A computationally efficient hypothesis testing method for epistasis analysis using multifactor dimensionality...

Area of Science:

• Genetics
• Epidemiology
• Bioinformatics

Background:

• Multifactor dimensionality reduction (MDR) is a nonparametric method for detecting epistasis in complex traits.
• Traditional MDR analysis relies on computationally intensive permutation testing for model significance.
• Genome-wide epistasis detection necessitates more efficient statistical evaluation methods.

Purpose of the Study:

• To develop and evaluate alternatives to large-scale permutation testing for assessing MDR model significance.
• To compare the power and type I error rates of a new hypothesis testing method against traditional permutation testing.
• To demonstrate the utility of the new method in a real-world genetic epidemiology study.

Main Methods:

• Simulated data from 70 epistasis models were used for comparison.
• Multifactor dimensionality reduction (MDR) was employed.
• A 1,000-fold permutation test was compared with hypothesis testing using an extreme value distribution (EVD).

Main Results:

• The EVD hypothesis testing method provides a viable alternative to the computationally expensive 1,000-fold permutation test.
• The new method is approximately 50 times faster than the traditional permutation test.
• The EVD method was successfully applied to a genetic epidemiology study of bladder cancer susceptibility.

Conclusions:

• Extreme value distribution (EVD) hypothesis testing offers a computationally efficient and statistically sound approach for assessing multifactor dimensionality reduction (MDR) models.
• This advancement is crucial for enabling large-scale, genome-wide epistasis analyses.
• The developed method accelerates the identification of gene-gene interactions influencing complex traits and diseases.