GWAS of Bacterial Cold Water Disease Resistance in Rainbow Trout Reveals the Potential for a Combined Approach Between Genomic Selection and Marker Assisted Selection

Genomic selection (GS) incorporates dense SNP marker genotypes with phenotypic data from related animals to predict animal-specific genomic breeding value (GEBV). Marker assisted selection (MAS) uses DNA markers linked to positive alleles at a single locus with a major effect on the total genetic variation for the trait. The genotyping cost associated with the application of MAS is lower than GS, but to date, it has only been effective for traits with single major-effect genes such as IPNV resistance in Atlantic salmon. Here, using a commercial rainbow trout breeding population, we conducted genome-wide association study (GWAS) to identify loci that account for at least 1% of the genetic variation for bacterial cold water disease (BCWD) resistance in a laboratory disease-challenge model. Combining the results from two GWAS models (weighted ssGBLUP and BayesB) we identified seven chromosomal regions with moderate to major effect on four different chromosomes (Omy03, 05, 08 and 25). We then evaluated the accuracy of GEBV predictions comparing eight SNP array configurations composed of 45K, 10K, 3K, 1K, 500, 300, 200, and a panel of 70 SNPs flanking the moderate-major QTL identified by GWAS. As expected, reducing the SNP density reduced the accuracy of the genomic predictions, but the panel of the 70 QTL-flanking SNPs had an accuracy of 0.66, which was only slightly lower than the accuracy of 0.71 from the full dataset of ~45K SNPs and was equal to the accuracy from the 10K SNPs configuration. Using a panel of only 70 QTL-flanking SNPs, we doubled the accuracy of the pedigree-based BLUP model predictions; however, further evaluation is required as we are currently uncertain of the extent of linkage disequilibrium between these SNPs and the BCWD resistance QTL from one generation to the next.