UCell: Robust and scalable single-cell gene signature scoring

Abstract

UCell is an R package for evaluating gene signatures in single-cell datasets. UCell signature scores, based on the Mann-Whitney U statistic, are robust to dataset size and heterogeneity, and their calculation demands less computing time and memory than other available methods, enabling the processing of large datasets in a few minutes even on machines with limited computing power. UCell can be applied to any single-cell data matrix, and includes functions to directly interact with Seurat objects. The UCell package and documentation are available on GitHub at https://github.com/carmonalab/UCell.

Keywords: Cell type; Gene set enrichment; Gene signature; Module scoring; Single-cell.

Fig. 1

Evaluating T cell signatures using UCell. A) UMAP representation of T subsets from the single-cell dataset by Hao et al. . B) UCell score distribution in UMAP space for five gene signatures (listed in Table 1) evaluated using UCell. C-D) Comparison of UCell score (C) and Seurat’s AddModuleScore (D) distributions for a two-gene CD8 T cell signature (CD8A, CD8B), evaluated on the complete T cell dataset (black outlines), or on the subset of CD8 T cells only (red outlines); UCell scores for CD8 T cell have the same distribution in the complete or subset dataset, while AddModuleScores are highly dependent on dataset composition. E-F) Running time (E) and peak memory (F) for UCell and AUCell (which produces similar results) on datasets of different sizes show that UCell is about three times faster and requires up to ten times less memory on large (>10⁴) single-cell datasets. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)