SketchData with BPcells taking days to run

[eheinrichs]

Hi,

I have a dataset of about 1.5 million cells that consists of around 100 samples. They are all being loaded in from matrices using BPcells and I am following along with the commands in the "Sketch integration using a 1 million cell dataset from Parse Biosciences" vignette.

Everything runs at a reasonable speed until I get to the SketchData() call. I have traced the function and found that the part that seems to be by far the slowest is:

for (lyr in layers.data) { try( expr = VariableFeatures(object = sketched, method = "sketch", layer = lyr) <- VariableFeatures(object = object[[assay]], layer = lyr), silent = TRUE ) }

which is taking about an hour per sample. Does anyone know what this bit does? It seems to me that it is setting the variable features of the sketched object to be the same as the variable features of the whole object, which should be fast since the variable features of the whole object are already calculated. That being said, I can't seem to find documentation of what the method = "sketch" portion does, so I am not sure how that impacts it.

I am using Seurat 5.1.0 and BPcells 0.2.0