Disease association tests by inferring ancestral haplotypes using a hidden markov model

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) commonly use single nucleotide polymorphism (SNP) analysis, but stringent thresholds miss true signals.
• Haplotype-based methods offer higher power but face challenges with rare haplotypes and block definition.
• Improved statistical methods are needed to increase power in genetic association studies while controlling for false positives.

Purpose of the Study:

• To develop a novel statistical method, Ancestral Haplotype Clustering (AncesHC), to enhance the power of genetic association studies.
• To address limitations of traditional SNP and existing haplotype-based methods.
• To provide a flexible tool applicable to various marker types and ploidies, handling missing data.

Main Methods:

• Developed the AncesHC method utilizing a Hidden Markov Model (HMM) for ancestral haplotype clustering.
• Clustered haplotypes into groups of common ancestral origin without assuming rigid block structures.
• Tested association by comparing case and control counts within each cluster (0, 1, or 2 chromosomes).
• Validated the method through simulations of case-control status across diverse disease models in mice.

Main Results:

• AncesHC demonstrated substantially greater power than single-SNP analyses in detecting disease associations.
• The method showed increased power compared to the CLADHC cladistic haplotype clustering approach.
• AncesHC effectively handles various marker types (biallelic/multiallelic) and ploidies, including missing genotypes.

Conclusions:

• AncesHC provides a powerful and flexible approach for genetic association studies, improving disease gene discovery.
• The method overcomes key statistical challenges associated with traditional SNP and haplotype analyses.
• AncesHC offers a valuable tool for researchers seeking to identify causal loci with improved statistical power.