comparing TF activity changes pre/post-knockout across multiple cell lines using run_ulm and limma

[attaff]

Dear decoupleR team,

good afternoon and I hope you are doing well !!

I'm using decoupleR to infer TF activities from bulk RNA-seq data in a knockout experiment across two cell lines , each with pre- and post-knockout replicates. I want to identify TFs with common activity changes (intersection of significant changes in both cell lines).

Question 1: What's the recommended workflow?
Option A: Compute per-sample TF activities with run_ulm on the full expression matrix, then use limma to compare activities (pre vs. post) per cell line, and intersect results.
Option B: Run limma DE per cell line (pre vs. post) to get t-values, then input those t-values to run_ulm for differential TF activities, and intersect.

Question 2: When using limma t-values as input to run_ulm, what does the p_value in the output represent? For filtering common changes, should I use p_value < 0.05 on TF activities from each cell line, then intersect (with FDR adjustment)?

Thanks for your help!,
Jiachen