Modeling additive genetic effects in animal models by closed skew normal distribution

• Published in: Communications in statistics. Simulation and computation Vol. 51; no. 3; pp. 1186 - 1198
• Main Authors: Pakbaz, F., Hosseini, F., Nematollahi, A. R.
• Format: Journal Article
• Language: English
• Published: Philadelphia Taylor & Francis 09-03-2022; Taylor & Francis Ltd
• Subjects: 92D50 62J12; Algorithms; Animal Model; Bayesian analysis; Bayesian approach; Closed skew normal; Computer simulation; Genetic random effects; Markov chains; Normal distribution; Parameter estimation; Skewed distributions; Statistical inference; Statistical models
• ISSN: 0361-0918; 1532-4141
• DOI: 10.1080/03610918.2019.1664576

Animal models are used commonly for modeling genetic responses. In these models the response variable can be Gaussian or Non-Gaussian, so these models belong to the generalized linear mixed models, where the genetic correlation structure of data is considered through random effects with the normal distribution. But in many applications, it is unclear whether or not the normal assumption holds. Wrong Gaussian assumptions cause bias in the component variance estimates and affect the accuracy of results. In this paper, we have proposed a closed skew normal distribution for the genetic random effects which is more flexible and includes the normal distribution. The Bayesian inference approach and the Markov Chain Monte Carlo algorithms are developed for the parameter estimations. The performance of the proposed models is illustrated by a simulation study and an example. The accuracy of the closed skew normal model is favorably compared with the normal model in the simulation and the real example.