Lab 15 — Single-Cell Preprocessing in R
Key concepts and tools
SingleCellExperiment(SCE) object:colData,rowData,reducedDimsscran,scater— R single-cell analysis packages- Quality control:
perCellQCMetrics,quickPerCellQC - Normalization:
computeSumFactors(scran pooling) - Feature selection: modeling gene variance, selecting highly variable genes
- PCA, UMAP via
scater - Graph-based clustering:
buildSNNGraph,igraphcommunity detection - Marker gene detection:
findMarkers - R Markdown (
.Rmd)
The R-side counterpart to Lab 13. Working in an R Markdown document, you will import the same single-cell counts data into a SingleCellExperiment object and repeat the core preprocessing steps using the Bioconductor ecosystem: QC filtering, sum-factor normalization with scran, highly variable gene selection, dimensionality reduction, and graph-based clustering. The lab also prepares the data structure needed for the pseudobulk analysis in Lab 16, introducing the concept of collapsing single-cell observations back to sample-level counts.