AF2 Validation

[okanders]

Hi,
I have been really excited to work with RFdiffusion, but am curious how you all went about the validation / filtering step through AF2. I noticed that the DL Binder Design project is mentioned (with initial guess) as an option, but was curious if there was an alternate process (without Pyrosetta / silent files). The general AF2 process is quite long and was hoping you could provide insight on how you dealt with this time and filtering dilemma? Thanks again!

[okanders]

In terms of AF2 recapitulation for de novo binder design, how was the MSA searching space handled (was it single sequence)? I found work by DL Binder Design and ColabFold that seem to reach similar levels of prediction (were these methods used)? Thanks!