Parallelization

Since cellula is mostly based on R/Bioconductor packages, it offers the BiocParallel parallelization backend through its parallel_param argument. In some cases, e.g. the Seurat integration, another type of backend has to be set up separately outside of the function call.

BiocParallel parallelization is implemented where possible, i.e. all the steps of the pipeline where it is sensible to use them (PCA, clustering, integration…).

sce <- integrateSCE(sce, batch = "individual", 
                    method = "fastMNN", 
                    parallel_param = MulticoreParam(workers = 4))