Accurate genomic predictions for BCWD resistance in rainbow trout are achieved using low-density SNP panels: Evidence that long-range LD is a major contributing factor

Roger L Vallejo; Rafael M O Silva; Jason P Evenhuis; Guangtu Gao; Sixin Liu; James E Parsons; Kyle E Martin; Gregory D Wiens; Daniela A L Lourenco; Timothy D Leeds; Yniv Palti

doi:10.1111/jbg.12335

Accurate genomic predictions for BCWD resistance in rainbow trout are achieved using low-density SNP panels: Evidence that long-range LD is a major contributing factor

J Anim Breed Genet. 2018 Jun 5. doi: 10.1111/jbg.12335. Online ahead of print.

Affiliations

¹ National Center for Cool and Cold Water Aquaculture, United States Department of Agriculture, Agricultural Research Service, Kearneysville, West Virginia.
² Animal and Dairy Science Department, University of Georgia, Athens, Georgia.
³ Troutlodge, Inc, Sumner, Washington.

PMID: 29869355
DOI: 10.1111/jbg.12335

Abstract

Previously accurate genomic predictions for Bacterial cold water disease (BCWD) resistance in rainbow trout were obtained using a medium-density single nucleotide polymorphism (SNP) array. Here, the impact of lower-density SNP panels on the accuracy of genomic predictions was investigated in a commercial rainbow trout breeding population. Using progeny performance data, the accuracy of genomic breeding values (GEBV) using 35K, 10K, 3K, 1K, 500, 300 and 200 SNP panels as well as a panel with 70 quantitative trait loci (QTL)-flanking SNP was compared. The GEBVs were estimated using the Bayesian method BayesB, single-step GBLUP (ssGBLUP) and weighted ssGBLUP (wssGBLUP). The accuracy of GEBVs remained high despite the sharp reductions in SNP density, and even with 500 SNP accuracy was higher than the pedigree-based prediction (0.50-0.56 versus 0.36). Furthermore, the prediction accuracy with the 70 QTL-flanking SNP (0.65-0.72) was similar to the panel with 35K SNP (0.65-0.71). Genomewide linkage disequilibrium (LD) analysis revealed strong LD (r² ≥ 0.25) spanning on average over 1 Mb across the rainbow trout genome. This long-range LD likely contributed to the accurate genomic predictions with the low-density SNP panels. Population structure analysis supported the hypothesis that long-range LD in this population may be caused by admixture. Results suggest that lower-cost, low-density SNP panels can be used for implementing genomic selection for BCWD resistance in rainbow trout breeding programs.

Keywords: bacterial cold water disease; genomic breeding value; genomic selection; linkage disequilibrium; progeny performance; reduced SNP array; salmonids.