scrna_pipeline: scrna_pipeline

In pipeComp: pipeComp pipeline benchmarking framework

Description

The 'PipelineDefinition' for the default scRNAseq clustering pipeline, with steps for doublet identification, filtering, normalization, feature selection, dimensionality reduction, and clustering. Alternative arguments should be character, numeric or logical vectors of length 1 (e.g. the function name for a method, the number of dimensions, etc). The default pipeline has the following steps and arguments:

• doublet: 'doubletmethod' (name of the doublet removal function)

• filtering: 'filt' (name of the filtering function, or filter string)

• normalization: 'norm' (name of the normalization function)

• selection: 'sel' (name of the selection function, or variable of rowData on which to select) and 'selnb' (number of features to select)

• dimreduction: 'dr' (name of the dimensionality reduction function) and 'maxdim' (maximum number of components to compute)

• clustering: 'clustmethod' (name of the clustering function), 'dims' (number of dimensions to use), 'k' (number of nearest neighbors to use, if applicable), 'steps' (number of steps in the random walk, if applicable), 'resolution' (resolution, if applicable), 'min.size' (minimum cluster size, if applicable). If using the 'scrna_alternatives.R' wrappers, the dimensionality can be automatically estimated by specifying 'dims = "method_name"'.

Usage

scrna_pipeline(saveDimRed = FALSE, pipeClass = c("seurat", "sce"))

Arguments

saveDimRed	Logical; whether to save the dimensionality reduction for each analysis (default FALSE)
pipeClass	'sce' or 'seurat'; which object class to use throughout the pipeline. Note that the 'alternatives' functions have to be built around the chosen class. Given that, if running the 'scrna_alternatives', the class of whole pipeline is determined by the output of the filtering, only this step is affected by this option.

Value

A 'PipelineDefinition' object to be used with 'runPipeline'.

Examples

pip <- scrna_pipeline()
pip

pipeComp documentation built on Nov. 8, 2020, 7:35 p.m.