Domain specific vs generic network visualization: an evaluation with metabolic networks

Bourqui, R., Purchase, H.C. and Jourdan, F. (2011) Domain specific vs generic network visualization: an evaluation with metabolic networks. In: Australasian User Interface Conference, Perth, Australia, 17-20 January 2011, pp. 9-18. ISBN 9781920682972

Abstract

Metabolic networks have been drawn manually for many years, and over time have developed representational conventions that make them familiar to biologists. With increasing current biological discoveries, these networks need to be frequently updated and modified, and automatic visualization algorithms are thus becoming a necessity. Many existing automatic graph layout algorithms exist, and it is not known whether such generic algorithms are sufficiently useful for biologists, or whether algorithms that specifically consider the existing representational conventions are necessary. No prior task efficiency evaluation studies have been performed on biological network visualizations. This paper reports on an experiment comparing the task efficiency of biologically relevant motif-search tasks using three layouts, two of which were produced using existing generic graph layout algorithms (Force Directed, Hierarchical), and one which was specifically designed to take existing metabolic representation conventions into account (MetaViz). Despite the search task favouring the easy identification of node connectivity in the Force Directed layout, the results showed no efficiency difference between Force Directed and MetaViz. We conclude that embodying the representational conventions in an automatic algorithm is not an impediment to task efficiency, and that some minor improvements to MetaViz would enhance its usefulness for biologists even further.

• Item Type: Conference Proceedings
• Status: Published
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Purchase, Dr Helen
• Authors: Bourqui, R., Purchase, H.C., and Jourdan, F.
• Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
• College/School: College of Science and Engineering > School of Computing Science
• ISBN: 9781920682972