Inspired by recent approaches for natural language processing and computer vision, we developed Annotatability, a framework that analyzes deep neural network training dynamics to interpret pre-annotated single-cell and spatial omics data. Annotatability identified erroneous annotations and ambiguous cell states, inferred trajectories from binary labels, and revealed underlying biological signals.
This is a preview of subscription content, access via your institution
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$32.99 / 30 days
cancel any time
Subscribe to this journal
Receive 12 digital issues and online access to articles
$119.00 per year
only $9.92 per issue
Buy this article
- Purchase on SpringerLink
- Instant access to the full article PDF.
USD 39.95
Prices may be subject to local taxes which are calculated during checkout
References
Kim, T. et al. scReClassify: post hoc cell type classification of single-cell rNA-seq data. BMC Genom. 20, 913 (2019). The paper discusses challenges in cell type identification from scRNA-seq data, focusing on the potential for mislabeling due to reliance on human expertise and computational methods.
Burkhardt, D. B. et al. Quantifying the effect of experimental perturbations at single-cell resolution. Nat. Biotechnol. 39, 619–629 (2021). A method developed to quantify the effects of perturbations at the single-cell level.
Swayamdipta, S. et al. Dataset cartography: mapping and diagnosing datasets with training dynamics. In Proc. 2020 Conference on Empirical Methods in Natural Language Processing (eds Webber, B. et al.) 9275–9293 (Association for Computational Linguistics, 2020). A paper that presents ‘data maps’ in the context of natural language processing and used training dynamics to assess dataset quality, which revealed ambiguous, easy-to-learn and hard-to-learn examples.
Gayoso, A. et al. A Python library for probabilistic analysis of single-cell omics data. Nat. Biotechnol. 40, 163–166 (2022). Annotatability used the 10x Genomics PBMC single-cell dataset, integrated and pre-processed by this library (scvi-tools), to reveal both errors and ambiguities in the original cell type annotations.
Tritschler, S. et al. A transcriptional cross species map of pancreatic islet cells. Mol. Metab. 66, 101595 (2022). Annotatability used this single-cell dataset to quantify the damage to individual mSTZ β cells, infer disease-related cell states and dissect treatment responses for individual cells.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This is a summary of: Karin, J. et al. Interpreting single-cell and spatial omics data using deep neural network training dynamics. Nat. Comput. Sci. https://doi.org/10.1038/s43588-024-00721-5 (2024).
Rights and permissions
About this article
Cite this article
Deep learning training dynamics analysis for single-cell data. Nat Comput Sci 4, 886–887 (2024). https://doi.org/10.1038/s43588-024-00728-y
Published:
Version of record:
Issue date:
DOI: https://doi.org/10.1038/s43588-024-00728-y
