the marker heterozygosity as a covariate) via Bayesian ridge regression. Furthermore, we built a framework of formulae and algorithms that use estimated real-world genetic parameters to fine-tune a simulation, in particular the mean and variance of dominance degrees (relative magnitude of biological dominance effects compared to biological additive effects) and the number of quantitative loci. We evaluated a full grid across the parameter space to find the most credible inputs so that the resulting simulation reflected variation found in a real-world dataset. This work sets the building blocks for future simulations of animal breeding programmes influenced by additive and dominance variation.

Period	13 Oct 2022
Event title	4th CiBreed Fall Workshop
Event type	Workshop
Location	Göttingen, GermanyShow on map
Degree of Recognition	International