Quantitative structure-activity relationship analysis of β-amyloid aggregation inhibitors

Inhibiting the aggregation process of the beta-amyloid peptide is a promising strategy in treating Alzheimer's disease. In this work, we have collected a dataset of 80 small molecules with known inhibition levels and utilized them to develop two comprehensive quantitative structure-activity relationship models: a Bayesian model and a decision tree model. These models have exhibited high predictive accuracy: 87% of the training and test sets using the Bayesian model and 89 and 93% of the training and test sets, respectively, by the decision tree model. Subsequently these models were used to predict the activities of several new potential beta-amyloid aggregation inhibitors and these predictions were indeed validated by in vitro experiments. Key chemical features correlated with the inhibition ability were identified. These include the electro-topological state of carbonyl groups, AlogP and the number of hydrogen bond donor groups. The results demonstrate the feasibility of the developed models as tools for rapid screening, which could help in the design of novel potential drug candidates for Alzheimer's disease.