| A novel approach based on latent variable modelling is presented for the analysis of multivariate quantitative and qualitative trait loci. The approach is general in the sense that it enables the joint analysis of many kinds of quantitative and qualitative traits (including count data and censored traits) in a single modelling framework. In the framework, the observations are modelled as functions of latent variables, which are then affected by quantitative trait loci. Separating the analysis in this way means that measurement errors in the phenotypic observations can be included easily in the model, providing robust inferences. The performance of the method is illustrated using two real multivariate datasets, from barley and Scots pine. |
