Techniques of biclustering in gene expression analysis

DNA microarrays allow to observe expression levels of large numbers of genes under different experimental conditions. One of the main goals in the gene expression analysis is clustering genes with similar expression patterns under specified conditions or clustering conditions according to expression of certain genes. Traditional clustering methods have some limitations, such as assumption that related genes behave similarly across all experimental conditions and partition of the genes into disjoint groups, which implying an association of the gene with a single biological process. In recent years, several biclustering methods were proposed to overcome the limitations of clustering. Biclustering allows for simultaneous grouping of genes and conditions, which leads to identification of subsets of genes exhibiting similar behavior across a subset of conditions. In gene expression analysis, the term biclustering was introduced in 2000 by Cheng and Church and since then several methods were developed. The paper presents a review of the algorithmic approaches to biclustering.