Single-cell RNAseq using Seurat, FindVariableFeatures and ScaleData before and after doublet prediction.

[FarzanehRah]

Hello,
Thank you so much for these wonderful workshops.
I am trying to use your pipeline and have the question:
In "seurat_01_qc", before predicting the doublet, we need to run NormalizeData, FindVariableFeatures and ScaleData,
in "seurat_02_dim_reduction", we run FindVariableFeatures and ScaleData before the reduction of dimension,
in "seurat_03_integration", before the integration, we run again NormalizeData, FindVariableFeatures and after the integration, ScaleData.
Until now, I thought it was sufficient to run NormalizeData, FindVariableFeatures and ScaleData only once after QC and data filtering, but it seems that my conclusion was not correct.
If we use SCTransform for normalization, should we run it before doublet prediction, before dimension reduction, and before integration as well?
I would also like to know if there is any rules for choosing the appropriate slot of Seurat object for each analysis?

Thank you so much.