An improved multi-factor dimensionality reduction approach to identify gene-gene interactions

Abstract

Genetic component of disease risk can be determined by gene–gene interactions. Multifactor dimensionality reduction (MDR) has widely used to identify gene–gene interactions based on the binary classification into high- or low-risk to evaluate the gene–gene interactions. However, the binary classification could not reflect the uncertainty of high- or low-risk classification. In this study, an improved classification method based on fuzzy sigmoid function was proposed to enhance MDR to identify GGIs. A total of 40 simulation data sets were used to compare the detection success rates of the improved MDR with the original MDR. The results expressed that our improved MDR obtained better detection success rates than MDR.