Comparison of GA-MLR and PSO-ANFIS methods in QSAR study of some derivatives of tricyclic pyridobenzo and di-pyrido-di-azepinones as HIV-1 inhibitors

In this study, the inhibition effect of some derivatives of tricyclic-pyridobenzo and di-pyrido-diazepinones as HIV-1 inhibitors using Quantitative Structure Activity Relationship (QSAR) is investigated. The study was performed on the extended series of 59 molecules of derivates of tricyclic-pyridobenzo and di-pyrido-di-azepinones as HIV-1 inhibitors. Combination of genetic algorithm- multiple linear regression (GA-MLR) and combination of particle swarm optimization algorithm- adaptive neuro-fuzzy inference system (PSO-ANFIS) was used to create the model as the linear and non-linear methods, respectively. In this study the results indicated that PSO-ANFIS method has the better results for prediction of the biological activity of new and non-synthesized tricyclic-pyridobenzo and dipyridodiazepinones derivatives.