Classification suport sytem base on SOM to 3-dimensional structure characteristic

Abstract

Because there is an enormous number of a conformation, it is difficult to estimate the pharmacophore and to do docking-study with the receptor using the 3D structure information that acquired from search of drug molecules' all stable conformation. So, there is the need to narrow down the comparison of conformation. In this research, we visualize the property-vector of pharmaceutical's 3D stable conformation using the Self-Organization Map(SOM: Kohonen map) that is a kind of the neural network. We have developed SOM data mining system. It specializes specialist's knowledge and analysis can be performed interactively. This time, we tried the analysis to three kinds of HIV virus protease Inhibitors.