Meta-analysis of genome-wide association from genomic prediction models

Area of Science:

• Animal Breeding
• Quantitative Genetics
• Genomics

Background:

• Genome-wide association (GWA) studies using genomic best linear unbiased prediction (GBLUP) are standard in animal breeding.
• GWA studies often have limited power to detect rare variants due to small effect sizes and sample size constraints in animal populations.
• Meta-analysis (MA) is widely used in human genetics but less implemented in animal breeding for GWA studies.

Purpose of the Study:

• To present methods for implementing a meta-analysis (MA) of genome-wide association (GWA) studies in animal breeding.
• To describe how to compute weights from multiple genomic evaluations using animal-centric GBLUP models for MA.
• To demonstrate the advantages of MA over population-level GWA for detecting genetic associations.

Main Methods:

• Developed methods for meta-analysis (MA) of genome-wide association (GWA) studies.
• Defined a proper approach to compute weights from multiple genomic evaluations based on GBLUP models.
• Scripts were created to account for association strength, sign, and heterogeneity in association phase.

Main Results:

• Meta-analysis (MA) significantly increases the power to detect genetic associations compared to population-level GWA.
• MA effectively accounts for population structure and heterogeneity of variance components across different populations.
• MA does not require access to raw genotype data, unlike joint analysis methods.

Conclusions:

• Meta-analysis (MA) of GWA studies is a powerful alternative for summarizing results from multiple genomic studies in animal breeding.
• MA overcomes limitations related to sample size, genotype data sharing, fixed effects definition, and trait measurement scales.
• This approach enhances the detection of quantitative trait loci (QTL) and rare variants, advancing genomic selection.