Enabling explicit modeling of crosslinks: announcement and a question
Closed this issue · 2 comments
Announcement: Crosslinks can sometimes improve AlphaFold3 models. Crosslinking is an experimental method often used as a source of distance restraints in modeling. In AlphaFold3, instead of using restraints, crosslinks can be included as covalent ligands. Thanks to Agnieszka Obarska-Kosinska for the idea.
Proof-of-concept implementation: https://github.com/KosinskiLab/alphafold3x
Technical report: https://www.biorxiv.org/content/10.1101/2024.12.03.626671v2
Question How should we call AlphaFold3-derived programs? E.g. I think OpenAI has clear rules that derived work can't use "GPT" in names. If my code above grows to a reusable program, can I keep calling it AlphaFold3x? Happy to rename it if necessary but I just wanted to indicate in the name that this is just an AlphaFold3 mod.
This is neat - amazing work! Glad to see things being built on top of AlphaFold 3 so quickly.
Our derived work naming guideline is avoiding the full "AlphaFold" in the name, but OK to use "AF", "AF3", "Fold", or "Alpha" in the name.
I also see you are using the num_seeds flag -- it looks like a useful feature and I will add that in our repo.
OK, thanks, I will rename the program!