A Network Pharmacology And Deep Learning-based Approach To Study The Mechanism Of Action Of Cinnamon In The Treatment Of Myocardial Ischemia-reperfusion Injury

Acute myocardial infarction frequently results in myocardial ischemia-reperfusion injury(MIRI),whose occurrence exacerbates cardiac dysfunction and raises the possibility of a bad prognosis.Despite significant advancements in recent years in our knowledge of the pathogenic mechanisms underlying MIRI,the clinical treatment outcomes are still somewhat restricted.Furthermore,cinnamon has been shown to offer potential for the treatment of MIRI due to its anti-oxidant,anti-inflammatory,and immune response-modulating properties.The network pharmacology approach is one of the most often utilized techniques to investigate the mechanism of action of cinnamon in the treatment of MIRI and,more significantly,to identify the key active components and their targets of action.Traditional network pharmacology methods,on the other hand,rely on existing drugs and disease databases to gather relevant components and targets,and the number of compounds and targets available is restricted.Contrarily,Drug-Target Interactions(DTI)prediction is the most straightforward method for identifying possible targets based on drug-target combinations with known interactions to forecast unknowable associations.Therefore,in this study,we developed a DTI prediction model to obtain more targets of cinnamon active ingredients to increase the accuracy of network pharmacology analysis and prediction.This is accomplished primarily by undertaking the following:(1)A deep learning-based approach for predicting DTI is proposed.Drug and target sequence information,or structural data,is frequently used by DTI prediction algorithms for feature learning.Modeling is also interesting for the intricate interactions between drug atoms and amino acids.The SMILES structural formula of a drug is transformed into a molecular graph as a result of the use of a joint model for feature representation learning.In addition,for the amino acid sequences of the targets,a pre-trained model is used for embedding representation,followed by Bi-LSTM for feature representation learning,and an attention mechanism is introduced to model the intermolecular interactions between atoms.Experimental comparisons reveal that GELA-DTI performs better than the cutting-edge baseline method.Furthermore,through ablation experiments,we are able to learn more about how each GELA-DTI component affects the body.Finally,a case study based on atherosclerosis was created to illustrate the usefulness of this research.(2)A network pharmacology approach incorporating prediction of DTI is proposed.The main source of pertinent active components and targets for the network pharmacology analysis method is drug and disease databases.Protein-protein interaction analysis is then carried out,and a threshold value is chosen to screen the key targets,following the intersection of the action targets of the ingredients and disease targets.Subsequently,these targets are used to construct the network of active ingredients-targets-signaling pathways,and topology analysis is employed to determine the fundamental ingredients and targets.Based on this,we add the DTI prediction method to the 873 action targets of cinnamon.After further analysis and molecular docking validation,we were able to identify the six main active ingredients of cinnamon(Oleic acid,Palmitic acid,Beta-sitosterol,Eugenol,Taxifolin,and Cinnamaldehyde),three action targets(PTGS2,GSK3 B,and MAPK14),and four related pathways(PI3KAkt,MAPK,IL-7,and HIF-1).According to the zebrafish experiment,taxifolin may have a beneficial impact on the therapy of MIRI.