QSAR Modeling of α-Ketooxazole Motif Analogues as Potent Anti-Alzheimer Agents

Abstract

Studies of quantitative structure-activity relationships are undeniably important in computational chemistry. QSAR were performed on a series of 35 α-ketooxazole analogues searching for improved anti-Alzheimer agents. PLS (Partial least square), MLR (Multiple linear regression) and FFNN (Feed forward neural network) were used to develop the models, including different descriptors such as electronic, lipophilic, and topological. The statistical significance and prediction capability of QSAR models were assessed. f = 34.2924, s = 0.406483, r= 0.922545, r2cv = 0.802496, r2 = 0.849 are the best MLR statistical expressions with strong prediction and authentication capability. MLR, PLS, and FNN have 2r (training and test-set) values of 0.8493, 0.8003, 0.8008, 0.763, 0.8577, and 0.8075, correspondingly, which indicate the model's soundness. The model indicates that the Kier chi 5 ring index, VAMP Heat of formation (whole molecule), and VAMP LUMO (whole molecule) greatly contribute to determining the anti-Alzheimer activity of FAAH (fatty acid amide hydrolase enzyme) antagonist. FAAH enzyme antagonists with increased potency as anti-Alzheimer agents may be developed using the model used in this work.