Research On The Virtual Screening Of Active Ingredients Of Wenpi Decoction In The Treatment Of Chronic Kidney Disease Based On Bioinformatics

Chronic kidney disease(CKD)is a global disease,which can progress to chronic renal failure(CRF).In recent years,it has been found that traditional Chinese medicine has potential application in the treatment of CKD.The clinical application of traditional Chinese medicine compound attaches great importance to the overall regulation,syndrome differentiation and treatment,which has the characteristics of multi-component,multi-target integration.Wenpi soup comes from Sun Simiao's prescription for preparing the urgent thousand gold,which is composed of aconite,dried ginger,ginseng,liquorice and rhubarb.Although Wenpi decoction has been reported for the treatment of CKD,the effective components of Wenpi Decoction for the treatment of CKD still need to be further explored.With the development of artificial intelligence and computer-aided drug design,the accuracy of drug virtual screening technology has been improved.This bioinformatics calculation provides a new strategy for revealing the active ingredients and mechanism of traditional Chinese medicine compound from the perspective of network pharmacology.Using database resources and bioinformatics research tools,the interaction network of "disease gene drug" was constructed to predict the key targets in the process of disease regulation and analyze the target drugs related to chronic kidney disease.Drug virtual screening can be divided into ligand based drug virtual screening and receptor based drug virtual screening,such as molecular similarity calculation,pharmacophore model technology and molecular docking pair.In this study,molecular similarity calculation,ligand prediction model based on machine learning and molecular docking method were used to screen the potential active components of Wenpi Decoction in the treatment of chronic kidney disease.Experiment1: Construction of Wenpi Decoction and Ligand Compounds Database Objective to collect the components of Wenpi Decoction and construct the key target ligand library.Methods through literature research,tcmsp and batman-tcm database,the potential target of Wenpi decoction was obtained,and the enrichment of KEGG pathway wasanalyzed by David database.Based on the literature and pathway enrichment results,we determined to collect the key target ligand library from NF-? B signaling pathway,MAPK signaling pathway,JAK/STAT signaling pathway,TGF-? /SMAD signaling pathway and oxidative stress pathway.Results a total of 588 compounds of Wenpi Decoction were collected and 156 compounds were screened by the drug like index and oral bioavailability index.A total of 12 biological experiments were obtained by PubChem,including 23525 cases of high-throughput screening of active compounds.Experiment 2: Screening of Active Components of Wenpi Decoction Based on Molecular Similarity Calculation Objective To construct a screening model of traditional Chinese medicine compounds for molecular similarity calculation,and to screen the compounds in Wenpi Decoction virtually.Methods Firstly,based on RDKit module,the compounds of Wenpi Decoction were clustered by molecular clustering algorithm,and the similarity between Wenpi Decoction compounds and active ligand library was calculated by Guben coefficient.Preliminary compounds with similarity of more than 50% were selected as active compounds of Wenpi Decoction.Based on TCMSP database and molecular similarity,the target prediction results of active ingredients were calculated.Cytoscape(V3.7.1)was used to construct the "compound-target" topological structure network.Results 588 known compounds and23525 active ligand libraries in Wenpi Decoction were calculated.Finally,4790 pairs of compounds with similarity greater than 30% were obtained,of which 360 pairs had similarity greater than 50%.The results of Ginsenoside Rf,Glycyrrhizic acid,Benzoyl neoaconitine and Aloe emodin search were similar.Compound target interaction network composed of the core components of Wenpi Decoction can form three clusters and four compound radiation points.Ginsenoside Rf,Ginsenoside Rh2 and ?-curcumin formed cluster A;Curcumin,Glycyrrhizin,Glycyrrhizin A,Glycyrrhizin B,Glycyrrhizin G,Emodin methyl ketone formed cluster B;6-Gingerol,Aloe Emodin,?