Investigations on pleiotropy and genome wide association analyses by random effects using QTL-MAS 2010 dataset

Abstract

Recent advances in molecular genetics have provided hundreds of thousands of single nucleotide polymorphisms to detect mutations at the vicinity of genes related with quantitative traits. Breeding values could be used as response variable in mixed model framework to detect possible associations with genomic relationship matrix. It is known that most of quantitative traits are correlated which leads to construct of networks and pathways of genes due to pleiotropy. Hence the main aim of this paper is to a) detecting pleiotropy by principal component analyses methods b) prediction of genomic breeding values by ridge regression c) detecting associations based on predicted genomic breeding values obtained from b) using QTLMAS 2010 simulated dataset. Most of the Quantitative Trait Locus (QTLs) were located at chromosome 1 and 3. Highest correlation between true breeding value and predicted breeding value were obtained by Gaussian Kernel function as 0.557. To detect pleiotropy we used first and second principal components as response variable and success rates found to be 0.2727 and 0.1714 and error rates found to be 0.5952 to 0.6400 for first two principal component loadings respectively. Using genomic breeding values as response variable gave better success rate and lower error rate compared with when using raw phenotypes. We found that using the most heritable and variable component (first component) had higher change to detect pleiotropic genes using QTLMAS-2010 dataset