REW-ISA V2: A Biclustering Method Fusing Homologous Information for Analyzing and Mining Epi-Transcriptome Data

Background: Previous studies have shown that N6-methyladenosine (m(6)A) is related to many life processes and physiological and pathological phenomena. However, the specific regulatory mechanism of m(6)A sites at the systematic level is not clear. Therefore, mining the RNA co-methylation patterns in the epi-transcriptome data is expected to explain the specific regulation mechanism of m(6)A. Methods: Considering that the epi-transcriptome data contains homologous information (the genes corresponding to the m(6)A sites and the cell lines corresponding to the experimental conditions), rational use of this information will help reveal the regulatory mechanism of m(6)A. Therefore, based on the RNA expression weighted iterative signature algorithm (REW-ISA), we have fused homologous information and developed the REW-ISA V2 algorithm. Results: Then, REW-ISA V2 was applied in the MERIP-seq data to find potential local function blocks (LFBs), where sites are hyper-methylated simultaneously across the specific conditions. Finally, REW-ISA V2 obtained fifteen LFBs. Compared with the most advanced biclustering algorithm, the LFBs obtained by REW-ISA V2 have more significant biological significance. Further biological analysis showed that these LFBs were highly correlated with some signal pathways and m(6)A methyltransferase. Conclusion: REW-ISA V2 fuses homologous information to mine co-methylation patterns in the epi-transcriptome data, in which sites are co-methylated under specific conditions.