Construction And Application Of Key Module Identification Algorithm Mediated By M~6A-Related LncRNAs Logical Regulation

As important gene regulatory factors,long noncoding RNAs(lncRNAs)play the critical role in the initiation and progression of various diseases.N6-Methyladenosine(m6A)methylation is one of the mRNA modifications,it plays important role in regulating RNA stability,splicing and translation.The m6A-related lncRNAs are the major regulatory factors involved in the progression of tumors.The interactions of m6Amodifications and lncRNAs can contribute to identify potential prognostic markers and therapeutic targets of tumors.While,the progression of tumors also involves the co-regulation of multiple lncRNAs.LncRNAs,target genes and m6A modifications jointly constitute the complex regulatory network in tumors.Presently,less attention was put on constructing diagnostic and prognostic markers for various tumor types based on the m6A-related lncRNAs logical regulatory network.In this work,the m6A-related lncRNAs logical regulatory model was constructed based on Matrix Factorization algorithm(MF)to identify tumor diagnostic signature modules,to explore whether m6A-related lncRNAs logical regulatory modules can be used as diagnostic biomarkers for mass tumor screening and their potential prognostic value.The m6A-related lncRNAs logical regulatory model comprehensively considered the regulation network of m6A-related lncRNAs and the co-regulation among m6A-related lncRNAs.The 4 m6A-related lncRNAs logical regulation modules(BRWD1-AS2,FOXP4-AS1,LEF1-AS1 and NRSN2-AS1)were identified as signature modules of tumor diagnosis.The tumor diagnosis signature module set showed high diagnostic value,it could distinguish tumor samples from normal(AUC,0.942;accuracy,0.873;sensitivity,0.818;specificity,0.912),distinguish tumor samples of each tumor type from all normal samples,and distinguish various tumor types among 23 tumor types.The validation data sets from the ICGC database and GEO database further validated the prognostic value of the tumor diagnosis classification model,which showed high accuracy,repeatability,reliability,and robustness.Among them,41 GEO validation data sets span 19 chip platforms and sequencing platforms,indicating that the tumor diagnostic classification model constructed in this work has high compatibility and wide adaptability for different data types.The tumor diagnosis signature module set also showed the potential application value of tumor staging diagnosis,with high accuracy and reliability of staging diagnosis in some tumors.The m6A risk score was calculated according to the core genes in the 4 tumor diagnosis signature modules,and it was used as the prognosis score of tumor patients.This work found that the tumor samples with higher m6A risk score showed higher tumor malignancy and a worse prognosis.Compared with clinicopathologic factors,the m6A risk score showed better predictive performance of tumor prognosis,with higher robustness,reliability,accuracy,and repeatability.It can be used as an independent prognostic feature of tumors,which was further demonstrated by the validation dataset from ICGC.Immunotherapy prediction results showed,the m6A risk score was positively correlated with the degree of immunotherapy benefits.In this study,TOP2A was identified as a potential target for tumor therapy,and the up-expression of TOP2A was significantly associated with poor prognosis of tumor patients.Taken together,tumor diagnostic signature modules identified by the m6A-related lncRNAs logical regulatory network can be used for mass tumor screening,staging diagnosis,and prognostic prediction.This work provides a new direction and strategy for the identification of diagnostic biomarkers and prognostic biomarkers for mass tumor screening.