Computational study of histamine H3-receptor antagonist with support vector machines and three dimension quantitative structure activity relationship methods

Support vector machine (SVM) was used to derive QSAR models for 144 histamine H3 receptor antagonists. Several additional descriptors determined by SVM method, such as highest occupied molecular orbit (HOMO) and lowest unoccupied molecular orbit (LUMO), combined with conventional fields of CoMFA and CoMSIA were employed to construct 3D-QSAR model. The results show that inclusion of HOMO and LUMO is meaningful for 3D-QSAR model. The validation of this model was testified by some structural diverse compounds, which were not included in the CoMFA and CoMSIA models. Therefore, the non-linear SVM method can be applied to the selection of reasonable additional descriptors for 3D-QSAR investigations. The combination of these techniques could dramatically improve the statistical properties of model.