Genome‑wide association study and genomic prediction for growth traits in yellow-plumage chicken using genotyping-by-sequencing

Abstract

Background Growth traits are of great importance for poultry breeding and production and have been the topic of extensive investigation, with many quantitative trait loci (QTL) detected. However, due to their complex genetic background, few causative genes have been confirmed and the underlying molecular mechanisms remain unclear, thus limiting our understanding of QTL and their potential use for the genetic improvement of poultry. Therefore, deciphering the genetic architecture is a promising avenue for optimising genomic prediction strategies and exploiting genomic information for commercial breeding. The objectives of this study were to: (1) conduct a genome-wide association study to identify key genetic factors and explore the polygenicity of chicken growth traits; (2) investigate the efficiency of genomic prediction in broilers; and (3) evaluate genomic predictions that harness genomic features. Results We identified five significant QTL, including one on chromosome 4 with major effects and four on chromosomes 1, 2, 17, and 27 with minor effects, accounting for 14.5 to 34.1% and 0.2 to 2.6% of the genomic additive genetic variance, respectively, and 23.3 to 46.7% and 0.6 to 4.5% of the observed predictive accuracy of breeding values, respectively. Further analysis showed that the QTL with minor effects collectively had a considerable influence, reflecting the polygenicity of the genetic background. The accuracy of genomic best linear unbiased predictions (BLUP) was improved by 22.0 to 70.3% compared to that of the conventional pedigree-based BLUP model. The genomic feature BLUP model further improved the observed prediction accuracy by 13.8 to 15.2% compared to the genomic BLUP model. Conclusions A major QTL and four minor QTL were identified for growth traits; the remaining variance was due to QTL effects that were too small to be detected. The genomic BLUP and genomic feature BLUP models yielded considerably higher prediction accuracy compared to the pedigree-based BLUP model. This study revealed the polygenicity of growth traits in yellow-plumage chickens and demonstrated that the predictive ability can be greatly improved by using genomic information and related features.

Genomic prediction with whole-genome sequence data in intensely selected pig lines

“…One such method is BayesRC [ 21 ], which extends BayesR by estimating the proportion of variants effects that are drawn from each normal distribution separately for each of several classes of variants defined based on a priori biological information or other criteria [ 17 , 20 ]. Similarly, genomic feature BLUP (GFBLUP) [ 77 ] could be used to incorporate prior biological information from either QTL databases or GWAS as genomic features [ 19 , 34 , 65 ]. The multi-breed multi-genomic relationship matrices genomic prediction model (MBMG) [ 26 ], which fits two genomic relationship matrices according to prior biological information, has also been proposed for multi-breed scenarios to improve genomic prediction in small populations.…”

“…Nevertheless, these examples indicate that identifying causal variants from WGS data could enhance genomic prediction. Whole-genome sequence data has already been applied in genome-wide association studies (GWAS) to identify variants associated with a variety of traits in livestock [ 2 , 32 – 34 ], including pigs [ 35 , 36 ]. However, the fine-mapping of causal variants remains challenging due to the pervasive long-range linkage disequilibrium across extremely dense variants [ 37 ].…”

Genomic prediction with whole-genome sequence data in intensely selected pig lines

“…One of such methods is BayesRC [21], which extends BayesR by assigning flatter prior distributions to classes of variants that are more likely to be causal [17,20]. Similarly, GFBLUP [77] could be used to incorporate prior biological information from either QTL databases or GWAS as genomic features [19,34,65]. The model MBMG [26], which fits two genomic relationship matrices according to prior biological information, has also been proposed for multi-breed scenarios to improve genomic prediction in small populations.…”

“…Nevertheless, this shows how identifying causal variants could enhance genomic prediction with WGS. Wholegenome sequence data has already been applied in genome-wide association studies (GWAS) to identify variants associated to a variety of traits in livestock [2, [33][34][35],…”

“…Nevertheless, these examples indicate how identifying causal variants could enhance genomic prediction with WGS. Whole-genome sequence data has already been applied in genome-wide association studies (GWAS) to identify variants associated to a variety of traits in livestock [2, [32][33][34], including pigs [35,36]. However, the fine-mapping of causal variants remains challenging due to the pervasive long-range linkage disequilibrium across extremely dense variation [37].…”