Elucidating context-dependent N6-methyladenosine (m6A) functions using machine learning
N6-methyladenosine (m6A) is the most abundant methylation in transcripts, existing in >25% of human mRNAs. Exciting recent discoveries indicate a close involvement of m6A in regulating many different aspects of mRNA metabolism and cancer. However, our current knowledge about how m6A levels are regulated and whether and how the regulation of m6A levels of specific genes can play a role in cancer is elusive. We have developed a host of computational and machine learning tools including widely used exomePeak for identifying transcriptome-wide m6A and differential m6A sites, predicting context-specific m6A-regulated genes and elucidate m6A-associated disease. In this talk, I will present some of these tools, focusing on m6A-express, a machine learning method for predicting context-dependent m6A-regulation of gene expression (m6A-reg-exp). Using m6A-express, we uncovered that human m6A readers, METTL3 and METTL14, collaboratively regulate transcription and RNA processing by mediating m6A-reg-exp of highly different target genes in Hela cells. In contrast, METTL3 regulates stress-related processes by mediating m6A-reg-exp of a distinct group of genes in HepG2 cells. We further examined the m6A-reg-exp in the human brain and intestine tissues and found highly tissue-specific targets and their involved functions. This study demonstrates the effectiveness of m6A-express in predicting context-specific m6A-reg-exp and highlights the complex, context-specific nature of m6A-mediated post-transcriptional regulation.