An analytical method and apparatus using principal component analysis of nuclear
magnetic resonance (NMR) data for rapid molecular structure/function pattern recognition.
The presence of a molecular substructure in an organic compound is determined by
comparing principal components calculated from chemical shift values of the substructure
in selected compounds with those calculated from the chemical shift values of the
organic compound. Alternatively, principal components are calculated from the intensities
of NMR signals for a full spectrum, or selected regions thereof, to determine whether
an organic compound belongs to or is excluded from a set of structurally related
compounds. Also, the presence of a pharmacophore in an organic compound can be
determined by comparing the principal components derived from data on a set of
compounds known to bind to a particular receptor, or have a common biological effect,
with the principal components of the data set of the organic compound.
