Quality and value of imputing gene tests for all animals


  • Jeff O'Connell University of Maryland School of Medicine
  • Paul VanRaden USDA-ARS Animal Genomics and Improvement Laboratory


Genomic selection is driven by genotyping arrays designed for uniform coverage of the genome because most quantitative trait loci (QTLs) underlying the heritability of the trait are unknown. Laboratories have improved the arrays since 2014 with custom content by adding selected QTLs discovered from whole-genome sequencing (WGS) and high-effect markers from higher-density arrays. Breed differences, missing data rates, and error rates were investigated for 8 QTL gene tests currently imputed for all genotyped animals of 5 breeds plus crossbreds. Gene content for each gene test was predicted for non-genotyped relatives using mixed model methods like those used in single-step genomic evaluations, allowing potential direct selection across all animals. For the 8 QTL studied, Mendel error rates were low except for polled in Jerseys and DGAT1 in most breeds. Allele effects for DGAT1 were smaller than 2 nearby flanking SNPs because DGAT1 was genotype quality was poor on several arrays. For yield traits, 79K predictions including selected markers and QTLs had 1-2% higher reliability than 45K or 35K predictions excluding those SNPs.