RFdiffusion icon indicating copy to clipboard operation
RFdiffusion copied to clipboard

Design small molecule binding proteins [potentials.guiding_potentials]

Open Lyueyang2020 opened this issue 1 year ago • 2 comments

Your work is excellent, and I want to design small molecule binding proteins with your moldes. But I'm afraid I can't understand the meaning of these parameters, after consulting Extended Data Fig. 6 and Methods 4.4. Can you explain this part in detail? Thank you very much for your reply.

# We specify the identity of the substrate molecule (from which to apply the potential)
potentials.guide_scale=1 'potentials.guiding_potentials=["type:substrate_contacts,s:1,r_0:8,rep_r_0:5.0,rep_s:2,rep_r_min:1"]' potentials.substrate=LLK

Lyueyang2020 avatar May 14 '24 07:05 Lyueyang2020

Run in the same problem. Trying to understand whether the substrate identity is taken from the pdb input, or the potential should be specified somehow differently? Looked into the documentation, but this "potentials.substrate=LLK" is mentioned only in the design_enzyme.sh. @Lyueyang2020, any advance so far?

FluffiestFluff avatar Jun 08 '24 19:06 FluffiestFluff

The somewhat described part regarding potentials can be found in README.md, line 398 ### Using Auxiliary Potentials .

Also, an extensive guide is available at https://huggingface.co/blog/AmelieSchreiber/rfdiffusion-potentials

Regarding the ligand:

Run in the same problem. Trying to understand whether the substrate identity is taken from the pdb input, or the potential should be specified somehow differently? Looked into the documentation, but this "potentials.substrate=LLK" is mentioned only in the design_enzyme.sh. @Lyueyang2020, any advance so far?"

The ligand is indeed taken from the input PDB. In the model example, input structure 5an7 contains a ligand "LLK".

FluffiestFluff avatar Jun 11 '24 19:06 FluffiestFluff