could you have example how to use fitGAM with 20 cores

[FADHLyemen]

I need to use 20 cores for fitGAM function:
I did
options(MulticoreParam=quote(MulticoreParam(workers=20)))

then run fitGAM

sce_PT <- fitGAM(sce_PT3, conditions = factor(colData(sce_PT3)$t2dctrl),
                 genes = rownames(sce_PT3)[rownames(sce_PT3) %in% hvg_genes ],BPPARAM = BiocParallel::bpparam())

but still, fitGAM used all the cores. any suggestions?

[HectorRDB]

Hi @FADHLyemen
Please refer to the vignette where we discuss parallelization, it should help you sort this out.
Best

[FADHLyemen]

I did

BPPARAM <- BiocParallel::bpparam()
BPPARAM 
BPPARAM$workers <- 10 # use 10 cores
sce_subset <- fitGAM(counts=sce_subset@assays@data$counts,pseudotime =t,cellWeights=w, 
                     conditions = factor(colData(sce_subset)$sampletype),
                 genes = rownames(sce_subset)[rownames(sce_subset) %in% hvg_genes_Tcells ],
               parallel=TRUE, BPPARAM = BPPARAM
             )

But still, fitGAM used all cores.

[HectorRDB]

Just to check, how do you verify how many cores are used? Depending on your settings, R might do multi-threading on top of parallelization, which would make more than 10 R processes appear, even though only 10 cores are used for example.