Title: A heuristic for gene selection and visual prediction of sample type

Authors: Jianping Zhou, Georges Grinstein, Kenneth Marx

Addresses: Department of Computer Science, Department of Chemistry, University of Massachusetts Lowell, 1 University Ave., Lowell, MA 01854, USA. ' Department of Computer Science, Department of Chemistry, University of Massachusetts Lowell, 1 University Ave., Lowell, MA 01854, USA. ' Department of Computer Science, Department of Chemistry, University of Massachusetts Lowell, 1 University Ave., Lowell, MA 01854, USA

Abstract: In this paper, we introduce a heuristic method for gene selection. We target this method, coupled with RadViz visualisation, to the visual prediction of tissue samples which may exist in normal and disease states. As a result of this coupling, the gene selection process, predictive model training and evaluation as well as the model|s application for tissue sample prediction can all be intuitively visualised. Such integrated visual analytics enhance the insight provided by classical statistics and machine learning methods. The case study shows our proposed method is cost effective and achieves competitive performance when compared with several widely used techniques.

Keywords: gene selection; visual prediction; visualisation; classifier design; classifier evaluation; feature evaluation; feature selection; bioinformatics; tissue sample prediction; tissue samples.

DOI: 10.1504/IJDMB.2011.041558

International Journal of Data Mining and Bioinformatics, 2011 Vol.5 No.4, pp.428 - 448

Received: 04 Jun 2009
Accepted: 15 Jan 2010

Published online: 24 Jan 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article