CINF 78 |
| Enzymatic reactions are generally classified by EC numbers, which are chemically meaningful, but based on rules often ambiguous and heterogeneous. Their use for diversity analysis of metabolic reactions (the reactome) is limited. We report the mapping of a genome-scale set of 3468 enzymatic reactions by a self-organizing map (or Kohonen neural network), and their classification in terms of EC numbers. Computer assignment of EC numbers from the reaction equation is essential for the reconstruction of metabolic pathways. Furthermore, we show how a map of enzymatic reactions can be used to identify similarities between reactions exhibiting strong differences in EC numbers. This work uses a method for reaction representation that avoids identification of the bonds and atoms involved in the reaction (reaction center). The approach shows a general compatibility with the well established EC numbers, and overcomes some of their limitations for diversity analysis of the reactome. |
|
Advances in Reaction Informatics – New Tools and Technologies to Improve Synthesis Planning
9:00 AM-11:30 AM, Wednesday, 29 March 2006 Georgia World Congress Center -- B302, Oral
Division of Chemical Information |