Back

CellBench-LS: Benchmark Evaluation of Single-cell Foundation Models for Low-supervision Scenarios

Xu, Y.; Li, Y.; Yuan, Y.; Yu, C.; Zang, Z.

2026-04-05 cell biology
10.64898/2026.04.01.714123 bioRxiv
Show abstract

While single-cell foundation models (SCFMs) have shown promise across various downstream tasks, their generalization performance in label-scarce settings remains a critical bottleneck. The absence of systematic benchmarks for these low-resource scenarios hinders their translation to realworld biomedical research. To bridge this gap, we present CellBench-LS, a comprehensive framework designed to rigorously evaluate SCFMs generalization under low-supervision conditions. This framework employ a stratified evaluation protocol to systematically compare traditional methods and foundation models. We evaluate their zero-shot representational abilities on cell clustering and batch correction tasks, and apply lightweight fine-tuning of task-specific heads for predictive tasks, such as celltype annotation, expression reconstruction, and perturbation prediction. Experimental results demonstrate a biologically stratified landscape, with foundation models showing distinct advantages in tasks critically reliant on celltype recognition, while traditional methods remain competitive in those requiring precise quantification of gene expression patterns. CellBench-LS provides critical guidance for developing more biologically grounded and generalizable computational approaches in single-cell analysis.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

1
Nature Methods
336 papers in training set
Top 0.9%
10.5%
2
Nature Communications
4913 papers in training set
Top 17%
10.2%
3
Nature Machine Intelligence
61 papers in training set
Top 0.3%
6.9%
4
Nucleic Acids Research
1128 papers in training set
Top 3%
6.4%
5
Nature Medicine
117 papers in training set
Top 0.4%
4.9%
6
Genome Biology
555 papers in training set
Top 2%
4.0%
7
Patterns
70 papers in training set
Top 0.2%
3.6%
8
Genome Research
409 papers in training set
Top 0.9%
3.6%
9
Proceedings of the National Academy of Sciences
2130 papers in training set
Top 20%
3.6%
50% of probability mass above
10
PLOS ONE
4510 papers in training set
Top 44%
2.7%
11
Advanced Science
249 papers in training set
Top 7%
2.6%
12
Briefings in Bioinformatics
326 papers in training set
Top 3%
2.4%
13
Cell Systems
167 papers in training set
Top 6%
2.1%
14
PLOS Computational Biology
1633 papers in training set
Top 14%
1.9%
15
Communications Biology
886 papers in training set
Top 6%
1.9%
16
Scientific Reports
3102 papers in training set
Top 53%
1.9%
17
Genomics, Proteomics & Bioinformatics
171 papers in training set
Top 3%
1.7%
18
Cell Reports Medicine
140 papers in training set
Top 4%
1.7%
19
eLife
5422 papers in training set
Top 43%
1.7%
20
Nature Biotechnology
147 papers in training set
Top 5%
1.3%
21
Bioinformatics
1061 papers in training set
Top 8%
1.3%
22
iScience
1063 papers in training set
Top 19%
1.3%
23
Nature Cell Biology
99 papers in training set
Top 3%
1.2%
24
Nature Genetics
240 papers in training set
Top 6%
1.2%
25
Nature
575 papers in training set
Top 14%
0.9%
26
Science Advances
1098 papers in training set
Top 26%
0.9%
27
npj Systems Biology and Applications
99 papers in training set
Top 2%
0.8%
28
Journal of Cell Biology
333 papers in training set
Top 4%
0.8%
29
Nature Computational Science
50 papers in training set
Top 2%
0.7%
30
Cell Reports
1338 papers in training set
Top 35%
0.6%