finemapping

Finemapping is a statistical approach used in genetics to identify the specific genetic variants within a genomic region that are most likely to causally influence a trait or disease. It typically follows genome-wide association studies (GWAS) that have localized an association to a region but cannot pinpoint the exact causal variant due to linkage disequilibrium (LD).

Inputs for finemapping usually include GWAS summary statistics (effect sizes and standard errors), an LD reference panel that matches the study population, and sometimes functional annotations. Methods are primarily Bayesian or frequentist. Bayesian finemapping computes the posterior inclusion probability (PIP) for each variant and often produces a credible set—a group of variants that together contain a specified probability (e.g., 95%) of containing the true causal variant. Some approaches allow for more than one causal variant per locus. Popular tools include FINEMAP, SuSiE (Sum of Single Effects), CAVIAR, and PAINTOR, which can incorporate annotations to improve resolution.

Outcomes of finemapping are prioritized lists or credible sets of candidate causal variants, guiding downstream functional validation and experimental studies. It also aids interpretation by clarifying which variants within a GWAS locus are most plausible drivers of the association, and it can be extended to multi-ancestry analyses or integrated with eQTL and other functional data to assess colocalization and regulatory potential.

Limitations include sensitivity to LD accuracy and the chosen model, potential multiple causal variants that are difficult to distinguish, and reliance on large, well-matched datasets. Results are probabilistic and suggest candidates rather than definitive proofs of causality.