Clustering methods classify patterns into clusters using the entire set of attributes of patterns in the similarity measurement. In plenty of cases, patterns are similar under a subset of attributes only. The class of methods that cluster patterns based on subsets of attributes is called biclustering. Biclustering simultaneously groups on both rows and columns of a data matrix and has been applied to various fields, especially gene expression data. However, the biclustering problem is inherently intractable and computationally complex. In recent years, several biclustering algorithms which are based on linear coherent model have been proposed. In this paper, we introduce a novel GA-based algorithm that uses hyperplane to describe the linear relationships between rows (genes) in a sub-matrix (bicluster). The performance of our algorithm is tested via simulated data, gene expression data and compared with several other bicluster methods.