Research And Implementation Of LncRNA-protein Association Prediction System Based On GCN

Long non-coding RNA(lncRNA)plays a very extensive regulatory role in many biological activities,such as chromatin modification,cell differentiation and proliferation,RNA progression and apoptosis.lncRNA protein interaction plays an important role in post transcriptional gene regulation,such as splicing,translation,signal transduction and the progress of complex diseases.Traditional research process includes target recognition,biomacromolecule discovery,preclinical research,clinical research and approval,while computer simulation method has the advantages of short time and low cost.In this experiment,the graph convolution neural network in deep learning is used to build the model.In the experiment,the data obtained were preprocessed to delete the duplicate value,deletion value and low correlation data,that is,the lncRNA without interaction with at least two proteins and the protein without interaction with at least two lncRNAs,Then Gauss algorithm and Smith Waterman algorithm are used to calculate the similarity of long non coding RNA and protein to get the corresponding similarity matrix.Then,the lncRNA protein interaction matrix,lncRNA similarity matrix and protein similarity matrix are used to construct the global heterogeneous graph,and the graph convolution neural network model is used to predict the potential association between long non coding RNA and protein,so as to get the final score.In the 5-fold cross validation experiment,the AUC of the model is 0.9095.Furthermore,the applicability of this model is verified on the independent test set,and the AUC value of 0.8758 is obtained.Compared with other models or algorithms,the experimental results show that this model has better performance than other models and algorithms in different test sets.According to the functional and non-functional requirements analysis,the whole lncRNA protein association prediction system is divided into five modules,which are registration module,data processing module,prediction module,result management module and system management module.Bootstrap and sweetalert framework are used to build the front-end of the system,and Python and PHP are used to complete the front-end and back-end interaction functions of the system.After the completion of the system,as far as possible a comprehensive test case is designed to test the system,in order to ensure that all parts of the system have no problems.