Gene set selection for classification with single-cell RNA-seq
Marker genes are important for analyzing and interpreting single-cell RNA-seq (scRNAseq) experiments. They are used to identify and categorize cell types and can also be useful for the design of downstream experiments. We present CellSieve, which tackles the problem of finding effective gene sets rather than distilling markers one at a time. While several "combinatorial" approaches to gene set selection have been proposed, we find that our approach, which leverages a combination of multinomial logistic regression and linear programming, is much more effective at finding small gene sets that can classify cell types.
Authors
- Lior S. Pachter (ORCID: https://orcid.org/0000-0002-9164-6231)
- Nikhila P. Swarna (ORCID: https://orcid.org/0000-0003-3300-0114)
Institutions
- California Institute of Technology (US)
Publication Details
- Journal
- bioRxiv (Cold Spring Harbor Laboratory)
- Published
- 2026-11-18
- DOI
- https://doi.org/10.64898/2026.11.16.767160
- Primary Topic
- Single-cell and spatial transcriptomics
- Type
- preprint
Funders
- National Institutes of Health
- National Human Genome Research Institute