Bi-Phase evolutionary biclustering algorithm with the NSGA-II algorithm

Abstract

The gene expression data analysis is significant in investigating the fundamental biological phenomena. Biclustering algorithm is one of the powerful tools to discover the consistent patterns, and has commonly been utilized in the analysis of gene expression data. In this paper, we introduce an innovative biclustering algorithm which incorporates a bi-phase evolutionary architecture and the Non-dominated sorting and sharing (NSGA-II) algorithm. The first phase of the evolution is designed for the population of columns and rows, the second phase of evolution is for the population of biclusters. The two populations are initialized by a hierarchical clustering (HC) algorithm, and then the two populations are treated as two independent population to evolve in two phase respectively. The proposed algorithm was implemented both on synthetic datasets and real datasets, comparative experiments between the proposed algorithm and several typical algorithms demonstrate the effectiveness of the proposed algorithm.