Computer simulation of marker-assisted selection utilizing linkage disequilibrium.
The face of animal breeding has changed significantly over the past few decades. Traditionally, the genetic improvement of both plant and animal species focussed on the selective breeding of individuals with superior phenotypes, with no precise knowledge of the genes controlling the traits under selection. Over the past few decades, however, advances in molecular genetics have led to the identification of genetic markers associated with genes controlling economically important traits, which has enabled breeders to enhance the genetic improvement of breeding stock through linkage disequilibrium marker-assisted selection. Since the integration of marker-assisted selection into breeding programmes has not been widely documented, it is important that breeders are able to evaluate the advantages and disadvantages of marker-assisted selection, in comparison to phenotypic selection, prior to the implementation of either selection strategy. Therefore, this investigation aimed to develop deterministic simulation models that could accurately demonstrate and compare the effects of phenotypic selection and marker-assisted selection, under the assumption of both additive gene action and complete dominance at the loci of interest. Six computer models were developed using Microsoft Excel, namely 'Random Mating,' 'Phenotypic Selection,' 'Marker-Assisted Selection,' 'Selection with Dominance,' 'Direct Selection' and 'Indirect selection.' The 'Random Mating' model was firstly used to determine the effects of linkage disequilibrium between two loci in a randomly mating population. The 'Phenotypic Selection' and 'Marker-Assisted Selection' models focused primarily on examining and comparing the response to these two selection strategies over five generations and their consequent effect on genetic variation in a population when the QTL of interest exhibited additive gene action. In contrast, the 'Selection with Dominance' model investigated the efficiency of phenotypic selection and marker-assisted selection under the assumption of complete dominance at the QTL under selection. Finally, the 'Direct Selection and 'Indirect Selection' models were developed in order to mimic the effects of marker assisted selection on two cattle populations utilizing both a direct and indirect marker respectively. The simulated results showed that, under the assumption of additive gene action, marker-assisted selection was more effective than phenotypic selection in increasing the population mean, when linkage disequilibrium was present between the marker locus and the QTL under selection and the QTL captured more than 80% of the trait variance. The response to both selection strategies was shown to decrease over five generations due to the decrease in genetic variation associated with selection. When the QTL under selection was assumed to display complete dominance, however, marker-assisted selection was markedly more effective than phenotypic selection, even when a minimal amount of linkage disequilibrium was present in the population and the QTL captured only 60% of the trait variance. The results obtained in this investigation were successful in simulating the theoretical expectations of markerassisted selection. The computer models developed in this investigation have potential applications in both the research and agricultural sectors. For example, the successful application of a model developed in this investigation to a practical situation that simulated markerassisted selection, was demonstrated using data from two Holstein cattle populations. Furthermore, the computer models that have been developed may be used in education for the enhancement of students understanding of abstract genetics concepts such as linkage disequilibrium and marker-assisted selection.