A Critical Look at Entropy-Based Gene-Gene Interaction Measures.

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Entropy-based measures are increasingly used for detecting gene-gene interactions, especially for complex diseases.
• These methods are favored for their potential to capture nonlinear relationships between genotypes and traits.

Purpose of the Study:

• To analytically characterize the properties of entropy-based measures for gene-gene interaction detection.
• To clarify the relationship between interactions identified by entropy-based measures and those detected by logistic regression models.

Main Methods:

• Analytical study of the properties of selected entropy-based measures.
• Comparison with traditional logistic regression models for gene-gene interaction analysis.
• Numerical simulations to validate theoretical findings.

Main Results:

• Entropy-based measures can capture nonlinear genotype-trait relationships.
• A lack of specificity was identified, where these measures may detect uninteresting signals as interactions.
• Theoretical findings were confirmed through numerical studies.

Conclusions:

• While promising for complex diseases, entropy-based measures require careful interpretation due to potential lack of specificity.
• Further research is needed to refine these measures for accurate gene-gene interaction detection.
• Understanding the limitations is crucial for reliable application in genetic association studies.