ACOCMPMI: An Ant Colony Optimization Algorithm Based on Composite Multiscale Part Mutual Information for Detecting Epistatic Interactions

Epistatic interaction detection plays a pivotal role in understanding the genetic mechanisms underlying complex diseases. The effectiveness of epistatic interaction detection methods primarily depends on their interaction quantification measures and search strategies. In this study, a two-stage ant colony optimization algorithm based on composite multiscale part mutual information (ACOCMPMI) is proposed for detecting epistatic interactions. In the first stage, composite multiscale part mutual information is developed to quantify epistatic interactions, and an improved ant colony optimization algorithm incorporating filter and memory strategies is employed to search for potential epistatic interactions. In the second stage, an exhaustive search strategy and a Bayesian network score are adopted to further identify epistatic interactions within the candidate SNP set obtained in the first stage. ACOCMPMI is compared with five state-of-the-art methods, including epiACO, FDHE-IW, AntEpiSeeker, SIPSO, and MACOED, using simulation data generated from 11 epistatic interaction models. Furthermore, ACOCMPMI is applied to detect epistatic interactions in a real dataset of age-related macular degeneration. The experimental results show that ACOCMPMI is a promising method for epistatic interaction detection.