Skip to content
DraconiticResearch
Search Sources API
203,234 studies31 sourcesUpdated 35 min ago
PreprintCC BYbioRxiv

scRM: A single-cell foundation model trained on scRNA-seq reads

  • D. He1
  • R. Patro
  • M. Nassar
  • H. Wu
  • Q. Yan
  • G. Xu
  • K. Torkenczy
  • J. Liu
  • S. Tyanova
  1. 1Altos Labs
Read full textPDFDOICite

Abstract

To our knowledge, scRM is the first single-cell foundation model trained on sequencing reads, demonstrating that raw data pretraining is not only feasible but competitive at a fraction of the computational cost. We anticipate that scaling this framework and exploring alternative architectures will unlock biological information inaccessible to count models and extend the dataset-specific pretraining it already makes possible. The source code is available at https://github.com/altos-labs/scRM.

Related

  1. Where single-cell foundation models help: a benchmark of annotation, perturbation and regulatory inference against simple baselines
    F. A. Cupaioli, I. Merelli
    bioRxivPreprintAll rights reservedFull textBioinformatics & computational biologybioRxiv
  2. CellPatch: a flexible and efficient framework for single-cell foundation models empowered by heuristic gene patching
    Hanwen Zhu, Yushun Yuan, Jiyuan Yang +12
    Genome BiologyPeer-reviewedCC BY 4.0Molecular biology & gene regulationGenetics & genomicsCrossref
  3. Benchmarking single-cell foundation models reveals an advantage in the low-data regime for cell-type and disease-state prediction
    H. Benhayon, M. M. Danziger, D. Ben-Zvi +1
    bioRxivPreprintCC BYFull textBioinformatics & computational biologybioRxiv
  4. ProtCell: a foundation model for single-cell proteomics
    B. Hu, Z. Li, Q. Jiang +3
    bioRxivPreprintCC BYFull textBioinformatics & computational biologybioRxiv
  5. Towards a knowledge-enhanced single-cell foundation model
    Hanqing Zhang, Jie Bao, Mei Ma +7
    arXivPreprintCC BY 4.0Full textGenetics & genomicsarXiv

Cite

BibTeXRIS
BibTeX
@misc{he2026scrm,
  title = {{scRM: A single-cell foundation model trained on scRNA-seq reads}},
  author = {D. He and R. Patro and M. Nassar and H. Wu and Q. Yan and G. Xu and K. Torkenczy and J. Liu and S. Tyanova},
  howpublished = {bioRxiv},
  year = {2026},
  month = oct,
  doi = {10.64898/2026.10.24.761590},
  url = {https://www.biorxiv.org/content/10.64898/2026.10.24.761590v1},
  note = {CC BY},
}
RIS
TY  - UNPB
TI  - scRM: A single-cell foundation model trained on scRNA-seq reads
AU  - D. He
AU  - R. Patro
AU  - M. Nassar
AU  - H. Wu
AU  - Q. Yan
AU  - G. Xu
AU  - K. Torkenczy
AU  - J. Liu
AU  - S. Tyanova
PY  - 2026
DA  - 2026/10/24
PB  - bioRxiv
DO  - 10.64898/2026.10.24.761590
UR  - https://www.biorxiv.org/content/10.64898/2026.10.24.761590v1
AB  - To our knowledge, scRM is the first single-cell foundation model trained on sequencing reads, demonstrating that raw data pretraining is not only feasible but competitive at a fraction of the computational cost. We anticipate that scaling this framework and exploring alternative architectures will unlock biological information inaccessible to count models and extend the dataset-specific pretraining it already makes possible. The source code is available at https://github.com/altos-labs/scRM.
ER  -