Improving the Biological Activity Prediction of Acetylcholinesterase and Butyl Cholinesterase Inhibitors Using Nonlinear Random Forest Algorithm

Due to the growing population of the elderly and the increasing trend of Alzheimer's disease, evaluation of acetylcholinesterase (AChE) and butyl cholinesterase (BChE) inhibitors, as major causes of Alzheimer's disease, is essential. Since the synthesis and investigation of each new compound is very costly and time-consuming, computational modeling techniques have been used to estimate biological activity. Up to now, various computational methods have been proposed which one of the major approaches, quantitative structure activity relationship, is based on the linear and non-linear methods using calculating the independent molecular descriptors. This study aimed to improve the biological activity prediction of AChE and BChE inhibitors using nonlinear random forest algorithm.

In order to predict the biological activity of AChE and BChE compounds, linear partial least squares and nonlinear random forest algorithms were used. To obtain more accurate and reliable results, 80% of the compounds were randomly used as a training sample, and the rest as a test sample, to construct the model and evaluate the predictive power of the model.

By applying nonlinear random forest model on AChE and BChE inhibitors, the accuracy of 89% was achieved. Finally, in order to examine more accurately the performance of the proposed model, the results were compared with the results obtained from the minimum partial error method, and the nonlinear random forest method had stronger performance than linear least squares method.

The observations indicated that the nonlinear random forest method could be very effective in predicting the biological activity of AChE and BChE compounds proposed by physicians and pharmaceutical chemists. Therefore, before animal and human testing of the proposed compound, the biological activity of the compound was estimated approximately to be 90%. Based on the estimated biological activity, it can be argued that a new drug combination, at the expense of time and money, has the potential of becoming a new drug or not.