Can RFDiffusion be used to optimize the structure of the homodimer predicted by af2-multimer?

[TernencezzZ]

Hi, thank you for your contribution and great work done on this project.

I have some homodimer structure prediction tasks, and the results of the af2-multimer output are not very good in terms of symmetry. I wonder if the partial structure diffusion in this repo can be used to optimize the structure of its specific region? However, it is troublesome to keep the sequence corresponding to the optimized structure unchanged. I don't know if this can be achieved, or is there any other way?