Supporting data for The Fate of Deleterious Variants in a Barley Genomic Prediction Population

Abstract

Targeted identification and purging of segregating deleterious genetic variants has been proposed as a novel approach to plant breeding. This approach is motivated in part by the observation that demographic events and strong selection associated with cultivated species pose a “cost of domestication.” This includes an increase in the proportion of genetic variants at phylogenetically-constrained sites where a mutation is likely to reduce fitness. Recent advances in DNA resequencing technology and sequence constraint-based approaches to predict the functional impact of a mutation now permit the identification of putatively deleterious SNPs (dSNPs) on a genome-wide scale. Using exome capture resequencing of 21 barley 6-row spring breeding lines, we identify 3,855 dSNPs among 497,754 total SNPs. The dSNPs are more frequent in portions of the genome with a higher recombination rate, as measured by cM/Mb, than in pericentromeric regions with lower recombination rate and gene density. Using 5,215 progeny from a genomic prediction experiment, we examine the fate of dSNPs over three breeding cycles. Average derived allele frequency is lower for dSNPs than any other class of variants. Adjusting for frequency, derived alleles at dSNPs reduce in frequency or are lost more frequently than other classes of SNPs. Using a linear mixed model applied to 677 lines phenotyped at 5 year-locations, we find that a genomic region with the strongest association with a fungal disease resistance trait that was selected for in the population also negatively impacts yield. Finally, the highest yielding lines in the experiment, as chosen by standard genomic prediction approaches, carry fewer homozygous dSNPs than randomly selected lines.