-sitosterol,Catechin formed cluster C;at the same time,Paeonol,Isoglycyrrhizin,Myristic acid and Emodinformed radiation points of compounds.Experiment 3: Screening of Active Components of Wenpi Decoction Based on Machine Learning Objective To construct a screening model of active compounds based on machine learning and screen the active compounds of Wenpi Decoction.Methods The experimental data of fibrosis related pathways were used as training set and test set.Based on SK-Learn module,machine learning algorithm is constructed.Random Forest(RF)model,Gradient Lifting Decision Tree(GBDT)model and Artificial Neural Network(ANN)model are constructed by selecting experimental data.The mixed models of RF+LR,GBD+LR,RF+ANN,GBDT+ANN are also constructed.The performance of each model was evaluated based on accuracy,recall,F1 value and ROC curve.Molecular docking experiments and model performance validation were carried out with known anti-fibrotic Chinese herbal compounds.Subsequently,the model was used to screen the compounds in Wenpi Decoction,and the compounds with a predicted probability of more than 90% were used as the active ingredients of Wenpi Decoction.Results The gradient lifting decision tree +logical review model had good predictive performance and stability.Its accuracy rate was0.80,recall rate was 0.80,F1 value was 0.79,AUC value was 0.872.Experiment 4: Virtual Screening of Wenpi Decoction Activity Based on Molecular Docking Objective To calculate the binding activity of 11 compounds in Wenpi Decoction to Smad3,Smad7,Nox4 and p22 phox targets based on molecular docking simulation.Methods Three-dimensional molecular structures of Catechin,Gallic acid,Emodin,Rhein,Hypaconitine,Aconitine,Glycyrrhizic acid,Ginsenoside Rg1,Ginsenoside Rf,6-Gingerol and Gingerone were obtained by PubChem.Molecular docking simulation experiments were carried out with Smad3,Smad7,Nox4 and p22 phox targets by CDocer software.Results The top six compounds in Wenpi Decoction with binding activity to p22 phox protein were Glycyrrhizic acid(24.9754),Ginsenoside Rf(22.6292),Gconitine(22.4404),Ginsenoside Rg1(20.4986),Hypaconitine(20.3279),6-Gingerol(18.2458).The top six compounds with binding activity to Nox4 protein were Ginsenoside Rf(54.0621),Glycyrrhizin(51.4306),Ginsenoside Rg1(45.433),Catechin(39.5441),Subaconitine(37.7516),Aconitine(34.0981).The top six compounds of Smad3 binding activity were Glycyrrhizic acid(41.5171),Ginsenoside Rg1(37.2358),Gnsenoside Rf(31.6412),Aconitine(30.3829),Hypaconitine(26.9489),Catechin(24.0615).The first six compounds of Smad7 binding activity were Ginsenoside Rf(51.2196),Ginsenoside Rg1.(46.8617),Glycyrrhizic acid(46.497),Aconitine(42.4877),Hypaaconitine(35.3275),Catechin(31.9764).Beief summaryDrug virtual screening is an important part of drug discovery and design,and has gradually become one of the important methods to explain the pharmacological mechanism.Drug virtual screening mainly includes ligand-based drug virtual screening technology and receptor-based drug virtual screening technology.In this paper,the active compounds of Wenpi Decoction in treating chronic failure were screened by the above methods.Firstly,an algorithm model for obtaining high throughput experimental data of active ligands was constructed,and 23525 data of related ligands were obtained.Based on literature,TCMSP and BATMAN-TCM,588 compounds of Wenpi Decoction were obtained.156 compounds with better bioactivity were screened by oral availability and drug-like index.Subsequently,a molecular similarity calculation model was constructed,and the similarity between Wenpi Decoction compounds and ligand compounds library was calculated.A total of 4790 pairs of compounds with similarity greater than 30% were obtained,of which 360 pairs had similarity greater than 50%.The results of ginsenoside Rf,glycyrrhizic acid,benzoyl neoaconitine and aloe emodin search were similar.Subsequently,a screening model for screening anti-fibrosis compounds was constructed by using molecular fingerprints and machine learning algorithm,and the active compounds of Wenpi Decoction were screened.Finally,the molecular docking simulation experiments of Wenpi Decoction compoundscatechin,gallic acid,emodin,rhein,hypaconitine,aconitine,glycyrrhizic acid,ginsenoside Rg1,ginsenoside Rf,6-curcumol and gingerone with Smad3,Smad7,Nox4 and p22 phox targets were carried out by CDocer software.The results showed that ginsenoside Rg1,ginsenoside Rf,glycyrrhizic acid and aconitine all showed good binding activity with the four targets.The hybrid intelligent method based on artificial intelligence combined with computer-aided drug design can better realize the screening task of active compounds in traditional Chinese medicine,and can provide basic research support for revealing the pharmacological mechanism of Wenpi Decoction system.