A Novel Imputation Strategy Optimized for Low-Coverage Sequence Information

Advances in sequencing technology enabled the advent of low-cost, genome-wide genotyping platforms, such as Genotyping-By-Sequencing (GBS, Elshire et al. 2011). However, the resulting data often has high rates of missingness and heterozygote undercalling, which result in a loss of power and potentially accuracy for downstream applications such as genome-wide association studies or whole-genome prediction. Published imputation methods (eg. Beagle v.4 (Browning and Browning 2013)) are capable of highly accurate imputation when supplied with appropriate haplotype information but these algorithms are designed for obligate outcrossing taxa, such as humans, and do not leverage inbred information commonly available in crop plants well. I present a method for imputation that is fast, reasonably accurate and takes advantage of sparse, inbred genotypic data typical of breeding programs and compare results with Beagle v.4.