To control false positives in gene-gene interaction analysis: two novel conditional entropy-based approaches

Area of Science:

• Genetics
• Statistical genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) often miss gene-gene interactions, limiting the identification of genetic disease components.
• Existing model-free methods for detecting genetic interactions risk inflated false positive rates, especially with strong individual gene effects.

Purpose of the Study:

• To develop novel statistical metrics for robust detection of gene-gene interactions.
• To address the limitations of existing methods, particularly the issue of inflated false positive rates.

Main Methods:

• Proposed two novel conditional entropy-based metrics for assessing gene-gene interactions.
• Conducted extensive simulations to evaluate false positive rates and statistical power under various disease models (rare and common).
• Validated the proposed metrics on real genetic datasets, comparing performance against established methods.

Main Results:

• The proposed conditional entropy metrics maintained correct false positive rates for rare diseases.
• For common diseases, the new metrics showed superior or comparable control of false positive errors compared to existing model-free methods.
• The proposed methods demonstrated higher statistical power in detecting interactions across various common disease models.
• Real data analyses confirmed the ability of the metrics to detect significant gene-gene interactions.

Conclusions:

• Conditional entropy-based metrics offer a promising alternative to current approaches for identifying epistatic effects.
• These novel metrics provide a more reliable tool for genome-wide gene-gene interaction analysis.
• The developed methods enhance the discovery of complex genetic architectures underlying diseases.