Properly prevent output from BASiCSEstimate() and
scDDSimulate() when verbose = FALSE
Minor adjustments to tests to set verbose = FALSE and specify
expected warnings
Replace the package man page with an auto-generated page
Add GitHub usernames and the Bioconductor package URL to description
Add @keywords internal and @noRd to function docs where
needed
Modernise code by restyling using styler and fix other minor issues suggested by BiocCheck
Fixed a bug in splatSimPathDE() where DE factors were not adjusted
based on the path origin (path.from parameter). This affected paths
where the path origin was not the simulation origin (i.e.
path.from != 0), particularly when the path DE was minimal.
With this fix paths should no longer drift towards the origin.
Fixed bugs in splatPopSimulate() where conditional group
assignments were incorrect when batch effects were applied
(from Christina Azodi)
Reduced core dependencies by importing scuttle rather than scater (scater is suggested) and making ggplot2 a suggested dependency.
Fixed a bug in BASiSSimulate() when spike.means is resampled
Fixed bugs in splatPopSimulate() with non-matching rownames and
when sampling batches
The splatPop simulation is now published doi.org/10.1186/s13059-021-02546-1!
Improved initalisation of Params objects (from Wenjie Wang)
Improved fitting of dropout in splatEstimate()
Better initialisation of fitting as suggested by the InferCNV package
Additional fallback method
Bug fixes for the splat simulation
Bug fixes for the the splatPop simulation (from Christina Azodi)
Updates to the splatPop simulation (from Christina Azodi)
Added functionality to simulate directly from empirical values
Added eqtl.coreg parameter to splatPop
Fixed a bug where too many cells were simulated in splatPop with multiple batches
Fixed duplicate cell names in splatPopSimulate
Improved checks for group.prob in SplatParams
Automatically rescale group.prob during setting if it doesn't sum to 1
Substantial updates to the splatPop simulation (from Christina Azodi)
Added ability to simulate data with complex multiplexed sequencing designs
Added simulation of “conditional” effects, where a subset of DE and eQTL effects are applied to only a subset of individuals (e.g. disease vs. healthy samples)
Added the ability to simulate different numbers of cells for each sample, sampled from a gamma distribution.
Updates to the splatPop vignette describing these changes
Logical matrices should now be handled correctly when minimising output SingleCellExperiment objects
Other minor fixes
Add the splatPop simulation. This is a extension to the splat simulation contributed by Christina Azodi and Davis McCarthy that adds population effects. It allows you to specify relatedness between individuals and generate cell-type specific eQTL effects.
Add a batch.rmEffect parameter to the Splat simulation. This allows generation of a paired simulation without any batch effects.
Add a new minimiseSCE function which can be used to remove unneeded information from simulation output (or any SingleCellExperiment)
All simulations now return sparse assay matrices by default when they would be smaller than the equivalent dense matrix. This is controlled by a new sparsify argument.
Users will now be automatically prompted to install packages if they try to use a simulation for which the suggested dependencies are not available
Add checks for cycles in the Splat path.from parameter.
Use alternative algorithm if fitting dropout fails in splatEstimate.
Adjust paths example in vignette.
Replace defunct functions in vignettes.
Minor fixes for compatibility with updates to other packages.
Add the (experimental) Kersplat simulation model. This model incorporates a gene network and other useful features.
Refactor the summariseDiff function and add the KS statistic.
Add variable gene correlation plot to compareSCEs and violins to other comparison plots.
Check for counts assay when estimating from SingleCellExperiment objects.
Fix where simpleSimulate stores parameters.
Fix bugs where parameters were not being passed correctly in BASiCSEstimate and sparseDCEstimate.
Replace the sc_example_counts dataset from scater with the mockSCE function.
Tidy and improve estimation function examples and add checks for suggested packages.
Various fixes for compatibility with updates to other packages.
Add a Splat parameters vignette
Rename the Splat path.length parameter to path.nSteps
Fix a bug with parameter order in setParams
Fix a bug where Splat groups were being simulated in alphanumeric order
Protect against integer overflow in simulation functions
Fix bug and improve normality testing in splatEstLib
Fixes for compatibility with the latest version of BASiCS, BASiCSEstimate now uses the regression method
Fix bug in getLNormFactors when reversing factors less than one
Various updates to tests and documentation
Move scater to Imports and add scater version
Remove lingering references to SCESets
Add option to use a normal distribution for library sizes in Splat simulations
Allow Splat dropout parameters to be specified by experiment, batch, group or cell
Add SparseDC simulation
Rename params in metadata slot of simulation to Params for consistency
Improve and colourise Params print output
Improve test coverage
Various other minor updates and bug fixes
Now published in Genome Biology!
Converted to the SingleCellExperiment object
Added new simulations: BASiCS, mfa, PhenoPath, ZINB-WaVE
Added batch effects to the Splat simulation. This required a change to the SplatParams object.
Improved scDD estimation
Added and improved comparison functions
Improved default Splat parameters and estimation
Improvements to the Lun2Params object
Added addGeneLength function
Updated simulation references
Various other minor updates and bug fixes
Splatter is a package for the simple simulation of single-cell RNA-seq data, including:
Multiple simulation models
Parameter estimation from real data
Functions for comparing simulations and real datasets
Simulation of complex groups and differentiation paths
Package prepared for Bioconductor submission.