3D-QSAR study of antifungal N-myristoyltransferase inhibitors by comparative molecular surface analysis

• Published in: Chemometrics and intelligent laboratory systems Vol. 69; no. 1; pp. 51 - 59
• Main Authors: Hasegawa, Kiyoshi, Morikami, Kenji, Shiratori, Yasuhiko, Ohtsuka, Tatsuo, Aoki, Yuko, Shimma, Nobuo
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 28.11.2003
• Subjects: Genetic algorithm; Kohonen neural network; Molecular electrostatic potential; N-myristoyltransferase; Partial least squares; Quantitative structure–activity relationship; Variable selection; Molecular electrostatic potential; Quantitative structure–activity relationship; Genetic algorithm; Kohonen neural network; Partial least squares; N-myristoyltransferase; Variable selection
• ISSN: 0169-7439; 1873-3239
• DOI: 10.1016/S0169-7439(03)00112-6

The correlation between the molecular electrostatic potential (MEP) on benzofuran inhibitors and the N-myristoyltransferase (Nmt) inhibitory activity is investigated by a three-dimensional quantitative structure–activity relationship (3D-QSAR) approach. To each compound, a Kohonen neural network (KNN) is used to project the MEP values on a Connolly surface into a two-dimensional map. Each node in the KNN map is coded by the associated MEP value of the occupying sampling point and all the coding nodes are collected together to define chemical descriptors. The best partial least squares (PLS) model with the selected descriptors is searched among the possible candidate models by genetic algorithm (GA). The resulting best model is significantly more predictive than the one with all chemical descriptors. The model has given additional information about the binding of a benzofuran inhibitor to Nmt.