Questions about peak calling and data harmonization

[fstrijth]

Hello, I am new to scATAC-seq data, so I have two basic questions. For context, I am working on a 10x multi-batch multi-omics data set. I am trying out pycistopic with the intention of running SCENIC+ later. Since I have multiple batches of data I am following a mix of the tutorial on the website and this old one (https://github.com/aertslab/pycisTopic/blob/main/notebooks/Cortex_pycisTopic.ipynb).

First question, what is the added advantage of doing peak calling yourself compared to using the 10x cell ranger output?

Second question: I did my preliminary analysis in Seurat (WNN analysis of RNA and ATAC), where I noticed a strong batch effect in only the ATAC data, and when I applied harmony I lost a lot of signal. How would you evaluate the result of harmonization in cistopic analysis?