Merging datasets + Differential gene expression with batch correction

[SzymonSzymura]

Hello,
I am not sure if this topic has been discussed already (if it has, I apologize for redundancy). Also sorry for long description but my experimental design is little complicated. I would really appreciate your patience and help:
I need to perform differential gene expression analysis of before vs. after treatment using multiple patient samples that were run on different chromium chips (before and after were run on one chip for given patient, but each patient is on different chip).
I follow integration instructions on Satija website with CCA algorithm. The resulting Seurat object has two main assays: integrated (with 2000 variable genes used for anchoring) and RNA. For the purpose of visualization with UMAP, integrated assay looks nice, with no batch effect. However, for the purpose of differential gene expression before vs after I don't fully understand which slots in Seurat object contain which sort of raw or modified data and also which slots are used for differential gene expression. I assume 'RNA' assay which I want to use for DE has unmodified raw data? I can set DefaultAssay to RNA and then do ScaleData with vars.to.regress set to my batches to regress out batch effect. After scaling, RNA assay will contain 'counts' and 'data' with scale.data. How can I use this scaled and corrected for batch effects RNA data to perform differential gene expression? Which setting in FindMarkers should be used? Assay = RNA, slot = scale.data ? Will this work with Default Wilcoxon test to yield the result that I am hoping to analyze?
I would really appreciate help.
Thank you.