COMPASS - A SHAPE-BASED MACHINE LEARNING TOOL FOR DRUG DESIGN

Building predictive models for iterative drug design in the absence of a known target protein structure is an important challenge. We presen t a novel technique, Compass, that removes a major obstacle to accurat e prediction by automatically selecting conformations and alignments o f molecules without the benefit of a characterized active site. The te chnique combines explicit representation of molecular shape with neura l network learning methods to produce highly predictive models, even a cross chemically distinct classes of molecules. We apply the method to predicting human perception of musk odor and show how the resulting m odels can provide graphical guidance for chemical modifications.