gLM2 and protein-protein interaction prediction

[amelie-iska]

Hey, just a thought. You could probably do something similar with the new gLM2 and Sergey's "Categorical Jacobian" method, and this would be very useful for filtering binders designed by RFdiffusion, AlphaProteo, etc.

[flyark]

Thanks for letting me know! They seem interesting!

I am wondering whether they work well for the interaction including flexible regions such as IDR. I can take a look at those when I have time.