Rosetta Atom Type Errors ('CA' and 'O') in De Novo Protein Design

[fatoufofana]

Hello Rosetta Community,

I have successfully installed Rosetta and have been working on a de novo protein design project targeting specific interfaces on a malaria parasite protein. Through molecular dynamics simulations, I identified five potential binding pockets that I would like to use for designing binders. However, when running my script, I encountered an error related to unrecognized atom type names, specifically 'CA' and 'O'. I tried adjusting the atom names based on Rosetta’s expected conventions, but the error persists. As I am still learning Rosetta, I started with nstruct set to 10 (planning to increase to 50–100 once everything works). Could anyone provide guidance on:

1- Common fixes for atom type naming issues in Rosetta?
2- Best practices for setting up de novo design targeting specific protein interfaces?

Attached is the Rosetta crash log file for details about the errors. Any suggestions would be greatly appreciated!

ROSETTA_CRASH.log

Thanks in advance for your help.

[roccomoretti]

The -extra_res_fa option takes a Rosetta ResidueType parameter file (params file), not a PDB file. Params files can be generated from SDF/MOL2 files of the ligand through the source/scripts/python/public/molfile_to_params.py script. (Or molfile_to_params_polymer.py file, if it's a non-canonical amino acid.)

But that's only if Rosetta doesn't already recognize all the chemical entities in your structures. Rosetta already has params files in the database for all the canonical amino acids and a number of other common residue types. If you're just doing regular protein design against a protein pocket, Rosetta should already have all the parameters it needs for the common amino acids, and does not require anything passed to -extra-_res_fa. (Unlike MD, you don't have to explicitly parameterize/make a topology file for each input -- Rosetta will do that automatically on read-in of the PDB file.)

[fatoufofana]

@roccomoretti thank you so much for the clarification this makes a lot of sense now. Since I am only working with standard amino acids for protein design, I will go ahead without the -extra_res_fa flag. Really appreciate your help!