SparsePro

SparsePro performs genome-wide fine-mapping by integrating GWAS summary statistics with functional annotations to identify causal variants despite linkage disequilibrium and multiple causal variants per locus.

Key Features:

• Integration of Functional Annotations: Incorporates functional annotations via joint estimation of enrichment weights to derive functionally-informed priors from summary statistics.
• Efficient Computational Approach: Extends the iterative Bayesian stepwise selection algorithm from SuSiE with a sparse projection technique and paired mean field variational inference to improve computational efficiency.
• Hyperparameter Estimation and Posterior Summaries: Implements strategies for estimating hyperparameters and summarizing posterior probabilities for fine-mapping results.
• Performance Evaluation: Evaluated through extensive simulations using UK Biobank resources, demonstrating improved fine-mapping power and reduced computation time relative to state-of-the-art methods.
• Real-World Application: Applied to fine-mapping five functional biomarkers associated with clinically relevant phenotypes.

Scientific Applications:

• Genetic architecture analysis: Identifies causal variants in GWAS loci to inform studies of complex traits.
• Functional follow-up and target prioritization: Integrates functional annotations to prioritize variants for downstream functional studies and therapeutic target identification.

Methodology:

Extends SuSiE by applying sparse projection, paired mean field variational inference, and joint enrichment-weight estimation to derive functionally-informed priors, together with hyperparameter estimation and posterior summarization.