adverSCarial: assessing the vulnerability of single-cell RNA-sequencing classifiers to adversarial attacks
Bioinformatics · 2025
DOI: 10.1093/bioinformatics/btaf168 ↗Kurz gefasst
Fachartikel in der Zeitschrift Bioinformatics, 2025. Verfasst von Ghislain Fievet und 3 weiteren Autor:innen. Beteiligte Einrichtung: INSERM U1256, Nutrition, Genetics, and Environmental Risk Exposure (NGERE), University of Lorraine, Nancy, 54500, France. Hauptthemen laut MEDLINE: Single-Cell Analysis, Sequence Analysis, RNA, Machine Learning. Die Arbeit nennt 26 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Contrat de Plan Etat-Régions-Biologie Santé du site Lorrain, Contrat de Plan Etat-Régions—Biologie Santé du site Lorrain.
Abstract (Originalsprache)
Motivation Several machine learning (ML) algorithms dedicated to the detection of healthy and diseased cell types from single-cell RNA sequencing (scRNA-seq) data have been proposed for biomedical purposes. This raises concerns about their vulnerability to adversarial attacks, exploiting threats causing malicious alterations of the classifiers’ output with defective and well-crafted input. Results With adverSCarial, adversarial attacks of single-cell transcriptomic data can easily be simulated in a range of ways, from expanded but undetectable modifications to aggressive and targeted ones, enabling vulnerability assessment of scRNA-seq classifiers to variations of gene expression, whether technical, biological, or intentional. We exemplify the usefulness and performance with a panel of attack modes proposed in adverSCarial by assessing the robustness of five scRNA-seq classifiers, each belonging to a distinct class of ML algorithm, and explore the potential unlocked by exposing their inner workings and sensitivities on four different datasets. These analyses can guide the development of more reliable models, with improved interpretability, usable in biomedical research and future clinical applications. Availability and implementation adverSCarial is a freely available R package accessible from Bioconductor: https://bioconductor.org/packages/adverSCarial/ or https://doi.org/10.18129/B9.bioc.adverSCarial. A development version is available at https://github.com/GhislainFievet/adverSCarial.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- INSERM U1256, Nutrition, Genetics, and Environmental Risk Exposure (NGERE), University of Lorraine, Nancy, 54500, France.
- Department of Biological Hematology, Laboratory Center, University Hospital of Nancy, Nancy, 54500, France.
- Department of Molecular Medicine, Division of Biochemistry, Molecular Biology, and Nutrition, University Hospital of Nancy, Nancy, 54500, France.
ORCID-Profile: Sébastien Hergalant
Förderung
- Contrat de Plan Etat-Régions-Biologie Santé du site Lorrain
- Contrat de Plan Etat-Régions—Biologie Santé du site Lorrain
Zitieren
APA
Fievet, G., Broséus, J., Meyre, D., & Hergalant, S. (2025). adverSCarial: assessing the vulnerability of single-cell RNA-sequencing classifiers to adversarial attacks. Bioinformatics. https://doi.org/10.1093/bioinformatics/btaf168
Vancouver
Fievet G, Broséus J, Meyre D, Hergalant S. adverSCarial: assessing the vulnerability of single-cell RNA-sequencing classifiers to adversarial attacks. Bioinformatics. 2025. doi:10.1093/bioinformatics/btaf168
BibTeX
@article{Fievet2025,
author = {Fievet, Ghislain and Broséus, Julien and Meyre, David and Hergalant, Sébastien},
title = {adverSCarial: assessing the vulnerability of single-cell RNA-sequencing classifiers to adversarial attacks},
journal = {Bioinformatics},
year = {2025},
doi = {10.1093/bioinformatics/btaf168}
}
EUVIMED