You might think you know the leading players in the field of computational epigenetics; however, each new season brings forward a squad of exciting new players eager to prove their worth. Recently, one of these players EpiSCORE-d the goal of extending in silico cell-type deconvolution of human epigenomic data to many tissues.
While reference-based cell-type deconvolution algorithms can help to identify specific DNA methylation changes, they are limited by the need for profiles of constituent cell types, which are currently only mainly developed for blood. Generating DNA methylation reference profiles for other tissue-/organ-specific cell types remains a challenging prospect due to a lack of knowledge regarding tissue composition/cell-specific markers and the cost and sparsity of single-cell methylomics data.
In response, a team led by Charles E. Breeze (National Cancer Institute) and Andrew E. Teschendorff (University College London, UK) previously described EpiSCORE, an algorithm that uses single-cell RNA-sequencing data to generate a DNA methylation reference matrix that allows the quantification of cell proportions and specific differential methylation signals in the lung and breast. The authors extended their novel approach to other human tissues in their new study.
Let’s hear how this method can EpiSCORE a goal for your research:
- Leveraging high-resolution tissue-specific single-cell RNA-sequencing data (using negative correlation with DNA methylation and vice-versa) to generate a DNA methylation atlas defining thirteen solid tissue types and forty cell types that can computationally “decompose” bulk DNA methylome data
- Comprehensive validation with independent bulk and single-nucleus DNA methylation datasets and comparisons with lower-resolution methods demonstrate the validity of this approach
- This high-resolution DNA methylation atlas provides biological inferences and clinical insights, including:
- Confirming that most olfactory neuroblastomas derive from immature neurons; however, a link between poor outcome and high basal fraction/stemness does support alternative cells of origin
- Establishing α and β endocrine cells as the source of pancreatic neuroendocrine tumors and identifying misdiagnosed pancreatic adenocarcinoma cases, thereby demonstrating how this DNA methylation atlas has diagnostic potential
- Revealing the association of cardiovascular diseases with poor outcomes with inflammation-induced degradation of the extracellular matrix and vasculature
- Corroborating the neuronal origin for schizophrenia and highlighting a critical role of DNA methylation in mediating genetic risk
Overall, this method scores a goal for epigenetics and extends its application across different fields spanning cancer research, neurological disorders, and cardiovascular disease. Charles Breeze (National Cancer Institute) noted that “this method bridges the gap between single-cell and population-level epigenomics, enabling in silico cell-type deconvolution for many different areas of disease research.”
See how this method has EpiSCORE-d another goal for epigenetics at Nature Methods, March 2022.