EUVIMEDEUVIMED entdecken
ÜbersichtsarbeitOpen Access · CC BY 4.0

From modality-specific to compositional foundation models for cell biology

Mojtaba Bahrami, Till Richter, Niklas A Schmacke, Aitana Egea Lavandera, Fabian J Theis

Cell Systems · 2026-02 · 17 (2) · 101534

DOI: 10.1016/j.cels.2026.101534 ↗

Kurz gefasst

Übersichtsarbeit in der Zeitschrift Cell Systems, 2026-02. Verfasst von Mojtaba Bahrami und 4 weiteren Autor:innen. Beteiligte Einrichtung: Institute of Computational Biology, Helmholtz Center Munich, Munich, Germany; TUM School of Life Sciences Weihenstephan, Technical University of Muni…. Hauptthemen laut MEDLINE: Cell Biology, Models, Biological. Die Arbeit wurde bisher 9-mal zitiert (Crossref) und nennt 198 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: European Research Council, Joachim Herz Stiftung, Helmholtz Association.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deriving principles governing cell biology from single-cell measurements across modalities, called multimodal modeling, can advance our understanding of cellular states in health and disease. Realizing the full potential of multimodal models requires learning generalizable representations across data types, diseases, and biological contexts. This perspective examines the potential of compositional AI as a modular design approach for constructing multimodal foundation models that unify biological modalities-such as chromatin accessibility, protein abundance, spatial transcriptomics, microscopy imaging, and textual annotations-into cohesive representations of cellular behavior. We present key deep learning modeling approaches, along with transformer-based attention strategies to implement them, while addressing challenges posed by limited data availability and structural differences between modality representations. We also discuss how to connect and align partially overlapping multimodal measurements to build a comprehensive representation space. By synthesizing these technical advancements, we chart a path toward agentic virtual cell models, offering insights into opportunities, limitations, and future directions for leveraging multimodal AI to decode the complexity of cellular systems.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Cell BiologyModels, BiologicalHumansSingle-Cell AnalysisDeep LearningAnimals

Stichwörter der Autor:innen

compositional AIsingle-cell foundation modelssingle-cell multi-omics

Autor:innen und Institutionen

  • Institute of Computational Biology, Helmholtz Center Munich, Munich, Germany; TUM School of Life Sciences Weihenstephan, Technical University of Munich, Munich, Germany.
  • Institute of Computational Biology, Helmholtz Center Munich, Munich, Germany; TUM School of Computation, Information and Technology, Technical University of Munich, Munich, Germany.
  • Institute of Computational Biology, Helmholtz Center Munich, Munich, Germany; Gene Center, Department of Biochemistry, Ludwig-Maximilians-Universität München, Munich, Germany.
  • Institute of Computational Biology, Helmholtz Center Munich, Munich, Germany.
  • Institute of Computational Biology, Helmholtz Center Munich, Munich, Germany; TUM School of Life Sciences Weihenstephan, Technical University of Munich, Munich, Germany; TUM School of Computation, Information and Techno…

Förderung

  • European Research Council
  • Joachim Herz Stiftung
  • Helmholtz Association
  • European Commission

Zitieren

APA

Bahrami, M., Richter, T., Schmacke, N. A., Egea Lavandera, A., & Theis, F. J. (2026). From modality-specific to compositional foundation models for cell biology. Cell Systems, 17(2), 101534. https://doi.org/10.1016/j.cels.2026.101534

Vancouver

Bahrami M, Richter T, Schmacke NA, Egea Lavandera A, Theis FJ. From modality-specific to compositional foundation models for cell biology. Cell Systems. 2026;17(2):101534. doi:10.1016/j.cels.2026.101534

BibTeX
@article{Bahrami2026,
  author = {Bahrami, Mojtaba and Richter, Till and Schmacke, Niklas A and Egea Lavandera, Aitana and Theis, Fabian J},
  title = {From modality-specific to compositional foundation models for cell biology},
  journal = {Cell Systems},
  year = {2026},
  volume = {17},
  number = {2},
  pages = {101534},
  doi = {10.1016/j.cels.2026.101534}
}
Kostenlos starten →