openpipeline_rapids
Best-practice workflows for single-cell multi-omics analyses using rapids-singlecell.
bbknn
Compute a batch-balanced nearest-neighbor graph of observations on the GPU.
Wraps rapids-singlecell's rsc.pp.bbknn.
Python
MIT
calculate_qc_metrics
Calculate quality control metrics.
Computes a number of per-cell (.obs) and per-gene (.var) quality control
metrics from a count matrix, such as the number of genes per cell, the
total counts per cell, the number of cells per gene, and the percentage
of counts dropped out per gene.
Wraps rapids_singlecell.pp.calculate_qc_metrics.
Python
MIT
filter_cells
Filter cell outliers based on counts and numbers of genes expressed.
Keep cells that have at least min_counts counts or min_genes genes expressed,
and at most max_counts counts or max_genes genes expressed.
Python
MIT
filter_genes
Filter genes based on number of counts or cells.
Keep genes that have at least min_counts counts or are expressed in at least
min_cells cells, or at most max_counts counts or are expressed in at most
max_cells cells.
Python
MIT
harmony_integrate
Integrate observations over one or more batch/covariate columns on the GPU
using Harmony.
Wraps rapids-singlecell's rsc.pp.harmony_integrate, which runs Harmony batch
integration on a PCA embedding stored in .obsm and writes the corrected
embedding to a new .obsm slot.
Python
MIT
highly_variable_genes
Annotate highly variable genes [Satija15] [Zheng17] [Stuart19] [Lause21] [Andrews19].
Expects logarithmized data, except when flavor is 'seurat_v3', 'seurat_v3_paper',
'pearson_residuals' or 'poisson_gene_selection', in which case count data is
expected.
Depending on flavor, this reproduces the R-implementations of Seurat [Satija15],
Cell Ranger [Zheng17] and Seurat v3 [Stuart19].
Python
MIT
log1p
Logarithmize the data matrix.
Computes X = log(X + 1), where log denotes the natural logarithm unless
a different base is given.
Python
MIT
neighbors
Compute a nearest-neighbor graph of observations on the GPU.
Wraps rapids-singlecell's rsc.pp.neighbors, whose neighbor search relies on
cuVS for fast (approximate) KNN search.
Python
MIT