Using Footprint Score to Infer TF-Mediated Repression (question)

[QiliShi]

Hi ，

Thank you for this powerful tool.

I have a question regarding the biological interpretation of the footprint score, especially in the context of transcriptional repression.

Since the footprint score is defined as the difference between flanking signal (W flank) and the middle signal (W middle), it reflects the likelihood of transcription factor binding at a given site. Binding generally creates a depletion in the center of the motif, resulting in a strong footprint.

My question is:
Can this score also be used to infer the repressive effect of a transcription factor on its target genes?
In other words:

If a TF acts as a repressor, would its knockout or knockdown lead to a reduction in the footprint score at its binding sites (due to loss of binding)?

Conversely, would overexpression of a repressive TF increase the footprint score?

I understand that footprint score primarily indicates binding, not regulatory function, but I'm wondering if changes in the footprint score across conditions (e.g., KO vs. WT) can suggest direct repressive effects on nearby genes.

[hschult]

Hi @QiliShi,

That's an interesting thought, but I wouldn't interpret the footprint score this way. As you correctly said, the footprint score is based on a specific pattern of fewer Tn5-cutsites flanked by more Tn5-cutsites. Areas that show this pattern get a high score, while areas with a poor fit will get a low score. So it is only intended for binding prediction. Interestingly, we found that certain TFs seem to produce "better" footprints e.g. CTCF, which may have to do with the duration and strength of the binding. For the activity of a TF you can compare the bound vs. unbound rate or do, e.g., gene set analysis on the genes in proximity to binding sites. However, to investigate the actual effect of a TF, you likely want to incorporate transcription data into your analysis.

[QiliShi]

Thank you for your detailed explanation!
Just to ask one more thing — does it mean that even the mean footprint score generated in the final table does not directly reflect the activity of a transcription factor?

[hschult]

Yes, the mean footprint score looks at all the locations where a TF is predicted to be bound. It gives you an estimate on how "clear" the individual footprints are in your respective condition. As I said, the activity would be the rate of bound vs. unbound sites.

[QiliShi]

Thank you for the explanation