A large number of genomic studies are underway to determine which genes are abnormally regulated by methylation in disease. However, our understanding of how disease-specific methylation changes potentially affect expression is poorly understood. We need better tools to explain specific variation in methylation that potentially affects gene expression in clinical sequencing. We have developed a model, Methylation Interpolated Gene Signatures (MIGS), that captures the complexity of DNA methylation changes around a gene promoter. Using data from the Roadmap Epigenomics Project, we show that MIGS significantly outperforms current methods to use methylation data to predict differential expression. We find that methylation changes at the TSS and downstream ~2kb are most predictive of expression change. MIGS will be an invaluable tool to analyze genome-wide methylation data as MIGS produces a longer and more accurate list of genes with methylation-associated expression changes.

MIGS: Methylation Interpolated Gene Signatures Determine Associations Between Differential Methylation and Gene Expression