Sequences of one-symbol abbreviations of amino acids are applied as the basis to build up predictive model of Angiotensin converting enzyme (ACE) inhibitory activity of dipeptides and antibacterial activity of group of polypeptides. The developed models are one-variable correlations between biological activity and descriptors calculated with so-called correlation weights of amino acids. The numerical data on the correlation weights are obtained by the Monte Carlo method. The Index of Ideality of Correlation (IIC) is a mathematical function of (i) the determination coefficient; and (ii) sums of positive and negative values of "observed minus predicted" endpoints values. The obtained results confirm that IIC can be applied to improve predictive potential of models for ACE inhibitor activity of dipeptides and antibacterial activity of polypeptides.
Keywords: ACE Inhibitory activity; Antibacterial activity; Bioinformatics; CORAL software; Monte Carlo method; Peptide; Quasi-SMILES.
Copyright © 2019 Elsevier B.V. All rights reserved.