Development of Robust Methods for Genomic Selection

Abstract

Genomic selection is an advance method of breeding where genome-wide dense markers information is used to predict genetic merit of an individual in a breeding programme. In today’s scenario genomic selection is a promising tool for improving genetic gain of individuals under study. Genomic selection is a form of marker-assisted selection in which genetic markers covering the whole genome are used to identify QTL which are in LD with at least one marker. Genomic selection predicts the breeding values of lines in a population by analyzing their phenotypes and high-density marker scores. Genomic selection process starts with building a statistical model from individuals having both genotypic and phenotypic data (i.e. training set), this model is further used for estimation GEBVs for individuals having only genotypic information. Individuals are then ranked on the basis of GEBVs and subsequently superior individuals are selected. Genomic selection procedures have proven useful in estimating breeding values and predicting phenotype with genome-wide molecular marker information. Number of parametric methods has been proposed to predict individual breeding values by modeling the relationship between individual genotype and phenotype. Normally in the method of estimation of breeding value for genomic selection, assumptions of the model (e.g. Normality, linearity, independence of explanatory variables) are violated which may provide false breeding value. It is also noticed that parametric methods only performs satisfactory when system under study have additive genetic architecture. However, some nonparametric methods have been developed to capture non additive (dominance & epistasis) variances, but these are generally fails to capture additive variances. The idea behind this study is to identify best suitable model from parametric and nonparametric model under different genetic architecture. First we have gone through various parametric models and evaluated there performance on real and simulated datasets. Under parametric models we have studied most commonly used methods i.e. linear regression, RR, BLUP, GBLUP, LASSO, Bayesian. For nonparametric models we have studied models like RKHS, SVM, NN and RF. It was observed from the result that for additive architecture, GBLUP performed quite well and among nonparametric methods, performance of SVM was found to be encouraging. Keeping these results in the mind, a robust model has been developed for genomic selection studies which can handle additive and epistatic effects simultaneously by minimizing their error variance. Developed integrated model has been evaluated using the prediction accuracy and error variance. It has been found that our proposed model is either performing better or at par with the existing models. It has also been observed that our proposed model is robust to the diverse genetic architecture i.e. additive and epistatic. Apart from this, impact of outlier on genomic prediction accuracy has also been explored. In this study, a new efficient method using meta-analysis for outlier detection in genomic data has been proposed. It has been shown that by implementing efficient diagnostic measure for outlier detection, accuracy of genomic selection model can be improved. Comparative study has been made among various existing methods of outlier detection in high dimensional genomic data for their impact on accuracy of genomic estimated breeding value. It has been observed that our proposed method has outperformed among existing methods.