NatureJClub
Universal cell embedding provides a foundation model for cell biology.
Nature · · Journal Article · Open access
Rosen, Roohani + more
Abstract ↗AI summary
The abstract is read at the publisher; the summary is JClub's.
Universal cell embedding (UCE) creates a single biological latent space that accurately represents 36 million cells across eight species, over 1,000 cell types, and diverse tissues.
- Why it matters: A comprehensive, species-agnostic cell representation is crucial for advancing cell biology, enabling cross-species comparisons and deeper understanding of cellular diversity without extensive data labeling.
- What they did: The team trained UCE using self-supervised learning on a vast dataset of single-cell transcriptomic data, capturing key biological variations while ignoring experimental noise, and applied it to build the Integrated Mega-scale Atlas.
- The result: UCE's embedding space reveals emergent biological insights, such as developmental lineages and cross-species data, facilitating analysis, annotation, and hypothesis generation for single-cell research.
The findingWhy it mattersWhat they didThe result
- Open access
- 15 cites