A comment on two-locus epistatic interaction models for genome-wide association studies

Area of Science:

• Genetics and Bioinformatics
• Computational Biology
• Statistical Genomics

Background:

• Genome-wide association studies (GWAS) aim to identify genetic variants associated with diseases.
• Detecting epistatic interactions, where multiple genes influence a phenotype, presents significant computational challenges in GWAS.
• Existing algorithms often rely on simulated data and various interpretations of disease models for performance evaluation.

Purpose of the Study:

• To elucidate subtle differences in the interpretation of disease models used in previous studies of epistatic interactions in GWAS.
• To propose standardized reporting of disease model versions and interpretations in future research.
• To provide a method for facilitating the parameter setting of disease models.

Main Methods:

• Comparative analysis of existing disease models and their interpretations in the context of epistatic interaction detection algorithms.
• Development of guidelines for explicit reporting of disease model specifics in GWAS studies.
• Introduction of a parameter-setting facilitation tool for disease models.

Main Results:

• Identified and clarified subtle, yet significant, discrepancies in the application and interpretation of disease models across studies.
• Highlighted the need for consistent methodology in evaluating epistatic interaction detection algorithms.
• Demonstrated a practical approach to simplify and standardize disease model parameterization.

Conclusions:

• Standardizing the definition and application of disease models is crucial for reproducible and comparable results in epistatic interaction research within GWAS.
• Adoption of explicit reporting standards will enhance the reliability of algorithm performance evaluations.
• The proposed facilitation method aims to improve the efficiency and accuracy of setting up disease models for simulations.