A New SVG Descriptor of Amino Acids and Its Application to Peptide QSAR

作者：Tong Xiaomei · 年份：2008 · 被引用次数：2 · 研究领域：Computational Drug Discovery Methods、Machine Learning in Bioinformatics、Protein Hydrolysis and Bioactive Peptides

To establish a new amino acid structure descriptor that can be applied to peptide quantitative structure activity relationship(QSAR) studies,a new descriptor,SVG,was derived from principal components analysis of the matrix of 74 geometrical indexes of amino acids.The scale was then applied in two panels of peptide QSAR that was molded by partial least square regression.The correlation coefficient(R2cum) and cross-validation correlation coefficient(Q2cum) of the obtained models were respectively 0.823 and 0.770 for 58 angiotensin-converting enzyme inhibitors;and 0.844 and 0.704 for 48 bitter tasting dipeptides.In addition,the estimation capability and generalization ability of the models were analyzed by external validation.The correlation coefficients of predicted values versus experimental ones of external samples(Q2ext) were 0.755 and 0.703.Satisfactory results showed that information related to biological activity can be systemically expressed by SVG scales.