BIOL 185 |
| Classification of enzymatic reactions enables the exploration of enzyme structure-function relationships, assists in the reconstruction of metabolic pathways from genomes, and allows for the comparison of reactomes. Processing chemical reactions by statistical or machine learning techniques, in order to their classification, requires a numerical, fixed-length representation. This is the case of the MOLMAP reaction descriptors (J. Chem. Inf. Model. 2005, 45, 1775-1783), on which we based a study with self-organizing maps (SOMs) and Random Forests for the classification of a genome-scale set of enzymatic reactions (Angew. Chem. Int. Ed. 2006, 45, 2066-2069). SOMs provide intuitive visualizations of the differences between reactions, and allowed for the identification of a few similar reactions with highly different EC numbers. Random Forests could be trained to predict EC numbers from the structures of reactants and products. We present an overview of the approach, and the latest developments concerning the optimization of the reaction descriptors. |
|
Enzymes
4:30 PM-6:30 PM, Wednesday, 13 September 2006 Moscone Center -- Hall D, Poster
Division of Biological Chemistry |