BARS

BARS applies Bayesian Adaptive Regression Splines to integrate single-locus and haplotype-based association statistics for detecting liability alleles and capturing linkage disequilibrium signals across genomic regions.

Key Features:

• Bridging Methodologies: Integrates single-locus and haplotype-based tests using Bayesian Adaptive Regression Splines to analyze common genomic regions.
• Non-Parametric Approach: Employs non-parametric regression via Bayesian Adaptive Regression Splines to model smooth genotype-phenotype association curves consistent with the data.
• Robust Signal Detection: Maintains specified test size by showing no signal when no liability allele is present and detecting signals when a liability allele exists, yielding high sensitivity and specificity.
• Reduction of Multiple Testing Issues: Combines single-locus association statistics into a unified analysis to diminish the multiple testing burden inherent in classical tests.
• Versatility Across Data Types: Applies to various data types, including genotype frequencies estimated from pooled samples.

Scientific Applications:

• Association scans over large genomic regions: Performs association scans across large genomic regions to capture signals encoded in linkage disequilibrium.
• Identification of genetic determinants of complex diseases: Enhances detection of liability alleles that contribute to complex disease phenotypes.
• Pooled-sample analysis: Enables analysis of studies that use pooled genotype frequency estimates.

Methodology:

Integrates single-locus association statistics within a common genomic region using Bayesian Adaptive Regression Splines and non-parametric regression to fit smooth curves consistent with the data, enabling detection of liability alleles while controlling test size.