CellCycleScoring issue

[honghh2018]

Hi @SeuratLabs, I get a question , when runing the CellCycleScoring on the integrated sample. and i want to regress the impact of cell cycle genes. I wonder which running can be selected on integrated analysis. Whether or not I was runing the CellCycleScoring for each sample before data integrated or after the data integrated. The code showing below,

Runing the CellCycleScoring before integrated

   pancreas.list <- lapply(X = pancreas.list, FUN = function(x) {

x <- CellCycleScoring(x, s.features = s.genes, g2m.features = g2m.genes, set.ident = TRUE) x <- ScaleData(x, vars.to.regress = c("S.Score", "G2M.Score"), features = features) x <- RunPCA(x, features = features, verbose = FALSE) })

Runing the CellCycleScoring after integrated

  anchors <- FindIntegrationAnchors(object.list = pancreas.list, reduction = "rpca", 
                              dims = 1:30)

Integrated <- IntegrateData(anchorset = anchors, dims = 1:30)

Now proceed with downstream analysis (i.e. visualization, clustering) on the integrated dataset. Commands are identical to the standard wor

kflow. Integrated<-CellCycleScoring(x, s.features = s.genes, g2m.features = g2m.genes, set.ident = TRUE) Integrated <- ScaleData(Integrated,vars.to.regress = c("S.Score", "G2M.Score") ) Just like the codes showing above, i cannot select the code with my integrated data analysis to remove the cell cycle impact without your help. Any advices would be appreciated. Best, hanhuihong

[honghh2018]

Hi @All, Can anyone hear this question ?wait for the reply.

[ktrns]

Hi @honghh2018

If I understand your question correctly, this issue might be your answer: https://github.com/satijalab/seurat/issues/2277

Best Katrin

[honghh2018]

Hi @ktrns , Sorry for the unclear description. As my case, i want to remove the cell cycle influence for data integrated of clustering, so, should i run the CellCycleScoring function before samples integrated or after the data integrated. if i run before data integrated, it would be run CellCycleScoring for each sample and running integrated workflow to merge multiple samples into one subsequently, and then runing the CellCycleScoring .
Based above issue, which way should i take , when runing CellCycleScoring . Best, hanhuihong

[torkencz]

Hi,

I would recommend cell cycle regression before integration on the individual samples. If you try the other route as well you can compare the scores and see if you see a significant difference.