A QSAR And MD Study For Drug Design Of Apoptosis Signal-regulating Kinase 1

Many stimulating factors can cause biological bodies and cells to produce stress responses in daily life,called stressors.The human body is often exposed to many stressors,so in the process of continuous evolution,many defence and delay mechanisms emerge as the times require.MAPK pathway（Mitogen-activated protein kinases,mitogen activated protein kinase pathway）is the most widely used and critical stress response signal pathway in eukaryotes.Apoptosis signal-regulated kinase 1（ASK1）,also called MAP3K5,is a member of the MAPK family,and its abnormal expression result in a variety of stress disorders,such as non-alcoholic steatohepatitis and multiple sclerosis.Thus,to inhibit it has become an excellent way to treat these diseases.The discovery and optimization of ASK1 inhibitors has become an important topic.It is currently in the beginning stages to research on them and only a few inhibitors are in clinical testing.This paper intends to use QSAR and molecular dynamics simulation methods to help rational design ASK1 inhibitors.This study’s quantitative structure-activity relationship（QSAR）is encoded and calculated in Matlab and R Studio software on the Windows platform.In this study,using various molecular descriptor screening methods and machine learning modeling methods,a series of descriptors related to the characteristics of ASK1 inhibitors,are selected and debugged into four linear and nonlinear modeling methods.Finally,four QSAR models of ASK1 inhibitors are builded,namely multiple linear regression（MLR）model,random forest（RF）regression model,support vector machine（SVM）regression model and artificial neural network（ANN）model.By establishing and optimizing the model,the typical characteristics of ASK1 inhibitors and which characteristics in the molecular structure of inhibitors will play a significant role in inhibiting ASK1 activity are discussed at the molecular level.The stability and prediction ability of the model are evaluated by characterizing the application domain of the model,the internal and external validation of the model,and the consistency correlation coefficient.The stability and prediction ability of the model are proved to ensure that there is no over-fitting phenomenon in the model.In the final four models,the R² of the training and test set are all more significant than 0.9.In leave one out method that the Q²_LOO is more significant than 0.8and the consistency correlation coefficient CCC_EXT is greater than 0.85.Then the model’s effectiveness is further verified from the application of the model.In this study,the best QSAR model（RF）is used to predict the biological activity values of more than2 million small molecules in the ZINC Lead-like database（largest library of organic small molecule compounds）,These small molecules are sorted according to the activity values(p IC₅₀)and the top four small molecules（ZINC000069655610,ZINC000091663896,ZINC000045406778 ZINC000220673524）are extracted for molecular dynamics simulation,RMSD calculation,and binding free energy.In comparison with known inhibitors,these molecules are identified as new candidate structures for new inhibitor development.The QSAR model of ASK1 inhibitors is established in this study,and four small molecules are recommended through molecular data screening and molecular dynamics simulation.This study helps provide theoretical support for the development of ASK1 inhibitors.