FindVariableFeatures post Integration

[santiiiiiii]

Hi,

I have integrated 9 datasets together using this method: https://satijalab.org/seurat/articles/integration_rpca.html
I would like to modify the features used to run PCA

DefaultAssay(combined) <- "integrated"
combined <- ScaleData(combined,verbose = FALSE)
combined <- RunPCA(combined, npcs = 20, features = features, verbose = FALSE)
combined <- RunUMAP(combined, reduction = "pca", dims = 1:20, n.neighbors=10)

Which option should I use?

Option 1:
DefaultAssay(combined) <- "RNA"
combined <- FindVariableFeatures(combined, selection.method = 'vst', nfeatures = 2000)
features <- VariableFeatures(combined)

Option 2: This works but I get a warning
DefaultAssay(combined) <- "Integrated"
combined <- FindVariableFeatures(combined, selection.method = 'vst', nfeatures = 2000)
features <- VariableFeatures(combined)

Warning messages:
1: In FindVariableFeatures.Assay(object = object[[assay]], selection.method = selection.method, :
selection.method set to 'vst' but count slot is empty; will use data slot instead
2: In eval(predvars, data, env) : NaNs produced
3: In hvf.info$variance.expected[not.const] <- 10^fit$fitted :
number of items to replace is not a multiple of replacement length

thank you!