Design and Implementation of Modified Sparse K-Means Clustering Method for Gene Selection of T2DM

Abstract

Clustering is a trouble-free procedure for extracting clusters from large databases. Data clustering is the task of segregating and classifying data elements on the basis of some aspect of resemblance between the elements in the group. If such data encountered with many features, we may face many problems for analysis, for example, “genomic” data in bioinformatics. Nowadays, substantial advancement was achieved in the identification of risk genes involved with Type II Diabetes Mellitus (T2DM). In this paper, we are designing and implementing a modified sparse K-means clustering method for gene selection by collecting microarray data and grouped into clusters to analyze which is the most relevant gene for susceptibility of T2DM. The identification of most relevant genes of the disease is based on expression levels by genome-wide analysis and to make optimal decision results.