rPCA IntegrateData causes R to crash with a relatively small dataset

[ChrisSteel-bio]

Hello all,

When running IntegrateData via rPCA on three ~7K cells samples R consistently crashes. I have 32gb of memory, and before running I have ~20gb free. Once integrateData is called memory usage ramps up and hovers around 30gb until the final integration step then is pushed 32 and crashes. From what I understand R shouldn't be struggling so much with relatively small dataset like this? Where am I going wrong? Any help would be greatly appreciated.

After QC filtering I run the below code (sce is a list of the three specimens).

options(future.globals.maxSize = 1024 * 1024 * 1024) # 1 GB SCT <- function(x){ s.genes <- cc.genes.updated.2019$s.genes g2m.genes <- cc.genes.updated.2019$g2m.genes x <- NormalizeData(x) x <- CellCycleScoring(x, s.features = s.genes, g2m.features = g2m.genes, set.ident = TRUE)
x <- SCTransform(x, vars.to.regress = c("percent.mt", "S.Score", "G2M.Score")) x } sce <- lapply(sce, SCT)

names(sce) <- c("combo1","veh1","veh2")

features <- SelectIntegrationFeatures(object.list = sce, nfeatures = 3000)

sce <- PrepSCTIntegration(object.list = sce, anchor.features = features)

sce <- lapply(X = sce, FUN = RunPCA, features = features)

sce <- FindIntegrationAnchors(object.list = sce, normalization.method = "SCT", anchor.features = features, dims = 1:30, reduction = "rpca", k.anchor = 20)

sce.combined <- IntegrateData(anchorset = sce, normalization.method = "SCT", dims = 1:30)