Genome-enabled prediction of quantitative traits in chickens using genomic annotation

Area of Science:

• Animal genetics
• Quantitative genetics
• Genomic prediction

Background:

• Genome-wide association studies (GWAS) identify genetic variants for complex traits.
• Previous research shows trait-associated SNPs enrich functional regions and deplete intergenic regions (IGR).
• Systematic examination of genomic region impact on predictive ability for complex phenotypes is lacking.

Purpose of the Study:

• To partition SNPs by annotation to characterize genomic regions with varying predictive power for three broiler chicken traits.
• To evaluate the predictive ability of different genomic regions using a whole-genome approach.
• To compare the predictive performance of genic regions versus intergenic regions (IGR) for specific traits.

Main Methods:

• Partitioning single nucleotide polymorphisms (SNPs) based on genomic annotation.
• Constructing additive genomic relationship kernels for genic regions.
• Employing kernel-based Bayesian ridge regression for genomic prediction.

Main Results:

• Predictive performance for breast meat ultrasound area was better using SNPs in genic regions compared to IGR.
• Intergenic regions (IGR) tagged by SNPs showed higher predictive ability for body weight and egg production than genic regions.
• The predictive ability using all markers (whole-genome) approached the best prediction achieved by a single genomic region.

Conclusions:

• Genomic regions exhibit differential predictive abilities for complex traits.
• Whole-genome regression methods, utilizing all quality-filtered SNPs, are effective for predicting complex traits.
• The whole-genome approach is a promising tool for genomic prediction in livestock, despite observed regional variations.