SparsePro: An efficient fine-mapping method integrating summary statistics and functional annotations.

Area of Science:

• Genetics and Genomics
• Computational Biology
• Statistical Genetics

Background:

• Identifying causal variants in genome-wide association studies (GWAS) is difficult due to linkage disequilibrium (LD) and multiple causal variants per locus.
• Functional genomic annotations can prioritize biologically relevant variants, enhancing GWAS fine-mapping.
• Existing fine-mapping methods face computational challenges with complex genetic architectures and LD patterns.

Purpose of the Study:

• To develop an efficient fine-mapping method integrating summary statistics and functional genomic annotations.
• To improve the power and reduce computational cost of identifying causal variants.
• To provide functionally-informed priors for more accurate fine-mapping.

Main Methods:

• Developed SparsePro by connecting the SuSiE algorithm with a mean field variational inference algorithm via sparse projection.
• Implemented strategies for hyperparameter estimation and posterior probability summarization.
• Incorporated functional annotations by jointly estimating enrichment weights for priors.

Main Results:

• SparsePro demonstrated improved fine-mapping power compared to state-of-the-art methods in simulations using UK Biobank data.
• The method achieved reduced computation time.
• Successful application of SparsePro to fine-map five functional biomarkers for clinically relevant phenotypes.

Conclusions:

• SparsePro is an efficient and effective method for fine-mapping genetic variants by integrating summary statistics and functional annotations.
• The developed method has broad utility for understanding complex trait genetics.
• SparsePro can increase the yield of functional follow-up studies for GWAS.