1. ¹Department of Computer and Information Science, Purdue University School of Science, Indianapolis, IN, U.S.A.
2. ²Indiana University School of Informatics, Indianapolis, IN, U.S.A.
3. ³Center for Computational Biology and Bioinformatics, Indiana University School of Medicine, Indianapolis, IN, U.S.A.
4. ⁴Center for Bio-computing, Indiana University – Purdue University, Indianapolis, IN, U.S.A.

Correspondence: Qian You, Department of Computer and Information Science, Purdue University School of Science, Indianapolis, IN 46202, U.S.A. Tel: +1 317 701 3894; E-mail: qiyou@cs.iupui.edu

Received 22 July 2007; Revised 11 November 2007; Re-revised 5 January 2008; Accepted 13 January 2008; Published online 6 March 2008.

Abstract

We propose a new network visualization technique using scattered data interpolation and surface rendering, based upon a foundation layout of a scalar field. Contours of the interpolated surfaces are generated to support multi-scale visual interaction for data exploration. Our framework visualizes quantitative attributes of nodes in a network as a continuous surface by interpolating the scalar field, therefore avoiding scalability issues typical in conventional network visualizations while also maintaining the topological properties of the original network. We applied this technique to the study of a bio-molecular interaction network integrated with gene expression data for Alzheimer's Disease (AD). In this application, differential gene expression profiles obtained from the human brain are rendered for AD patients with differing degrees of severity and compared to healthy individuals. We show that this alternative visualization technique is effective in revealing several types of molecular biomarkers, which are traditionally difficult to detect due to 'noises' in data derived from DNA microarray experiments.