Ranking genes based on kernels

Issue title: Advances in medical intelligent decision support systems

Guest editors: Vassilis Kodogiannisx, Ilias Petrouniasy and John Lygourasz

Article type: Research Article

Authors: Iakovidou, Nantia^{a; *} | Nanopoulos, Alexandros^b | Manolopoulos, Yannis^a

Affiliations: [a] Department of Informatics, Aristotle University, GR-54124 Thessaloniki, Greece | [b] Institute of Computer Science, University of Hildesheim, D-31141 Hildesheim, Germany | [x] University of Westminster, United Kingdom | [y] The University of Manchester, United Kingdom | [z] Democritus University of Thrace, Greece

Correspondence: [*] Corresponding author. Tel.: +30 2310 991924; Fax: +30 2310 991913; E-mail: niakovid@csd.auth.gr

Abstract: Retrieval queries in microarray databases can rank genes according either to their similarity by detecting functionally related genes, or to their importance by detecting genes with significant regulation role. Although both rankings are useful, they can be contradicting. For instance, similar highly ranked genes may have low importance and vice versa. Thus, we propose a Web-inspired kernel method for fusing the two rankings according to the user needs.

Keywords: Gene similarity, gene importance, ranking

DOI: 10.3233/IDT-2009-0058

Journal: Intelligent Decision Technologies, vol. 3, no. 4, pp. 233-238, 2009