Description Usage Arguments Details Value Comments on the three matrices Additional arguments to Python Author(s) References Examples
Perform RNA velocity calculations with the scVelo package.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29  scvelo(x, ...)
## S4 method for signature '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 signature 'SummarizedExperiment'
scvelo(
x,
...,
assay.X = "counts",
assay.spliced = "spliced",
assay.unspliced = "unspliced"
)
## S4 method for signature 'SingleCellExperiment'
scvelo(x, ..., sf.X = sizeFactors(x), dimred = NULL, use.dimred = NULL)

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 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 
sf.spliced 
A numeric vector containing size factors for the spliced counts for each cell.
Defaults to 
sf.unspliced 
A numeric vector containing size factors for the unspliced counts for each cell.
Defaults to 
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 lowdimensional representation of the cells with number of rows equal to the number of cells in 
ncomponents 
Numeric scalar indicating the number of principal components to obtain.
Only used if 
BPPARAM 
A BiocParallelParam object specifying whether the PCA calculations should be parallelized.
Only used if 
BSPARAM 
A BiocSingularParam object specifying which algorithm should be used to perform the PCA.
Only used if 
assay.X 
An integer scalar or string specifying the assay of 
assay.spliced 
An integer scalar or string specifying the assay of 
assay.unspliced 
An integer scalar or string specifying the assay of 
use.dimred 
String naming the entry of 
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 steadystate 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:
Size factorbased normalization with sf.*
values and normalizeCounts
.
For "X"
, logtransformation is performed as well, while for the others, only scaling normalization is performed.
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.
If dimred=NULL
, the PCA step on the logexpression 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 logexpression 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 PCArelated arguments (ncomponents
, BSPARAM
) and subset.row
.
However, if a lowdimensionality 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 troubleshooting.
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.
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 exonintron 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 reusing 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 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.
Aaron Lun, Charlotte Soneson
Bergen V. et al. (2019). Generalizing RNA velocity to transient cell states through dynamical modeling. bioxRiv, https://www.biorxiv.org/content/10.1101/820936v1
1 2 3 4 5 6 7 8 9 
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.