Abstract
A non-linear quantitative structure activity relationship (QSAR) model based on 350 drug molecules was developed as a predictive tool for drug protein binding, by correlating experimentally measured protein binding values with ten calculated molecular descriptors using a radial basis function (RBF) neural network. The developed model has a statistically significant overall correlation value (r > 0.73), a high efficiency ratio (0.986), and a good predictive squared correlation coefficient (q2) of 0.532, which is regarded as producing a robust and high quality QSAR model. The developed model may be used for the screening of drug candidate molecules that have high protein binding data, filtering out compounds that are unlikely to be protein bound, and may assist in the dose adjustment for drugs that are highly protein bound. The advantage of using such a model is that the percentage of a potential drug candidate that is protein bound (PB (%)) can be simply predicted from its molecular structure.
Keywords: ANN, drug-protein binding, in silico modelling, QSAR, screening, theoretical molecular descriptors.
Related Journals
Current Computer-Aided Drug Design
33