Although not a general benchmark, in my small test BioEmu did not detect a mutation-induced secondary structure shift. I saw similar results with other methods like MSA-subsampling. The BioEmu paper focused more on mutation-induced stability/unfolding.

Not at all. Please feel free to share it.

I saw your post and was reminded that I wanted to uploaded my Colab integration of Alphaflow:
colab.research.google.com/github/mar4m...