Flanks of motif CWMs from chrombpnet model are rich in GCs?

[09oulin]

After getting my prediction results from cell-type specific chrombpnet model, I further processed using TF-Modisco and FiNeMo. When I checked the hit calling report, I found the two flanks of the CWMs are rich in GC bases, does it mean my chrombpnet model not right? I am really confused about it.

[adamyhe]

Mammalian CRE sequences are often GC-rich, so random degenerate GC patterns in attribution scores are pretty common.

[akundaje]

Are these from the count head or profile head?

[09oulin]

Are these from the count head or profile head?

From the count head.

[09oulin]

Mammalian CRE sequences are often GC-rich, so random degenerate GC patterns in attribution scores are pretty common.

Thanks, I see.

[09oulin]

By the way, I couldn't find any negative patterns in the TF-Modisco results for all my cell-type specific models (about 200 cell-types from embryo tissues). I think it's strange and unreasonable. Could you please give some advice?