STS-BN: An efficient Bayesian network method for detecting causal SNPs.

Area of Science:

• Genetics and Genomics
• Computational Biology
• Biostatistics

Background:

• Identifying causal single nucleotide polymorphisms (SNPs) is crucial for understanding complex disease pathogenesis, prevention, diagnosis, and treatment.
• Current methods for large-scale genome-wide association studies (GWAS) often lack sufficient accuracy in causal SNP detection.
• There is a significant need for more powerful and accurate computational methods for identifying disease-associated SNPs.

Purpose of the Study:

• To develop and evaluate a novel, accurate method for identifying causal SNPs associated with complex diseases using large-scale GWAS data.
• To improve upon the accuracy and stability of existing methods for SNP detection in case-control study designs.

Main Methods:

• A score-based, two-stage Bayesian network approach was developed to learn the structure of a disease-centered local Bayesian network.
• The method integrates principles from constraint-based and score-and-search algorithms for network structure learning.
• Performance was assessed through simulation experiments and application to real-world genome-wide association data.

Main Results:

• The proposed Bayesian network method demonstrated improved accuracy and stability compared to several common alternative methods.
• The algorithm achieved lower false-positive rates while successfully detecting all correct causal loci.
• Application to real-world GWAS data confirmed the algorithm's robust performance in handling large datasets.

Conclusions:

• The developed score-based two-stage Bayesian network method is effective for identifying SNPs related to complex diseases.
• This novel approach offers superior accuracy compared to existing methods applicable to large-scale genome-wide association studies.