We develop computational methods that read cell identity from the epigenome: which cells make up a tissue or a tumour, where a cancer came from, and how individual cells differ.
Estimating the cell-type composition of DNA methylation data, from aggregated profiles down to individual sequencing reads, with and without reference panels.
Using the epigenetic memory carried by tumour cells to trace cancers back to the normal cell types they arose from.
Quantifying epigenetic heterogeneity and integrating multiomic measurements in single cells and tissues.
How mutations in histone genes, such as H3.3-G34W in giant cell tumours of bone, rewire the epigenetic landscape.
Assigning a DNA fragment to its most likely cell of origin gets harder as the number of possible cell types grows. The solution turned out to be letting the model know when to remain unsure.
Read the postRead-level DNA methylation classification and cell-type deconvolution with deep learning, built on data-driven soft labels.
Read-level DNA methylation pattern identification and tumour deconvolution with a Transformer model.
R/Bioconductor package for systematic aggregation and analysis of bisulfite sequencing data.
Continuable benchmark of DNA methylation sequencing workflows against an experimental gold standard, with interactive results.