Consecutive KEGG pathway models for the interpretation of high-throughput genomics data

Acommon strategy to deal with the interpretation of gene lists is to look for overrepresentation of Gene Ontology (GO) terms or pathways. In related computational approaches the cell is formalized as genes that are grouped into functional categories. As output, alist of interesting biological processes is provided, which seems to be mostlycovered by the supplied gene list. However, it is more natural to model the cell as anetwork that reflects relations between genes. For many biological processes such information is available, but it is not used to the full extent in interpretational analyses. In this paper, we propose to interpret gene lists in network terms to provide the most probable scenario of gene interactions based on the available information about the topology of metabolic pathways. The proposed approach is an effort to exploit the biological information available in public resources to agreater extent in comparison to the existing techniques. Applying our approach to experimental data, we demonstrate that the currentlywidely employed strategy produces an incomplete interpretation, whilst our procedure provides deeper insights into possible molecular mechanisms behind the experimental data.