No population-specific enriched motifs in marker peaks

[mitchTwoTimes]

Hi,

Wondering if the maintainer or anyone else has tips for improving the results of enriched motifs in marker peaks (Sect 12.2). I've used both the default cisbp motifSet, as well as a custom ChIP-Seq annotation. I've pasted an example of what I'm consistently seeing, where the last two populations get little to no enrichment.

My thresholds in cutoff are already pretty liberal:

markersPeaks <- getMarkerFeatures(
  ArchRProj = proj, 
  useMatrix = "PeakMatrix", 
  groupBy = "Clusters_atac_all",
  bias = c("TSSEnrichment", "log10(nFrags)"),
  testMethod = "wilcoxon"
)

enrichMotifs <- peakAnnoEnrichment(
  seMarker = markersPeaks,
  ArchRProj = proj,
  peakAnnotation = "TF_binding", # Motif
  cutOff = "FDR <= 0.1 & Log2FC >= 0.5"
)

Would increasing the resolution of my clusters help? Should I tweak parameters in getMarkerFeatures? Or is there a different interpretation I am missing?

[rcorces]

Its a normalized enrichment, so I dont think its that they get no enrichment, just that their enrichment is tiny compared to your other clusters. You see the same thing in GMP cells in the tutorial:
https://www.archrproject.com/bookdown/motif-enrichment-in-marker-peaks.html
I think this is probably related to both biological and technical variables in the experiment.