Simulate single-cell RNA-seq count data using the method described in Lun and Marioni "Overcoming confounding plate effects in differential expression analyses of single-cell RNA-seq data".
Usage
lun2Simulate(
params = newLun2Params(),
zinb = FALSE,
sparsify = TRUE,
verbose = TRUE,
...
)Arguments
- params
Lun2Params object containing simulation parameters.
- zinb
logical. Whether to use a zero-inflated model.
- sparsify
logical. Whether to automatically convert assays to sparse matrices if there will be a size reduction.
- verbose
logical. Whether to print progress messages.
- ...
any additional parameter settings to override what is provided in
params.
Details
The Lun2 simulation uses a negative-binomial distribution where the means and
dispersions have been sampled from a real dataset
(using lun2Estimate). The other core feature of the Lun2
simulation is the addition of plate effects. Differential expression can be
added between two groups of plates (an "ingroup" and all other plates).
Library size factors are also applied and optionally a zero-inflated
negative-binomial can be used.
If the number of genes to simulate differs from the number of provided gene parameters or the number of cells to simulate differs from the number of library sizes the relevant parameters will be sampled with a warning. This allows any number of genes or cells to be simulated regardless of the number in the dataset used in the estimation step but has the downside that some genes or cells may be simulated multiple times.
References
Lun ATL, Marioni JC. Overcoming confounding plate effects in differential expression analyses of single-cell RNA-seq data. Biostatistics (2017).
Examples
sim <- lun2Simulate()
#> Getting parameters...
#> Simulating plate means...
#> Simulating library size factors...
#> Simulating cell means...
#> Simulating counts...
#> Creating final dataset...
#> Sparsifying assays...
#> Automatically converting to sparse matrices, threshold = 0.95
#> Converting 'counts' to sparse matrix: estimated sparse size 0.13 * dense matrix
#> Skipping 'CellMeans': estimated sparse size 1.5 * dense matrix
#> Converting 'TrueCounts' to sparse matrix: estimated sparse size 0.13 * dense matrix
#> Done!