An Improved Method for Clustering Gene Microarray Data Based on Intra-Cluster Distance and Variance

A major use of microarray data is to classify genes with similar expression profiles into groups in order to investigate their biological significance. Cluster analysis is by far the most used technique for gene expression analysis. It has grown to be an important research topic in a wide variety of fields owing to its wide applications. A number of clustering methods exist with one or more limitations, such as, dependence on initial parameters, inefficiency in presence of noisy data, to name a few. This paper proposes a novel clustering algorithm for gene microarray data which is free from the above limitations. Besides, it is simple to implement, and is has been proved to be very effective even in the presence of noisy data. Further, it is extremely exhaustive and is hence, less likely to get stuck at local optima.