A Hybrid Clustering Algorithm and Functional Study of Gene Expression in Lung Adenocarcinoma

摘要

DNA Microarray technology provides a convenient way to investigate expression levels of thousands of genes in a collection of related samples during different biological processes. Researchers from diverse disciplines such as computer science and biology have found it interesting as well as meaningful to group genes based on the similarity of their expression patterns. Hierarchical clustering and k-means clustering are commonly used algorithms to group genes with similar expression patterns. However, in spite of having some advantages such as producing tighter cluster than other algorithms, k-means clustering has some limitations also. The performance of k-means clustering algorithm largely depends on the selection of the value of k i.e., the number of clusters. In this research work, we proposed a new method to combine k-means clustering with hierarchical clustering to overcome the limitation. To test the algorithm, we used microarray data on lung adenocarcinoma, the most common type of non-small-cell cancers. We identified a number of representative genes from the group of normal tissue and from the group of KRAS mutation tissues. Genes for both of these groups were clustered using our proposed method. Finally we conducted functional investigation of the differentially expressed genes using Gene Ontology database to find changes in the enrichment of molecular functions of the genes contained in each cluster of both normal and KRAS positive groups. We discovered that our proposed method can group genes with similar expression pattern together and hence it can be used in future for clustering microarray data.