Application Of Text Mining In Drug Active Gene Screening And Case Study On Rapamycin

The outbreak of drug-resistance has increased the drug shortage even in the face of the problem of drug shortages.However,new drug R & D is always at a high cost.The drug repositioning,the discovery of new therapeutic activity of the marketed drug,is gradually emerging for its low-cost drug discovery process.Due to the fact that the drug has entered the clinical stage,the problem of drug safety has been solved,which means that the drug repositioning method has the advantages of short time and low cost.However,there are many challenges in the current drug repositioning,among which,how to construct a drug disease relationship through a gene is one of the important research directions in the field of drug repositioning.In this paper,the gene pathway information was introduced,and the drug-genepathway-disease relationship was constructed,and the purpose of the drug-disease-related construction was achieved,and the drug repositioning was completed.The vast majority of such data is stored in the form of text during the course of the study.Manual reading to extract the relationship of the biological entities from mass text is high-cost and low-efficiency.Therefore,the application of the text mining method is an important way to solve this problem through the knowledge extraction in the text.In that paper,the text mining methods comprise the following steps of: firstly,searching drug relevant documents in PubMed to obtain 303,443 summary texts by using drug names of 19 typical drug repositioning drugs,and excavating genes containing the interaction relation from the text through a text mining method as a drug-related gene set;and then,Enriching the drug gene set on the gene pathway to obtain an ordered list of drug gene pathways;and then,associating the drug to the disease according to the direct association information of the gene pathway and the disease,thereby completing the goal of building the drug disease relationship.In this process,the main work focuses on:1.The effect of four kinds of text mining methods on the drug discovery strategy in this paper is compared.The fourmethods are Co-occurrence in Abstract based on PubTator,Co-occurrence in Sentence based on PubTator,Co-occurrence in Dependency tree structure based on PubTator and the Turku Event Extension System 2.1(TEES).The evaluation of the mining methods is two aspects: On the one hand,the accuracy of the gene set obtained for the text mining on the obtained drug-known gene set is obtained.The gene set of the four kinds of text mining methods sequentially calculates the ratio of all the gene sets in the blank control group gene set(all the gene sets in the human pathway in the KEGG database)to the accuracy of the known target gene set of the drug,The results are: 12.322,14.062,32.547 and 101.193.The higher the ratio,the better the effect is,and the TEES method is the best.Another aspect is the ranking of the drug-known pathways in the pathway list obtained by comparing the results of the genetic pathway enrichment.The known drug gene pathway is topper in the enriched result,the better the method is,the results show that the TEES method is the best.2,taking the drug rapamycin as an example,the case analysis is carried out.The active gene set of rapamycin is obtained by the TEES method,and then the active gene set is enriched and analyzed by a nine-channel enrichment method to obtain a corresponding ordered gene pathway list.And F-value the IPF_box method which are designed by us and the P-value are closer to the known drug channel list provided by the CTD database.All three enrichment results,the first five disease-pathways such as breast cancer are verified by the literature as a rapamycin indication.In addition,we have verified that the text mining has the function of drug disease prediction with the case of breast cancer-rapamycin relationship.3.The development of the text network visualization tool is used in the drug-disease relationship verification of the strategy.The disease gene network is constructed from the disease through the Literature Network text network visualization tool,the condition of the drug-related gene in the disease gene network is observed,and the possible mechanism of the drug disease action is inferred to achieve the purpose of verifying the drug-disease relationship.