The Associated Prediction Of MicroRNA-Disease Based On Complicated Correlation Network

microRNAs(miRNAs)are single-stranded RNAs(ssRNAs)of 19–25 nucleotides in length that are generated from endogenous hairpin-shaped transcripts.They can regulate the expression of a target messenger RNAs(mRNAs)by base pairing with their 3'-UTRs,triggering their translational repression or degradation.Accumulated studies have shown that miRNAs play critical roles in various fundamental biological processes,such as cell development,proliferation,differentiation,apoptosis,signal transduction,and so on.Not surprisingly,the dysregulation of mi RNAs is obviously associated with the development and progression of complex diseases,such as neoplasm,cardiovascular diseases,metabolic diseases,autoimmune diseases,etc.It is impressed that more and more miRNAs have been discovered with the development of biotechnology.However,most of these miRNAs are rarely confirmed that they are associated with different diseases,and their biological functions are still unclear.So,it will be a meaningful work that predicting the functions of miRNAs and uncovering the associations between mi RNAs and diseases by bioinformatics.Based on the assumption(1)the majority of miRNA clusters are transcribed as a single unit and share common biological functions.(2)miRNAs are highly conserved in nucleotide sequences(especially the seed region of miRNAs)and secondary structures among closely related species over evolutionary time.(3)the members of miRNA family or cluster have the highly similar sequences and structures and are more likely to associate with the similar diseases.(4)functionally similar miRNAs tend to be associated with semantically similar diseases.In this paper,we download the datasets from mi RBase and HMDD and collect the nucleotide sequences,hairpin structures and the associations between miRNAs and diseases.Considered the following works,1.According to the nucleotide sequences of mature miRNAs and hairpin structure from pre-miRNAs,we calculate the similarity among miRNAs and propose the genes discriminant analysis(GDA)to assign newly detected miRNA into its specific miRNA families.Since those members who are from same family share same or similar biological functions,classifying new miRNAs into their corresponding families will be helpful for their further functional analysis in bio-experiments.2.Further developing an accurate miRNA structure network and combining it with miRNA function network which are proposed by Wang et al.,we construct a precise miRNA global similarity network(MSSN&MFSN)and train it on the classical algorithms to capture more potential disease-related miRNAs.To evaluate the effectiveness and feasibility of methods that we proposed,we employ K-fold cross-validation machine learning to achieve the accuracy rates of 80.74% for the original miRNA families with no less than three members.What's more,we test new miRNA similarity network on the different models that predicting the associated between miRNAs and diseases through LOOCV.Eventually,AUCs of 0.8212 and 0.9657 are obtained,respectively.Besides,cases studies show the results from our experiments,and some of them are verified by recent references.Therefore,whether genes discriminant analysis or miRNA global similarity network,all of them are an advantageous tool for predicting the disease-related miRNAs.Although more accurate results still need to be verified experimentally,the work of this paper provides direction for biological experimental verification and saves more resources for biologists or medical scientist.