Prediction Of MiRNA Disease Potential Association Based On Q Kernel Similarity And Matrix Completion

Studies have shown that microRNA(miRNA)is closely related to the emergence and development of many human diseases,and the emergence and development of a variety of diseases will be accompanied by abnormal expression of miRNA.Therefore,the study of miRNA disease association is helpful to the diagnosis and treatment of related diseases.However,we do not fully understand the role and potential molecular mechanism of miRNA in disease development.Ordinary biological experiments usually need higher cost,so researchers prefer to use computational methods to quickly and effectively predict the potential association between miRNA and disease at a lower cost,and can be used as a useful reference for experimental methods.Therefore,it is of great practical significance to use computational methods to predict the potential association of miRNA diseases.Aiming at the problem of miRNA disease potential association prediction,this paper proposes a new method of matrix completion algorithm based on q-kernel similarity(QIMCMDA).Based on the basic biological hypothesis that similar diseases tend to interact with similar miRNAs,QIMCMDA uses advanced q-kernel similarity to construct a complete similarity network between miRNAs and diseases.Then,we use matrix complement to score all the associations.A higher score means that there is a greater possibility of a potential connection.This paper uses cross validation to draw receiver operating characteristic curve and calculate area under the curve(AUC)value to prove the effectiveness of QIMCMDA.Leave-one-out cross validation shows that AUC can reach 0.9235,which has obvious advantages compared with other four common methods(IMCMDA,RLSMDA,TLHNMDA,WBSMDA).In addition,QIMCMDA case studies on three human diseases(Breast Neoplasms,Hepatocellular carcinoma,Colon Neoplasms)were carried out in this paper.The first 50 prediction results were verified based on dbDEMC and HMDD v3.2,and the effectiveness could reach 92%,90%and 96%respectively.The results show that our method performs well in inferring the potential interaction between miRNA and disease.It is expected that QIMCMDA will become an excellent supplement in the field of miRNA and disease potential association prediction in the future.