A spatial probit model for fine-scale mapping of disease genes

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Linkage disequilibrium (LD) mapping is crucial for identifying disease susceptibility loci by exploiting correlations between genetic variants.
• Understanding LD structure is complex due to variations in recombination rates, mutation, and genetic drift.
• Case-control studies are widely used but require robust statistical methods for accurate genetic association analysis.

Purpose of the Study:

• To present a novel Bayesian multivariate probit model for LD mapping in case-control studies.
• To develop a method that flexibly accounts for local spatial correlation between genetic markers.
• To identify genomic regions with significant differences in marker frequencies between cases and controls.

Main Methods:

• A Bayesian multivariate probit model was developed to analyze LD.
• A retrospective model was employed to align with the case-control sampling scheme.
• Information-theoretic distance measures were used to quantify marker frequency differences.
• The model naturally handles unphased or missing genotype data.

Main Results:

• The proposed method successfully identified regions associated with disease status in simulated data.
• Application to real data from the CYP2D6 region demonstrated its utility.
• The Bayesian approach effectively accommodates complex LD patterns and data imperfections.

Conclusions:

• The novel Bayesian statistical method provides a flexible and robust approach for LD mapping in genetic association studies.
• This method enhances the ability to identify disease susceptibility loci, particularly in complex genomic regions.
• The approach is valuable for genetic research, including studies of pharmacogenomics like the CYP2D6 region.