Fine-mapping methods for complex traits: essential adaptations for samples of related individuals

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genomics

Background:

• Fine-mapping causal variants from genome-wide association studies (GWAS) is challenging in related populations due to violated assumptions of standard methods.
• Existing tools often exhibit poor accuracy in livestock and other related populations, hindering genetic improvement.

Purpose of the Study:

• To introduce a comprehensive Bayesian framework for accurate fine-mapping in populations with complex relatedness.
• To develop novel methods adaptable for both individual-level data and summary statistics from GWAS.

Main Methods:

• Developed BFMAP-Shotgun Stochastic Search using linear mixed models (LMM) and simulated annealing for individual data.
• Introduced FINEMAP-adj and SuSiE-adj for summary statistics, utilizing LMM-derived inputs and relatedness-adjusted linkage disequilibrium.
• Proposed gene-level posterior inclusion probability (PIPgene) to aggregate variant signals and enhance detection power.

Main Results:

• Simulations showed substantial outperformance of new methods compared to existing tools (FINEMAP, SuSiE, GCTA-COJO) in related individuals.
• Multi-breed populations significantly enhanced fine-mapping accuracy over single-breed populations.
• PIPgene markedly improved candidate gene identification, validated by functional enrichment analysis in Duroc pigs.

Conclusions:

• The developed Bayesian framework and associated software provide robust and validated methods for accurate fine-mapping in populations with complex relatedness.
• These methods significantly improve the identification of biologically relevant variants and candidate genes, with practical utility in livestock breeding.