ProtCell: a foundation model for single-cell proteomics

Abstract

Single-cell proteomics provides a direct functional readout of cellular states, but targeted assays measure limited and non-identical protein panels, fragmenting information across experiments. Here, we present ProtCell, a foundation model developed using approximately 380 million cells from 98 datasets for pretraining and evaluation. ProtCell combines protein-identity and continuous-abundance embeddings in a transformer encoder and learns cell- and protein-level representations through masked abundance modeling. ProtCell first connects fragmented profiles through transferable annotation and batch correction while preserving biological organization. It then extends limited profiles through abundance reconstruction and inference of an unmeasured cytotoxic protein program enriched in natural killer cells and memory CD8⁺ T cells. The learned protein dependencies further guide the selection of compact panels that retain information for profile reconstruction and biological classification. ProtCell thus provides a framework for connecting fragmented proteomic profiles, extending information beyond measured panels and guiding more informative measurements.

Single-Cell Proteomics Reveals Heterogeneous and Bimodal Proteome Responses to DNA Damage
K. R. Adoni, G. H. Charlton, J. E. Ditcham +5
bioRxiv20 Jul 2026PreprintAll rights reservedFull textCell biologybioRxiv
Single Cell Proteomics reveals partial uncoupling of the Glycerol-3-Phosphate shuttle to compensate for Electron Transport Chain dysfunction in Alexander Disease GFAP mutations
C. A. van Gelder, E. Daoutsali, Y. Attour +8
bioRxiv5 Oct 2026PreprintAll rights reservedFull textNeurosciencebioRxiv
Protocol for Single-Cell Patch-Proteomics: identify specific proteins associated with electrophysiological properties from individual neurons
J. C. Althaus, Y. Cao, Y. Song +5
bioRxiv24 Sep 2026PreprintCC BY-NC-NDFull textNeurosciencebioRxiv
Single-cell proteomics reveals cell-type-specific functional coordination in PBMCs
L. Koury, A. Leduc, S. Khan +4
bioRxiv11 Sep 2026PreprintCC BY-NC-NDFull textSystems & synthetic biologybioRxiv
scRM: A single-cell foundation model trained on scRNA-seq reads
D. He, R. Patro, M. Nassar +6
bioRxiv24 Oct 2026PreprintCC BYFull textBioinformatics & computational biologybioRxiv

Cite

BibTeX RIS

BibTeX

@misc{hu2026protcell,
  title = {{ProtCell: a foundation model for single-cell proteomics}},
  author = {B. Hu and Z. Li and Q. Jiang and Q. Yin and Z. Liu and R. Jiang},
  howpublished = {bioRxiv},
  year = {2026},
  month = oct,
  doi = {10.64898/2026.10.22.760932},
  url = {https://www.biorxiv.org/content/10.64898/2026.10.22.760932v1},
  note = {CC BY},
}

RIS

TY  - UNPB
TI  - ProtCell: a foundation model for single-cell proteomics
AU  - B. Hu
AU  - Z. Li
AU  - Q. Jiang
AU  - Q. Yin
AU  - Z. Liu
AU  - R. Jiang
PY  - 2026
DA  - 2026/10/22
PB  - bioRxiv
DO  - 10.64898/2026.10.22.760932
UR  - https://www.biorxiv.org/content/10.64898/2026.10.22.760932v1
AB  - Single-cell proteomics provides a direct functional readout of cellular states, but targeted assays measure limited and non-identical protein panels, fragmenting information across experiments. Here, we present ProtCell, a foundation model developed using approximately 380 million cells from 98 datasets for pretraining and evaluation. ProtCell combines protein-identity and continuous-abundance embeddings in a transformer encoder and learns cell- and protein-level representations through masked abundance modeling. ProtCell first connects fragmented profiles through transferable annotation and batch correction while preserving biological organization. It then extends limited profiles through abundance reconstruction and inference of an unmeasured cytotoxic protein program enriched in natural killer cells and memory CD8^+ T cells. The learned protein dependencies further guide the selection of compact panels that retain information for profile reconstruction and biological classification. ProtCell thus provides a framework for connecting fragmented proteomic profiles, extending information beyond measured panels and guiding more informative measurements.
ER  -

Abstract

Related

Cite