Thank you! I would love that 🥰🥰🥰 Excited for future collaboration opportunities too ❤️❤️❤️

This paper is so cool! I’ll try to read it a few more times to fully digest the new ideas here. Very inspiring work.

We’d also like to give a shout out to all the amazing work that our study built upon, including but not limited to ChromBPNet, scBasset, TOBIAS, DNase footprinting, and many others! We hope that together we can use our understanding of gene regulation to advance human health.

This was an amazing team effort led by @yanhu97.bsky.social, @maxhorlbeck.bsky.social, and @ruochiz.bsky.social, joined by colleagues in our lab, the Wagers Lab, and GRO@Broad, supported by HSCRB, HSCI, @broadinstitute.org, @igvfconsortium.bsky.social, @bostonchildrens.bsky.social, EWSC, and GRO.

Because PRINT/seq2PRINT are applicable to any standard bulk or single-cell ATAC-seq dataset, we hope many people will try it out on their existing or future data! Code, tutorials, and examples are available at github.com/buenrostrola...

We believe PRINT will be a powerful tool to study gene regulation/dysregulation in complex biological systems, rare cell types, as well as diseases, opening up new opportunities to answer biological questions.

Interestingly, seq2PRINT captured de novo sequence motifs resembling composite motifs involving Runx, Ets, and Gata, many of which were supported by structural data from PDB or AlphaFold3 predictions, suggesting physical cooperations between TFs.

Finally, in collaboration with the Wagers Lab at HSCRB, we examined the CRE alterations during murine hematopoietic stem cell aging. We observed global gain of Gata/AP-1/Runx/Ets/NF-I binding, loss of Ctcf/Nrf1/Yy1 binding, as well as weakened nucleosome footprints.

If we rank pseudobulks along the same differentiation lineage by their pseudo-time, we can reconstruct a “movie” of how TFs and nucleosomes reorganize during differentiation. We observed stepwise establishment of hemoglobin CREs during erythroid differentiation.

We found that instead of having only two states, open vs closed, each CRE can be bound by several distinct TF combinations across cell states/types. Individual CREs thus occupy complex regulatory states undetectable by simply quantifying overall accessibility.

The really exciting part is the combination of seq2PRINT with single cell data. By pseudobulking cells and using low-rank adaptation to tune seq2PRINT to the differences among cell states, we tracked the changes in TF binding across cell types in human hematopoiesis.

Unexpectedly, the deep learning model, which we named seq2PRINT, also captured the binding of “invisible” TFs that do not leave a visible footprint in scATAC-seq. We took this opportunity to build a highly accurate TF binding predictor using seq2PRINT.

Building upon foundational work on DNA sequence models by the @anshulkundaje.bsky.social & others, we trained a deep learning model that predicts footprints from DNA sequence. We examined sequences that drive footprint predictions and saw that the model relies on the organization of TF motifs.

Here is an example region showing multi-scale footprints of CTCF and flanking nucleosomes:

Here we present PRINT. By improving the statistical approach and varying the footprinting kernel size, we detect proteins and complexes across diverse sizes (TFs and nucleosomes).

Footprinting is a powerful method that detects protein binding through protection of bound DNA from enzymes or chemicals. However, it’s typically done on bulk samples, limiting insight into gene regulation in complex systems, and many TFs are “invisible” to footprinting.

cis-regulatory elements (CREs) regulate gene expression by binding to regulatory proteins such as transcription factors (TFs) and histones. A tool that can track the binding dynamics of regulatory proteins with ultra-high resolution across cell states is long sought after.