epiBAT: Multi-objective bat algorithm for detection of epistatic interactions
Jozef Sitarčík, Mária Lucká · 2019
Detection of epistatic interactions associated with diseases can improve prevention and diagnosis of those diseases. Epistatic interactions are nonlinear interaction effects of single nucleotide polymorphisms (SNPs), which are substitution mutations occurring at some specific position in the genome. Detecting associations between them is very computationally expensive, as more complex diseases can be associated only with epistatic interactions of two and more SNPs, thus making a very large quantity of possible SNP combinations needed to test. To cope with such high computational complexity, current methods are also based on bio-inspired algorithms. In this paper we propose epiBAT, a new algorithm based on bat algorithm with multiple objectives and tabu search. We apply our algorithm on different testing data sets and compare it with other existing methods. The experiments have shown that the new epiBAT method achieves similar or better results than the compared methods.