The Key Predictive Technologies Development Of Systems Pharmacology And Its Applications In Traditional Chinese Medicine Research

The key predictive technologies development of systems pharmacology based onlarge-scale integrated biological data can accelerate the development of novel drugs andtherapeutic targets, which efforts yet have not reached full fruition. In this work, we report asystematic approach that efficiently integrates the chemical, genomic and pharmacologicalinformation to conduct in silico ADME and target prediction and further decipher the actionmechanisms of Traditional Chinese Herbs. In this study, our work includes three aspects: oralbioavailability prediction, multiple drug-target interactions identification and analyzing themechanisms of action of Chinese Herbs.(1) Oral bioavailability prediction: In this work, theP-glycoprotein (P-gp) and cytochrome P450s, the main line of defense by limiting the oralbioavailability (OB) of drugs, were brought into construction of QSAR (Quantitativestructure–activity relationship) modeling for human OB based on805structurally diversedrug and drug-like molecules. The linear (multiple linear regression: MLR, and partial leastsquares regression: PLS) and nonlinear (support-vector machine regression: SVR) methodsare used to construct the models with their predictive ability verified with five-foldcross-validation and independent external tests. The performance of SVR is slightly betterthan that of MLR and PLS, as indicated by its correlation coefficient (R2) of0.80and standarderror of estimate (SEE) of0.31for test sets. For the MLR and PLS, they are relatively weak,showing prediction abilities of0.60and0.64for the training set with SEE of0.40and0.31,respectively. Our study indicates that the MLR，PLS and SVR-based in silico models havegood potential in facilitating the prediction of oral bioavailability and can be applied in futuredrug design.(2) Multiple drug-target interactions identification: The in silico models weredeveloped based on Random Forests (RF) and Support Vector Machine (SVM), and werevalidated by using the internal five-fold cross-validations and external independentverifications. The optimal models show impressive performance of prediction for drug-targetinteractions. The consistence of the performances of the RF and SVM models demonstratesthe reliability and robustness of the obtained models. In addition, the validated models were employed to systematically predict known/unknown drugs and targets involving the enzymes,GPCRs, ion channels, and nuclear receptors, which can be further mapped to functionalontologies such as target-disease associations and target-target interaction networks. Thisapproach is expected to help fill the existing gap between chemical genomics and networkpharmacology and thus accelerate the drug discovery processes.(3) Analyzing themechanisms of action of Chinese Herbs: In this part, we took CVD as an example todemonstrate the application of these obtained models. By incorporating the chemicalpredictors, target predictors and network construction approaches, we constructed apharmacological system, which generated64bioactive ingredients from the Chinese herbalLigusticum chuanxiong Hort., Dalbergia odorifera T. Chen and Corydalis yanhusuo WTWang, and predicted54potential targets related to the CVD. Based on these obtained results,we uncovered the potential action mechanism of these three important Chinese herbs for thetreatment of CVD, and provides basis for an alternative approach to investigate novelTraditional Chinese Medicine (TCM) formula on the network pharmacology level.