Skip to contents

RNA velocity with scVelo

Source: R/scvelo.R
scvelo.Rd

Perform RNA velocity calculations with the scVelo package.

Usage

scvelo(x, ...)

# S4 method for class 'ANY'
scvelo(
  x,
  subset.row = NULL,
  sf.X = NULL,
  sf.spliced = NULL,
  sf.unspliced = NULL,
  use.theirs = FALSE,
  mode = c("steady_state", "deterministic", "stochastic", "dynamical"),
  scvelo.params = list(),
  dimred = NULL,
  ncomponents = 30,
  BPPARAM = SerialParam(),
  BSPARAM = bsparam()
)

# S4 method for class 'SummarizedExperiment'
scvelo(
  x,
  ...,
  assay.X = "counts",
  assay.spliced = "spliced",
  assay.unspliced = "unspliced"
)

# S4 method for class 'SingleCellExperiment'
scvelo(x, ..., sf.X = sizeFactors(x), dimred = NULL, use.dimred = NULL)

Arguments

x

A named list of three matrices of the same dimensions where genes are in rows and cells are in columns. The list should contain "spliced" and "unspliced" entries containing spliced and unspliced counts, respectively. It should also contain an "X" entry containing the “usual” count matrix, see details below.

Alternatively, a SummarizedExperiment object containing three such matrices among its assays.

...

For the generic, further arguments to pass to specific methods. For the SummarizedExperiment and SingleCellExperiment methods, further arguments to pass to the ANY method.

subset.row

A character, integer or logical vector specifying the genes to use for the velocity calculations. Defaults to all genes.

sf.X

A numeric vector containing size factors for usual count matrix. Defaults to librarySizeFactors on the "X" matrix in x.

sf.spliced

A numeric vector containing size factors for the spliced counts for each cell. Defaults to librarySizeFactors on the "spliced" matrix in x.

sf.unspliced

A numeric vector containing size factors for the unspliced counts for each cell. Defaults to librarySizeFactors on the "unspliced" matrix in x.

use.theirs

Logical scalar indicating whether scVelo's gene filtering and normalization should be used.

mode

String specifying the method to use to estimate the transcriptional dynamics.

scvelo.params

List of lists containing arguments for individual scVelo functions, see details below.

dimred

A low-dimensional representation of the cells with number of rows equal to the number of cells in x, used to find the nearest neighbors.

ncomponents

Numeric scalar indicating the number of principal components to obtain. Only used if use.theirs=FALSE and dimred=NULL.

BPPARAM

A BiocParallelParam object specifying whether the PCA calculations should be parallelized. Only used if use.theirs=FALSE and dimred=NULL.

BSPARAM

A BiocSingularParam object specifying which algorithm should be used to perform the PCA. Only used if use.theirs=FALSE and dimred=NULL.

assay.X

An integer scalar or string specifying the assay of x containing the usual count matrix.

assay.spliced

An integer scalar or string specifying the assay of x containing the spliced counts.

assay.unspliced

An integer scalar or string specifying the assay of x containing the unspliced counts.

use.dimred

String naming the entry of reducedDims(x) to use for nearest neighbor calculations. Ignored if dimred is supplied.

Value

A SingleCellExperiment is returned containing the output of the velocity calculations. Of particular interest are:

  • the velocity_pseudotime field in the colData, containing the velocity pseudotime for each cell.

  • the velocity entry of the assays, containing the velocity vectors for each cell.

The output will always have number of columns equal to the number of cells supplied in x, though the number of rows will depend on whether any subsetting (if subset.row is supplied) or feature selection (if use.theirs=TRUE) was performed.

Details

This function uses the scVelo Python package (https://pypi.org/project/scvelo/) for RNA velocity calculations. The main difference from the original velocyto approach is that the dynamical model of scVelo does not rely on the presence of observed steady-state populations, which should improve the reliability of the velocity calculations in general applications.

For consistency with other Bioconductor workflows, we perform as many standard steps in R as we can before starting the velocity calculations with scVelo. This involves:

  1. Size factor-based normalization with sf.* values and normalizeCounts. For "X", log-transformation is performed as well, while for the others, only scaling normalization is performed.

  2. Subsetting all matrices to subset.row, most typically to a subset of interest, e.g., highly variable genes. Note that, if set, any subsetting is done after normalization so that library sizes are correctly computed.

  3. If dimred=NULL, the PCA step on the log-expression values derived from the "X" matrix, using the specified BSPARAM to obtain the first ncomponents PCs.

This allows us to guarantee that, for example, the log-expression matrix of HVGs or the PCA coordinates are the same as that used in other applications like clustering or trajectory reconstruction.

Nonetheless, one can set use.theirs=TRUE to directly use the entire scVelo normalization and filtering pipeline. This ignores all of the size factors arguments (sf.*), all of the PCA-related arguments (ncomponents, BSPARAM) and subset.row. However, if a low-dimensionality result is supplied via dimred or use.dimred, the scVelo PCA will always be omitted.

