AUTHOR=Santana Bruna , Palti Yniv , Gao Guangtu , Tripathi Vibha , Martin Kyle E. , Fragomeni Breno O. 

TITLE=Genome-wide association analysis of resistance to bacterial cold-water disease in an important rainbow trout aquaculture breeding population

JOURNAL=Frontiers in Genetics

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2025.1582138

DOI=10.3389/fgene.2025.1582138

ISSN=1664-8021

ABSTRACT=Bacterial cold-water disease (BCWD) outbreaks in salmonid aquaculture have resulted in significant losses in commercial populations. Currently, there is no commercially available vaccine for the disease caused by Flavobacterium psychrophilum. BCWD resistance in rainbow trout exhibits moderate heritability and has been the focus of selection efforts. The understanding of key genomic regions associated with BCWD resistance has advanced since the integration of genomic information into genetic evaluations, proving successful in enhancing BCWD resistance in some commercial lines. Here, we report the results of a genome-wide association study for BCWD resistance in an important commercial rainbow trout line to further our understanding of the genetic architecture of the trait and infer a selective breeding strategy for this line. Different scenarios were tested, including the use of all single-nucleotide polymorphisms (SNPs) passing quality control, removal of SNPs with major effect, elimination of consistent “major SNPs” in subgroups of the population, and exclusion of SNPs within haplotypes with major effect. Prediction accuracy was evaluated with different SNP weighting strategies, utilizing cross-validation groups formed either randomly or based on principal components and cluster analyses of genotypic data. Comparative analysis of cross-validation methods suggested that partitioning of the dataset using K-means clustering reduced overfitting. The incorporation of SNP weighting further confirmed the oligogenic nature of the trait under investigation. Prediction accuracy with pedigree-based best linear unbiased prediction (PBLUP) was 0.27 and increased to 0.36 with genomic information. The accuracy obtained with a single largest effect haplotype was 0.23. Moreover, a decrease in accuracy was observed upon excluding major SNPs and haplotypes, providing supplementary evidence of their importance on phenotypes. The two largest association peaks on OmyA31/Omy25 and Omy8 were consistent with previous reports.