Study On The Structure-activity Relationship Between Benzodiazepine And Cannabinoid Psychoactive Substances

In recent years,the increasing and abused New Psychoactive Substances(NPS)in the international illicit drug market has brought a series of challenges to global public health.On the other hand,the identification of these substances requires a great deal of experimental research,which leads to a lot of manpower and material resources.In order to quickly identify these NPSs,and assist toxicologists or forensic scientists to quickly understand new substances,a reliable Quantitative Structure-Activity Relationships(QSAR)model can be constructed through computational methods.QSAR study builds a bridge between chemistry and biology,using statistical methods to quantitatively relate the molecular structure of compounds with its biological activity or other physiological properties,so as to study the effects of structural fragments or property descriptors of compounds on its activity.The increasing of experimental data and the development of computer related disciplines have greatly promoted the application of QSAR model in such fields as drug design,environmental science,toxicology and so on,and achieved good results.First charter of the paper begins with a brief induction of background knowledge of NPS and highlights the difficulty in the rapid identification of NPS.Then,the development history of QSAR model is summarized,and the general process of QSAR modeling,the methods commonly used in modeling and model evaluation,as well as the research on the interpretability of the model are introduced.Finally,the application of QSAR model in the field of forensic medicine is reviewed.The second charter first introduces the mechanism of benzodiazepines(BDZs)as drug development and related biological background,emphasizing that the existing prediction models are almost all for the activity of compounds binding to GABA_A receptor,while ignoring the differences in the mode of action of compounds binding to the BDZ site of GABA_Areceptor.In order to better distinguish between agonists and non-agonists at the BDZ site,we developed machine learning classification models by collecting data and calculating molecular descriptors.These models could distinguish agonists and non-agonists with high accuracy.Finally,an interpretable study was conducted on the established model,and some substructure fragments that are more important for agonists were found,which were helpful for rational drug design in the future.The third charter firstly introduces the background of new psychoactive substances synthetic cannabinoids and their harm to the human body,highlights the importance of predicting binding affinity of synthetic cannabinoids(SCs)to cannabinoid type 1(CB1)receptors,and the research progress of QSAR model in predicting the activity of compounds binding to CB1 receptor is summarized.On this basis,we collected relevant data in the Ch EMBL database,then calculated eight molecular fingerprints of compounds,and built the QSAR regression model.Finally,model with the best predictive performance was selected to predict the binding activity value of SCs with unknown activity,and to preliminarily judge the biological activity of these substances and evaluate their abuse potential.The main content of the full text is to build a series of QSAR models by machine learning method,and then apply the models to the prediction of the activity values of substances that bind to GABA_A receptors and CB1 receptors in the central nervous system.On the one hand,the constructed QSAR model can be applied to the screening of drugs.On the other hand,it can assist in judging the activity and potential abuse of new psychoactive substances,providing reference for toxicologists or forensic scientists to research and control such substances.