Inference from genome-wide association studies using a novel Markov model.

Area of Science:

• Genetics
• Statistical Genomics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) are crucial for identifying genetic variants associated with diseases.
• Traditional methods often struggle to pinpoint causal loci accurately, especially with complex genetic architectures.
• Bayesian modeling offers a powerful framework for integrating complex data and uncertainty in genetic analyses.

Purpose of the Study:

• To propose a novel Bayesian modeling approach for analyzing single nucleotide polymorphism (SNP) data in GWAS.
• To develop a flexible and computationally feasible model for accurate causal locus identification.
• To evaluate the model's performance on both synthetic and real disease phenotype datasets.

Main Methods:

• Developed a 'latent seed model' integrating k-means clustering, hidden Markov models (HMMs), and logistic regression within a fully Bayesian framework.
• Employed Markov chain Monte Carlo (MCMC) simulation with Metropolis-Hastings updates for model fitting.
• Utilized fast algorithms for HMMs to ensure computational feasibility and model extensibility.

Main Results:

• The latent seed model successfully identified causal genetic loci in analyzed datasets.
• Demonstrated superior performance compared to single SNP analysis, particularly in challenging cases.
• Showcased promising results on datasets with both synthetic and real disease phenotypes and SNP data.

Conclusions:

• The proposed Bayesian latent seed model provides a robust and flexible approach for GWAS analysis.
• The method effectively pinpoints causal loci, offering advancements over existing techniques.
• This approach holds significant potential for improving our understanding of genetic disease associations.