| name | bioconductor-singler |
| description | Performs unbiased cell type recognition from single-cell RNA sequencing data, by leveraging reference transcriptomic datasets of pure cell types to infer the cell of origin of each single cell independently. |
| when_to_use | Use when: Annotating unlabelled single-cell datasets using pre-labelled reference datasets (e.g., HumanPrimaryCellAtlasData()).; Propagating biological knowledge to new datasets in an automated manner without manually interpreting clusters.; Resolving closely related cell type labels using fine-tuning.. Not for: For unsupervised clustering without a reference (use scran instead because SingleR requires known labels).; For manual marker gene definition and cluster interpretation (use Seurat instead because SingleR automates this using a reference). |
| user-invocable | false |
SingleR
Dependencies & Environment
Package-intrinsic requirements from the Bioconductor landing page — reproduce in any R environment.
- Version: 2.14.0 · Bioconductor: 3.23 · R: ≥ 4.6
- Depends: SummarizedExperiment
- Imports: Matrix, BiocGenerics, S4Vectors, DelayedArray, Rcpp, beachmat
- System requirements: C++17
- Install:
BiocManager::install("SingleR")
When to Use
- Annotating unlabelled single-cell datasets using pre-labelled reference datasets (e.g.,
HumanPrimaryCellAtlasData()).
- Propagating biological knowledge to new datasets in an automated manner without manually interpreting clusters.
- Resolving closely related cell type labels using fine-tuning.
When NOT to Use
- For unsupervised clustering without a reference (use
scran instead because SingleR requires known labels).
- For manual marker gene definition and cluster interpretation (use
Seurat instead because SingleR automates this using a reference).
Data Requirements
- Test dataset: Unlogged counts are acceptable, as only ranks are used by
SingleR().
- Reference dataset: Must be normalized and log-transformed (e.g., using
normalizeRnaCounts.se()).
Key Parameters
- test: The unlabelled single-cell dataset to be annotated.
- ref: The pre-labelled reference dataset.
- labels: The cell type labels corresponding to the reference dataset.
- assay.type.test: The assay index or name to use from the test dataset.
- de.method: The marker detection mode (e.g.,
"wilcox" for single-cell references).
Best Practices
- Perform cell-based quality control before running
SingleR().
- Normalize and log-transform reference datasets using
normalizeRnaCounts.se().
- Inspect assignment confidence using
plotScoreHeatmap() and plotDeltaDistribution().
- Remove poor-quality or ambiguous assignments using
pruneScores().
- Examine the expression of marker genes for each label in the test dataset using
plotMarkerHeatmap().
Common Pitfalls
- Using raw counts for the reference (fix: log-transform and normalize the reference first using
normalizeRnaCounts.se()).
- Default marker detection failing on low-coverage data (fix: use
de.method="wilcox" to consider variance of expression across cells).
- Uncertain assignments for cells whose true label does not exist in the reference (fix: inspect per-cell deltas and use
pruneScores() to replace low-quality assignments with NA).
Alternatives
Seurat (for manual marker gene definition and cluster interpretation).scmap (for projection-based single-cell annotation).
Citations
- Aran, D., et al. 2019. "Reference-based analysis of lung single-cell sequencing reveals a transitional profibrotic macrophage." Nat. Immunol. 20 (2): 163–72.
- Grun, D., et al. 2016. "De Novo Prediction of Stem Cell Identity using Single-Cell Transcriptome Data." Cell Stem Cell 19 (2): 266–77.
References
- Homepage: bioconductor.org/packages/SingleR
- Vignette: vignette_0_9645ffcc.txt
Run this on BioMate
This skill is the knowledge layer — when, why, and how to use singler. To run this analysis on your own data with managed compute, automated QC, and reproducible outputs, use BioMate — free to start.
▶ Open singler on BioMate →
