- First public release of
DropSifton Bioconductor. - Implements SVM-based nuclei selection from single-nucleus RNA-seq (snRNA-seq) data.
- Supports input from 10x Genomics, Optimus H5AD, and dense DGE formats.
- Automated feature extraction, including:
- UMI counts
- % intronic reads
- Mitochondrial content
- Empty gene module scores
- Flexible classifier initialization, handling:
- With/without CellBender remove-background.
- High ambient RNA contamination cases.
- Efficient sparse matrix processing with
Matrixanddata.table. - Comprehensive visualization tools:
- Quality control plots.
- Feature distributions for classifier training.
- Selection probability visualization.
- Test dataset (
svmNucleusCallerInputs) included for reproducible examples.
- Improved input validation & error handling.
- Ensured Bioconductor compliance, including:
- Fully runnable examples.
- Properly formatted documentation.