Looking for methods on to create n-terminals that package more kb in a viral capsid using RFDiffusion.

[Hanoriega]

Thank you for this software. I have been using motif scaffolding and binder design. However, my main focus is to create n-terminals that would potentially package more kb preferably >5kb, to a viral protein that normally would only package 4.7. Is there a way?, and I want to disclose that I am not an avid coder, but I am learning. I want to use RFDiffusion to create n-terminals that package more kb in a viral capsid. How would I go about creating something like this using RFDiffusion? Do I have to re-create and train model myself? Or is there an option for this specific direction with this code? Please let me know your thoughts and advice, I appreciate it.

Heather

[roccomoretti]

The purpose of RFDiffusion is to generate novel backbones. So the key in applying RFDiffusion to your biological problem is translating your specification ("package more kb") into a set of requirements for what that implies about the backbone of your protein structure. As it stands currently, that connection (how does a novel backbone help you to package more kb?) is too diffuse to be useful. But by thinking about your particular biological system, you may be able to come up with an approach. (e.g. "If I were able to take this domain here and make it smaller while preserving these contacts I could free up some space" or "If I could extend this domain here, I could expand the size of the diameter of the capsid.")

Depending on what those requirements are, you may be able to make an set of parameters which allows RFDiffusion to generate novel backbones which meet those requirements. As long as you can rephrase your problem in terms of novel backbone generation, you likely won't need to recreate or retrain the model. But if you can't rephrase it as a novel backbone generation problem, then RFDiffusion isn't the tool you need to use.