Clustering of CD spectral data as a prototype QSAR model for neuropeptides

An analytical method that might eventually qualify as a general quality control assay procedure for polypeptide drug forms was described in the companion article to this paper. The detector is visible range circular dichroism spectroscopy. Multivariate data analysis reduced the spectral data to essentially four principal components (or factors) that are characteristic of each analyte. The level of analytical selectivity achieved among 51 analytes is very high. Using an alternative factor analysis algorithm, the selectivity is even more conveniently accomplished in the form of a 2-D cluster diagram presentation that has the potential of being a prototypical predictive in vitro model for correlating experimental data with structure-activity or structure-function relationships. Clustering of the analytes is a consequence not only of the chiral interactions associated with ligand exchange in the immediate primary coordination sphere of the host derivatizing reagent, but also of long-range intermolecular interactions between the coordination architecture of the host and the chiral polypeptides.