EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Capturing cell heterogeneity in representations of cell populations for image-based profiling using contrastive learning

Robert van Dijk iD, John Arevalo iD, Mehrtash Babadi iD, Anne E Carpenter iD, Shantanu Singh iD

PLOS Computational Biology · 11.11.2024 · 20 (11) · e1012547

DOI: 10.1371/journal.pcbi.1012547 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS Computational Biology, online erschienen am 11.11.2024. Verfasst von Robert van Dijk und 4 weiteren Autor:innen. Beteiligte Einrichtung: CellVoyant Technologies, Bristol, United Kingdom. Hauptthemen laut MEDLINE: Single-Cell Analysis, Computational Biology, Image Processing, Computer-Assisted. Die Arbeit wurde bisher 9-mal zitiert (Crossref) und nennt 45 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: NIGMS NIH HHS, NIH, Massachusetts Life Sciences Center.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Image-based cell profiling is a powerful tool that compares perturbed cell populations by measuring thousands of single-cell features and summarizing them into profiles. Typically a sample is represented by averaging across cells, but this fails to capture the heterogeneity within cell populations. We introduce CytoSummaryNet: a Deep Sets-based approach that improves mechanism of action prediction by 30–68% in mean average precision compared to average profiling on a public dataset. CytoSummaryNet uses self-supervised contrastive learning in a multiple-instance learning framework, providing an easier-to-apply method for aggregating single-cell feature data than previously published strategies. Interpretability analysis suggests that the model achieves this improvement by downweighting small mitotic cells or those with debris and prioritizing large uncrowded cells. The approach requires only perturbation labels for training, which are readily available in all cell profiling datasets. CytoSummaryNet offers a straightforward post-processing step for single-cell profiles that can significantly boost retrieval performance on image-based profiling datasets.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Single-Cell AnalysisComputational BiologyImage Processing, Computer-AssistedHumansMachine LearningAlgorithms

Autor:innen und Institutionen

  • CellVoyant Technologies, Bristol, United Kingdom.
  • Imaging Platform, Broad Institute of MIT and Harvard, Cambridge, Massachusetts, United States of America.
  • Data Sciences Platform, Broad Institute of MIT and Harvard, Cambridge, Massachusetts, United States of America.

ORCID-Profile: Robert van Dijk, John Arevalo, Mehrtash Babadi, Anne E Carpenter, Shantanu Singh

Förderung

  • NIGMS NIH HHS
  • NIH
  • Massachusetts Life Sciences Center

Zitieren

APA

van Dijk, R., Arevalo, J., Babadi, M., Carpenter, A. E., & Singh, S. (2024). Capturing cell heterogeneity in representations of cell populations for image-based profiling using contrastive learning. PLOS Computational Biology, 20(11), e1012547. https://doi.org/10.1371/journal.pcbi.1012547

Vancouver

van Dijk R, Arevalo J, Babadi M, Carpenter AE, Singh S. Capturing cell heterogeneity in representations of cell populations for image-based profiling using contrastive learning. PLOS Computational Biology. 2024;20(11):e1012547. doi:10.1371/journal.pcbi.1012547

BibTeX
@article{vanDijk2024,
  author = {van Dijk, Robert and Arevalo, John and Babadi, Mehrtash and Carpenter, Anne E and Singh, Shantanu},
  title = {Capturing cell heterogeneity in representations of cell populations for image-based profiling using contrastive learning},
  journal = {PLOS Computational Biology},
  year = {2024},
  volume = {20},
  number = {11},
  pages = {e1012547},
  doi = {10.1371/journal.pcbi.1012547}
}
Kostenlos starten →