Computed TF activity is high but TF expression is low in scRNA-seq data

[kasayadior]

Dear support team,

I have a question related to the interpretation of TF activity using the scRNA-seq data.
Here is my initial result from the most variable TF activity in the data set:

It looks promising so I further plot some of the TFs in the RNA assay to see if the TF is being expressed:

Not every TF has a higher expression when the computed activity is high. On the contrary, some of the TF being expressed high across cell types is computed with low activity.

Both of the expression are average scaled expression. Thus they are relative within the cells being plotted.

Can someone provide an interpretation of this observation? For further plan, should I restrict the experiment on the TFs that expressed high and computed with high activity? Or I should keep the selection open as TF transcript level often times is not a reflection of protein level and activity.

Is it possible to implement a function to link TF expression when TF activity is predicted active/score high in the current package?

Thank you for your valuable thoughts.