Mixed models are one of the most widely used
tools in applied statistics. Applications range from biostatistics (clinical
trials, longitudinal studies), agricultural/biological (animal and plant
breeding), environmental/epidemiological (spatial nutrient, pollutant and
disease mapping) to educational and social sciences (multi-level modelling).
Basic mixed models which had their origins in the analysis of designed
experiments have been expanded to allow for non-Gaussian random effects, larger
data-sets or complex variance modelling. The explosion of applications in the
bioinformatics area where mixed models are being applied to the analysis of QTL
experiments or the analysis of microarray data require both complex variance
modelling and computational efficiency to handle the larger numbers of random
effects. In this track we present an overview of the origins of mixed models,
through to more recent innovations with particular reference to the analysis of
genomics data and non-parametric regression.