We are excited to share our new preprint introducing methylTFR, an R package that infers transcription factor (TF) activity from DNA methylation.
DNA methylation shapes where TFs bind and is an important part of gene regulation, but it is hard to analyze at the level of single loci, especially in sparse low-input or single-cell data. methylTFR fixes this by combining methylation signals across all binding sites of a given TF, which gives a low-dimensional and interpretable TF activity profile for each sample. Across 147 human immune cell methylomes, methylTFR identified cell-type-specific regulators, such as CEBP and ETS family factors in myeloid cells. It also traced the naive-to-memory trajectory of CD4+ T cells and pointed to AP-1 factors as key memory regulators. Applied to sparse single-cell methylomes, the TF activity scores can be combined with gene expression and chromatin accessibility in joint factor models, giving multimodal views of lineage regulators.
This work was led by Irem B. Gündüz, together with Regina Nitsch and Sarath Kumar Murugan. We thank Jörn Walter and the members of the Integrative Cellular Biology and Bioinformatics group for their helpful feedback. The single-cell use case builds on data from our earlier collaboration with the Greenleaf and Ecker labs, and the bulk analyses use public data from the BLUEPRINT project. The work was funded by the ERA-NET TRANSCAN-3 project EPILUNAR (01KT2409) and a NanoBioMed Young Investigator Grant from Saarland University.
methylTFR is available on GitHub, with documentation and vignettes at https://epigenomeinformatics.github.io/methylTFR/.
Read the full preprint on bioRxiv: https://doi.org/10.64898/2026.09.29.755279