Evaluation and comparison of gene clustering methods in microarray analysis

• DOI: 10.1093/bioinformatics/btl406
• Publication Date: 2006-08-02T18:47:44Z
• Authors: Anbupalam Thalamuthu; Indranil Mukhopad...; Xiaojing Zheng; George C. Tseng
• Keywords: Models, Statistical; Models, Genetic; Gene Expression Profiling; Software Validation; 0206 medical engineering; Reproducibility of Results; 02 engineering and technology; Sensitivity and Specificity; Artificial Intelligence; Multigene Family; Cluster Analysis; Computer Simulation; Algorithms; Software; Oligonucleotide Array Sequence Analysis

Abstract Motivation: Microarray technology has been widely applied in biological and clinical studies for simultaneous monitoring of gene expression thousands genes. Gene clustering analysis is found useful discovering groups correlated genes potentially co-regulated or associated to the disease conditions under investigation. Many methods including hierarchical clustering, K-means, PAM, SOM, mixture model-based tight have used literature. Yet no comprehensive comparative study performed evaluate effectiveness these methods. Results: In this paper, six are evaluated by simulated data from a log-normal model with various degrees perturbation as well four real datasets. A weighted Rand index proposed measuring similarity two results possible scattered (i.e. set noise not being clustered). Performance assessed predictive accuracy through verified annotations. Our show that consistently outperform other both while SOM perform among worst. provides deep insight complicated problem profile serves practical guideline routine microarray cluster analysis. Contact: [e-mail] Supplementary information: available at Bioinformatics online.