Upon first use, this function will instantiate a conda environment containing the scVelo package. This is done via the basilisk package - see the documentation for that package for trouble-shooting.

Comments on the three matrices

Strictly speaking, only the spliced and unspliced matrices are necessary for the velocity calculations. However, it is often the case that the spliced matrix is not actually the same as a “usual” count matrix (e.g., generated by summing counts across all exons for all mapped genes). This is due to differences in the handling of ambiguous reads that map across exon-intron boundaries, or to genomic regions that can be either exonic or intronic depending on the isoform; the spliced count matrix is more likely to exclude such reads.

We request the usual count matrix as the "X" entry of x to ensure that the PCA and nearest neighbor detection in scVelo are done on the same data as that used in other steps of the large analysis (e.g., clustering, visualization, trajectory reconstruction). In practice, if the usual count matrix is not available, one can often achieve satisfactory results by simply re-using the spliced count matrix as both the "X" and "spliced" entries of x.

Note that if reduced dimensions are supplied in dimred, any "X" entry is only used to create the AnnData object and is not used in any actual calculations.

Additional arguments to Python

Additional arguments to scVelo functions are provided via scvelo.params. This is a named list where each entry is named after a function and is itself a named list of arguments for that function. The following function names are currently recognized:

  • "filter_and_normalize", for gene selection and normalization. This is not used unless use.theirs=TRUE.

  • "moments", for PCA and nearest neighbor detection. The PCA is not performed if dimred or use.dimred is already supplied.

  • "recover_dynamics"

  • "velocity"

  • "velocity_graph"

  • "velocity_pseudotime"

  • "latent_time"

  • "velocity_confidence"

See the scVelo documentation for more details about the available arguments and the examples below for a syntax example.

Supported operating systems and architectures

scVelo dependencies are pinned in a Conda environment to ensure reproducibility.

Differences in packages and versions available from Conda require different environments for different operating systems and architectures. basilisk.utils is used to determine the operating system and architecture of the computer used to run scvelo(), using to the appropriate Conda environment.

As of the latest velociraptor update (24 May 2024):

All environments

tqdm and ipywidgets are installed to suppress the message "Unable to create progress bar".

Linux

scVelo v0.3.2 from conda-forge is used. This is the latest version available to date. libtiff is pinned to v4.5.1 and pillow is pinned to v10.0.0 (https://github.com/conda-forge/libtiff-feedstock/issues/104#issuecomment-2375893029), scipy is pinned to v1.13.1 (https://github.com/theislab/scvelo/issues/1260).

Linux AArch64

scVelo v0.3.2 from conda-forge is used. This is the latest version available to date. libtiff is pinned to v4.5.1 and pillow is pinned to v10.0.0 (https://github.com/conda-forge/libtiff-feedstock/issues/104#issuecomment-2375893029), scipy is pinned to v1.13.1 (https://github.com/theislab/scvelo/issues/1260). Please note that this environment has not been validated yet; it is derived from the environment from Linux (above) and requires additional testing to identify a working environment before pinning all the packages and versions in the environment.

macOS

scVelo v0.3.2 from conda-forge is used. This is the latest version available to date. scipy is pinned to v1.13.1 (https://github.com/theislab/scvelo/issues/1260).

macOS ARM

scVelo v0.3.2 from conda-forge is used. This is the latest version available to date. scipy is pinned to v1.13.1 (https://github.com/theislab/scvelo/issues/1260).

Windows

scVelo v0.2.5 from bioconda is used. Later versions of scVelo depend on jaxlib which is not supported on Windows (https://github.com/google/jax/issues/438). matplotlib is pinned to v3.6.3 (https://github.com/scverse/scanpy/issues/2411), pandas is pinned to v1.5.2 (https://stackoverflow.com/questions/76234312/importerror-cannot-import-name-is-categorical-from-pandas-api-types), and numpy is pinned to v1.21.1 (https://github.com/theislab/scvelo/issues/1109).

References

Bergen, V., Lange, M., Peidli, S. et al. Generalizing RNA velocity to transient cell states through dynamical modeling. Nat Biotechnol 38, 1408–1414 (2020). https://doi.org/10.1038/s41587-020-0591-3

Author

Aaron Lun, Charlotte Soneson, Kevin Rue-Albrecht

Examples

# Using mock data to demonstrate the process:
library(scuttle)
sce1 <- mockSCE()
sce2 <- mockSCE()

spliced <- counts(sce1)
unspliced <- counts(sce2)

out <- scvelo(list(X=spliced, spliced=spliced, unspliced=unspliced))

# make scvelo use 10 rather than the default 30 neighbors to compute moments for velocity estimation:
out <- scvelo(list(X=spliced, spliced=spliced, unspliced=unspliced),
              scvelo.params=list(neighbors=list(n_neighbors=10L)))