Description
Hi Protenix Team,
Thank you for open-sourcing Protenix! It's a fantastic resource for the community, and we're excited about its potential.
We're currently working with a set of novel synthetic ligands that we've designed and docked/co-crystallized with known PDB protein structures. Our main goal is to fine-tune the Protenix model using these new protein-ligand complexes (approximately 20,000, none of which are in the official PDB CCD) to improve predictions within our specific chemical space. We've begun thinking about the best way to tackle this fine-tuning process and would greatly appreciate your guidance.
For our ligands that are not in the standard PDB CCD, we're looking to confirm the best way to prepare them for Protenix. Specifically, is it necessary to create custom parameter files or entries (perhaps similar to CCD entries) to define their properties like ones in components.cif file?
Any pointers, examples, or best practices you could share for this use case, particularly confirming if our intended approach for handling non-CCD ligands is sound, would be helpful.
Thanks again for your great work!