Genomic Prediction Using LD-Based Haplotypes in Combined Pig Populations

Area of Science:

• Animal Breeding and Genetics
• Genomics
• Quantitative Genetics

Background:

• Reference population size is crucial for genomic prediction accuracy.
• Combining multiple populations can enhance genomic prediction but is often suboptimal in pigs due to differing linkage disequilibrium (LD) patterns.
• Imputed whole-genome sequencing (WGS) data offers potential for improved genomic prediction strategies.

Purpose of the Study:

• To investigate the impact of SNP density, variant representation (SNPs vs. haplotypes), and reference population size on genomic prediction accuracy in combined pig populations.
• To evaluate the effectiveness of LD-based haplotypes constructed from WGS data for genomic prediction in multi-population scenarios.
• To identify optimal strategies for improving genomic prediction of reproduction traits in pigs.

Main Methods:

• Utilized imputed whole-genome sequencing (WGS) data to construct LD-based haplotypes.
• Compared genomic prediction accuracy using single nucleotide polymorphisms (SNPs) versus haplotype alleles.
• Assessed the influence of varying SNP densities and reference population sizes on prediction ability.
• Employed genomic best linear unbiased prediction (GBLUP) and haplotype-based methods for multi-population genomic evaluation.

Main Results:

• Genomic best linear unbiased prediction (GBLUP) with WGS data improved multi-population prediction accuracy but not within-population accuracy.
• Haplotype-based methods using 80K chip data and GBLUP showed higher multi-population prediction accuracy (3.4-5.9%) compared to within-population (1.2-4.3%).
• Haplotype methods based on WGS data in multi-population settings demonstrated superior genomic prediction performance.
• Optimal variable selection for reproduction traits in Yorkshire pigs was achieved using haploblocks constructed with a low LD threshold (r² = 0.2-0.3).

Conclusions:

• Both haplotype methods (chip data) and GBLUP (WGS data) benefit multi-population genomic prediction in pigs.
• Combining haplotype methods with WGS data represents a superior strategy for multi-population genomic evaluation.
• The study highlights the importance of LD patterns and data types for effective genomic prediction in diverse pig populations.