Clustering Gene Expression Data with Min-max-median Initialized Fuzzy C-means Algorithms

DNA microarray technologies have made it possible to analyze simultaneously thousands of gene expression patterns. This paper presents the successful application of a new clustering algorithm to gene expression data involving 72 data points in a 6817 dimensional space. We compare the new clustering algorithm to other clustering algorithms that have been used for expression analysis and show its effectiveness. The paper also investigates the effect of the dimensionality of feature space on clustering performances. Our results show that the clustering performance may be degraded as the dimensionality of feature space increases. For microarray data, we use principal component analysis (PCA) methods to reduce the dimensionality and apply our clustering algorithm to the PCA-reduced feature space, and show promising results.