Novel Alignment Method of Small Molecules Using the Hopfield Neural Network
Masamoto Arakawa, Kiyoshi Hasegawa, Kimito Funatsu · Journal of Chemical Information and Computer Sciences · 2003
Molecular alignment is an important step in three-dimensional quantitative structure-activity relationship (3D-QSAR) such as comparative molecular field analysis (CoMFA). A reasonable molecular alignment is necessary for building a 3D-QSAR model. In this paper, a novel method for molecular alignment using the Hopfield Neural Network (HNN) is introduced. Four kinds of chemical properties are assigned to each atom of a molecule. Then, those properties between two molecules correspond to each other using HNN. To validate our method, HNN was applied to 12 pairs of enzyme inhibitors cited from the Protein Data Bank (PDB). As a result, our method could successfully reproduce the real molecular alignments obtained from X-ray crystallography.