Plate IV.99 · Publications

UniversalEPI: a generalized attention-based deep ensemble model to accurately predict enhancer-promoter interactions across diverse cell types and conditions

Software in Zenodo, 24 March 2025.

SoftwareResearch area: Representations, Single Cell

Authors

  1. Aayush Grover
  2. Lin Zhang
  3. Till Muser
  4. Simeon Haefliger
  5. Minjia Wang
  6. Josephine Yates
  7. Eliezer Van Allen
  8. Fabian Theis
  9. Ignacio Ibarra
  10. Ekaterina Krymova
  11. Valentina Boeva

Abstract

UniversalEPI is an attention-based deep ensemble designed to predict enhancer-promoter interactions up to 2 Mb, which can generalize across unseen cell types using only DNA sequence and chromatin accessibility (ATAC-seq) data as input. The source code and model checkpoints corresponding to the main manuscript are released here.

Cite

@misc{grover2025universalepi,
  title = {{UniversalEPI: a generalized attention-based deep ensemble model to accurately predict enhancer-promoter interactions across diverse cell types and conditions}},
  author = {Grover, Aayush and Zhang, Lin and Muser, Till and Haefliger, Simeon and Wang, Minjia and Yates, Josephine and Van Allen, Eliezer and Theis, Fabian and Ibarra, Ignacio and Krymova, Ekaterina and Boeva, Valentina},
  howpublished = {Zenodo},
  publisher = {Zenodo},
  year = {2025},
  month = mar,
  doi = {10.5281/zenodo.14622040},
  url = {https://doi.org/10.5281/zenodo.14622040}
}
.bib file