New autocorrelation QTMS-based descriptors for use in QSAM of peptides

The autocorrelation analysis of the quantum topological molecular similarity (QTMS) has been used to discover some new indices for amino acids. QTMS descriptors include both physical and topological property molecules and produce a matrix of variables per molecule. To handle cubic array of QTMS descriptor calculated for a list of 20 naturally occurring amino acids, autocorrelation analysis has been performed. The proposed indices were validated by construction of quantitative sequence-activity model for three sets of peptides with different chain sizes. Modeling procedure used genetic algorithm-partial least square as variable selection and regression tools. The obtained models were of better or similar qualities compared with previously reported models for the same data sets. Also, the proposed indices in this article were able to determine active site region of the peptides under study