Systematic Study of the Quality of Various Quantum Similarity Descriptors. Use of the Autocorrelation Function and Principal Component Analysis

The quality of various quantum similarity descriptors is studied systematically for a series of peptide isosteres important in pharmacology. To cope with the drawbacks of the Carbó and Hodgkin-Richards type indices, those being the time-consuming three-dimensional integration and the importance of the relative orientation, position, and conformation of the molecules, the use of the autocorrelation function is investigated. In combination with the application of principal component analysis, this approach is proven to be a more practical and faster tool for generating similarity sequences.