SMART REGION DEFINITION - A NEW WAY TO IMPROVE THE PREDICTIVE ABILITYAND INTERPRETABILITY OF 3-DIMENSIONAL QUANTITATIVE STRUCTURE-ACTIVITY-RELATIONSHIPS

This report describes a new methodology aimed at grouping 3D-QSAR inte raction energy descriptors into regions of neighbor variables bearing the same chemical and statistical information. These regions represent the structural variability of the series better than individual descr iptor variables and can advantageously replace them in the chemometric analysis. The algorithm used to generate such regions is described, t ogether with their application for improving the quality of GOLPE vari able selection. The method is illustrated on a series of 47 glucose an alogues, inhibitors of glycogen phosphorylase b, and is shown to impro ve both the predictive ability and the interpretability of the 3D-QSAR models obtained, comparing favorably with other methods previously de scribed.