Compound models and Pearson residuals for single-cell RNA-seq data without UMIs
Recent work employed Pearson residuals from Poisson or negative binomial models to normalize UMI-based scRNA-seq data. To extend this approach to non-UMI data, we model the amplification step with a compound distribution: we assume that captured RNA molecules follow a negative binomial distribution and are replicated following an amplification distribution. This model leads to compound Pearson residuals, yielding meaningful gene selection and embeddings of Smart-seq2 datasets. Furthermore, we show that amplification distributions across several sequencing protocols can be described by a broken power law. The resulting compound model captures previously unexplained overdispersion and zero-inflation patterns in non-UMI data.
Authors
- Philipp Berens (ORCID: https://orcid.org/0000-0002-0199-4727)
- Dmitry Kobak (ORCID: https://orcid.org/0000-0002-5639-7209)
- Christoph Ziegenhain (ORCID: https://orcid.org/0000-0003-2208-4877)
- Jan Lause (ORCID: https://orcid.org/0000-0003-0946-412X)
- Leonard Hartmanis
Institutions
- HOGENT University of Applied Sciences and Arts (BE)
- Karolinska Institutet (SE)
- Ghent University (BE)
- VIB-UGent Center for Plant Systems Biology (BE)
- VIB-UGent Center for Inflammation Research (BE)
- Hertie Institute for Clinical Brain Research (DE)
- VIB-UGent Center for Medical Biotechnology (BE)
- University of Tübingen (DE)
Publication Details
- Journal
- Genome biology
- Published
- 2026-06-27
- DOI
- https://doi.org/10.1186/s13059-026-04161-4
- Primary Topic
- Single-cell and spatial transcriptomics
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- European Commission
- Eberhard Karls Universität Tübingen
- Vetenskapsrådet
- National Institute of Neurological Disorders and Stroke