The invention is a method for predicting the biological, chemical, and physical properties of molecules from their chemical shift data using through-bond and spatial distance connectivity patterns. In this method, predicted NMR chemical shift data that has already been structurally assigned in the process of developing the spectral predictions is used to construct a model that predicts biological, chemical and physical properties of the molecule. Since the structural assignments are only used to established molecular distance connectivity relationships, models can be developed for sets of molecules that do not share a common backbone geometry. In model development and use there is no molecular docking step. These models correlate particular molecules with desired "endpoints," including receptor-ligand binding, cancer effects, drug absorption and others. The new technique is a three dimensional Quantitative Structure Data-Activity Relationship (QSDAR) based on the spectrum-activity leg in the triangular structure-spectrum-activity relationship. The invention provides a quantitative relationship between spectra and certain properties or activities of the molecule, and will have important implications in the search for new therapeutic drugs. 3D-QSDAR Modelling is a very rapid objective process compared to conventional predictive methods. In comparable published results, the 3D-QSDAR model quality consistently exceeds that of conventional QSAR predictive methods.
U.S. Patent: No. 7,996,156 issued 2011-08-09
Bill Ronnenberg, JD-MIP, MS
FDA Technology Transfer Program
10903 New Hampshire Ave.
Building WO1, Rm 4214
Silver Spring, MD 20993
OTT Reference No: E-297-2001/0
Updated: August 10, 2015