The Identification Of Target Druggability And Construction Of Novel Algorithm For Target Discovery By Strict Validation Of Drug Targets

The deciphering of human genome and the rapid development of the Omic techniques including genomics,transcriptomics,proteomics and so on have provided opportunities and challenges for drug R&D and target discovery.However,with the persistent efforts on drugs and targets research,the process of target discovery is largely delayed due to several key issues,such as 1)elusive definition of therapeutic targets,2)inadequate coverage of clinical trial drugs and targets,3)insufficient drug resistance mutation information,4)incomplete information on target gene expression profiles and target combinations,5)lack of knowledge on the druggability features of successful and clinical targets and target combinations,6)low effectiveness of the existing target prediction tools,and so on.To address these issues,a systematic analysis about target druggablity based on target strict validation,and a new target prediction algorithm were conducted,which is discussed as follow.Firstly,a new strategy of target validation based on the association among drugs,targets,and diseases was proposed.For the validation of target,we not only takes the drug activity into consideration,but also verify the roles of targets in disease model via in vivo experiments(such as knockout etc.),and then find the evidence of efficacy of drug on the target in disease model(cell,in vitro or in vivo).Secondly,based on the new strategy,a large number of FDA approved and clinical trial drug targets were validated,and a variety of information for target research was collected for further analysis.These information includes: a large number of clinical drugs and clinical targets;cross-links of target and drug entries to the corresponding pathway entries;drug resistance mutations;target gene expression profiles;target combinations.These data were further integrated into the therapeutic target database(TTD),so that other scientific researchers was convenient to access to the database.Meanwhile,the TTD was reconstructed via more advanced website technologies and more reasonable architecture to improve the access efficiency.Thirdly,the progression and characteristics analysis of innovative targets was conducted based on the successful innovative targets between 2004 and 2017,and the research about target druggability was carried out.The influences of target type,drug type,targeted disease type,government regulatory policies,etc.on clinical trial progress have been analyzed.The function of target in disease,the feature of drug binding to target,biochemistry,structure,sequence,biological system factors,and so on have been also systemically further analyzed,forming the quantitative characteristics of distinguishing fast and slow target in clinical trial progression significantly.Fourthly,druggability study of target combinations from multi-target drugs and drug combinations approved by FDA in Kinome has been done.After analyzing drug-target network constructed by multi-target drugs and drug combinations and phylogenetic tree,we have established the most popular targets in multi-target drugs and drug combinations,and then the different pattern of targets combination from multi-target drugs and drug combinations has been revealed.Meanwhile,the differences of mechanism of action of approved multi-target drugs and drug combinations have also been compared.This result would help to use the next generation of network pharmacology ideas to discover more effective new target combinations.Fifthly,a target prediction algorithm was proposed.After the completion of the strict target validation and TTD update,there was a amount of accurate and sufficient data for subsequent target prediction.On this basis,a new method based on Pfam database to construct virtual negative data set was used and we have constructed a novel algorithm for target prediction,of which the false positive was significantly reduced.After completing the analysis of the innovative targets,these characteristics of significantly distinguishing fast and slow target in clinical trial progression have been identified.These characteristics include the features of biological system(the numbers of similar proteins,distributed human tissues,and affiliated biological pathways)and network features of protein,etc.These characteristics will be applied in the development of target prediction tools to further improve the target prediction accuracy in the near future.Target druggability was a special and more complex protein function,and based on the new target prediction algorithm,the protein prediction tool SVM-Prot has been greatly updated.This update highly improved the prediction accuracy of SVM-Prot,which has paved the way for the development of subsequent target prediction tools.In summary,towards the existing problems in the discovery of drug targets,this article established a new strategy of targets strict validation,and then collected or acquired important data by computation for the target discovery(clinical drugs and clinical targets;cross-links of target and drug entries to the corresponding pathway entries;drug resistance mutations;target gene expression profiles;target combinations),and then the data were integrated into the TTD.The completion of targets strict validation and the integration of important data related to the target,would promote the target druggability studies.According to the innovative targets from nearly fourteen years,the characteristics of target with the powerful capability of distinguishing fast and slow target in clinical trial progression were obtained,which laid a solid foundation for establishing target druggability standard.Based on the FDA approved multi-target drugs and drug combinations in kinome,the commonalities and differences of multi-target drugs and combination drugs were systematically compared in drug-drug target networks and phylogenetic tree.Finally,a new algorithm for target prediction has been mainly constructed.This algorithm in protein function prediction showed great predictive performance,which made a good preparation for developing better target prediction tools.