Enzyme Active Site Prediction with SVMs
Abstract
Support Vector Machines (SVMs) are used to predict active sites of enzymes without homology information. Used descriptors are amino acid charge, average electron density, electrostatic momentum, average electrostatic potential, average electron density difference, average potential difference, solvent accessibility, and secondary structure. Calculations require three-dimensional coordinates of atoms. The result of this study may directly be applied in the drug discovery fields.
