|
Abstract:
|
DNA methylation plays a critical role in cancer development. To better understand the shared and specific mechanisms of various cancers, it is important to investigate and compare differential methylation (DM) patterns between normal and case samples across different cancer types (known as pan-cancer analysis). Despite the rapid accumulation of cancer genomic data, current pan-cancer analyses call DM separately for each cancer type, which suffers from lower statistical power and fails to provide a comprehensive view of DM patterns across cancers. In this work, we propose a novel rigorous statistical model, PanDM, to characterize the similarities and differences of DM patterns among cancers by jointly modeling DNA methylation profiles of diverse cancer types. PanDM also clusters CpG sites accordingly, which in turn enhances the DM detection for each cancer type. We apply PanDM to DNA methylation data of 12 cancer types collected from The Cancer Genome Atlas project and discover novel DM patterns across major cancer types. Moreover, as PanDM works on the summary statistics, the same framework can in principle be applied to pan-cancer analyses of other functional genomic profiles.
|
