Pathway information f D

databases is a great boon to biologists, it also presents several important challenges. Almost 170 ‘‘pathway’ ’ databases exist, abases are the earliest and perhaps the best-known. Metabolic databases generally do not represent higher order cellular pro-pathways, about which rich datasets have been collected. A few metabolic pathway databases, for example KEGG FEBS 29327 FEBS Letters 579 (2005) 1815–1820E-mail address: pathways_febs@cbio.mskcc.org.which differ widely in form and content. This multiplicity of information sources can be daunting to researchers who simply wish to find information about genes or pathways of interest. The lack of uniform data models and data access methods makes pathway data integration extremely difficult, both mechanistically and semantically. To address these issues, it is useful to review the current landscape of pathway data and techniques for data integra-tion, and then to extrapolate the shape of desirable pathwa