Marker Features variance by cluster, sample, and treatment group

Question

[naomiboldon]

Hello! I am comparing scATAC-seq gene markers in ArchR using Seurat's getMarkerFeatures function. After applying Log2FC and FDR thresholds, I get different output depending on if I groupBy Clusters, Samples, or Treatment Group. This is especially confusing when I want to compare individual sample genes (within a treatment group) to that output by treatment group (a subset group of samples created before running getMarkerFeatures). How have you reconcilled these differences using significant gene markers? Thanks so much for your insight!