An algorithm for genetic variance estimation of reproductive traits under a threshold model

Abstract

Some binary traits are determined by a large number of loci. The standard approach, in this case, is to model an unobservable variable (liability) with a fixed threshold. We present a method for the estimation of the genetic additive variance component under a threshold animal model with fixed effects included. The method can be applied to data with repeated measurement per animal. The unknown parameters of the model have been estimated by Gibbs sampling. The numerical properties of the method are investigated on simulated data for a large real pedigree of breeding stock of laying hens. The algorithm shows good mixing properties, producing consistent estimates from many distinct runs. The application of the method is exemplified on fertility data recorded for the same pedigree.