Virtual Screening For New Drug Candidates Against Alzheimer's Disease Based On STITCH Database

It is considered that alpha7 nicotinic acetylcholine receptor is a drug target for the treatment of Alzheimer's disease. Through pharmacophore search, Lipinski's"rule of 5"and molecular docking, our subject has eventually singled out 179 small molecules based on stitch database. Then, we searched the function reported in the literatures of the 179 small molecules. For example, molecular of No.41 contributes to muscle activity as sparteine analogues. Molecular of No.85 enhances the activity of muscarinic. Molecular of No.123, which is Melperone, has been used as antipsychotics in the treatment of schizophrenia.In order to further analyze the properties of small molecules related to the combination withα7 nAChRs, we've built a quantitative structure-activity relationship model. Eventually we filter out 5 descriptors, they are Van der Waals surface area, Weiner path number, Number of rotatable single bonds, Largest vertex eccentricity in graph and Total negative 1 vdw surface area. The result of model calculation shows: our classification model can accurately distinguish better docking molecules from worse docking molecules, its accuracy rate is 98%, and result of validation set is basically same as test set 1. So it proves our model basically does not exist overfitting and possess reliable predictive power. The accuracy rate of test set 2 is 85%, which means our model can filter out most potential drugs and our docking results are reliable. Most molecules that can be used as drug candidates actually have good docking results.