Evaluation of associative classification-based multifactor dimensionality reduction in the presence of noise

Abstract

The advancements in genetic epidemiology have focused more on understanding the associations and functional relationships among the genes. Identifying the susceptible genes and their interaction effects over the complex traits remains statistically and computationally challenging. An associative classification-based multifactor dimensionality reduction method (MDRAC) was proposed to improve the identification of multi-locus interacting genes associated with a disease. The method was evaluated for one to six loci by varying heritability, minor allele frequency, case–control ratios, and sample size. The experimental results demonstrated significant improvements in the accuracy over the previous methods. However, the performance of MDRAC in the presence of noise due to genotyping error, missing data, phenocopy, and genetic heterogeneity is unknown. The goal of this study is to evaluate MDRAC for identifying single nucleotide polymorphism interactions in the presence of noise. Several experiments are conducted on simulated datasets and on a published dataset to demonstrate the performance of MDRAC. On average, the results showed improved performance over the previous MDR method in all the models. However, the performance of MDRAC is reduced in the presence of phenocopy and genetic heterogeneity, or their combinations with other sources of noise.