CAVIAR

CAVIAR models multiple causal genetic variants at GWAS loci to estimate per-variant posterior probabilities and identify high-confidence sets of causal variants for mapping complex trait associations.

Key Features:

• Flexible Causal Variant Estimation: Allows an arbitrary number of causal variants at each locus instead of assuming a single causal variant, enabling more accurate posterior probability estimation.
• Improved Identification Accuracy: Accommodating multiple causal variants yields improved identification accuracy, with simulations reporting a 20–50% improvement over traditional methods at loci with multiple causal variants.
• High Confidence Predictions: Predicts a set of variants per locus that, under stated assumptions, is intended to include all true causal variants with high confidence (e.g., 95%).
• Empirical Validation: Validated using empirical expression quantitative trait loci (eQTL) data from the CHI3L2 gene, identifying causal variants affecting gene expression at that locus.

Scientific Applications:

• GWAS fine-mapping: Refines genome-wide association study signals to prioritize likely causal variants for downstream biological interpretation.
• eQTL analysis: Identifies causal variants affecting gene expression in eQTL studies, as demonstrated for CHI3L2.
• Translational genetics: Informs investigations of disease etiology, potential therapeutic targets, and personalized medicine by providing more accurate causal variant sets.

Methodology:

Statistical framework that allows an arbitrary number of causal variants per locus, estimates posterior probabilities of causality, constructs locus-level variant sets intended to include all true causal variants with specified confidence (e.g., 95%), and was validated on empirical eQTL data from CHI3L2.