Implementation of Spectral Clustering on Microarray Data of Carcinoma Using Self Organizing Map (SOM)

The Microarrays technology is growing rapidly in bioinformatics. Microarray is a tool for measuring thousands gene expressions level of a sample. Microarray can be used to diagnose cancer including carcinoma. Carcinoma is one of cancer type that originated from epithelial tissue. Microarray data of carcinoma which highly dimensionality would be clustered to help diagnosing carcinoma patients. A highly dimensional data usually need a long computation time. In this paper, carcinoma microarray data would be clustered using spectral clustering method since it had a good capability to reduce data dimension. The result of spectral clustering would be partitioned using Self Organizing Map (SOM) algorithm. SOM is a popular implementation of artificial neural network for clustering. The advantage of SOM algorithm is that it efficiently handle big data and robust to data noise. This research aims to implement spectral clustering and SOM to classify microarray data of carcinoma genes expression from 7457 genes. The result of this study obtained three clusters of carcinoma genes.