Fakerotlib for NCAA Parametrization: Issues with Mainchain Scoring durin Structure Prediction

[dnbadu24]

Dear Rosetta Community,

I used the fakerotlib.py script to generate a .params file and rotamer for PHO (which I renamed from Phenylalanine for testing purposes), and the generation was successful. I treated this as new NCAA, update residutypes.txt with path to PHO.params file, keep PHO-rotamer.pdb in same folder, and when I tried to predict a cyclic peptide containing PHO using the following command:

../../../Rosetta-371/main/source/bin/simple_cycpep_predict.static.linuxgccrelease @input.flags -nstruct 1

and two input files for above command:

input.flags
-cyclic_peptide:sequence_file inputs/design_ncaa.txt
-symmetric_gly_tables true
-score:weights ref2015.wts
-in:file:fullatom
-cyclic_peptide:genkic_closure_attempts 1000
-cyclic_peptide:genkic_min_solution_count 1
-cyclic_peptide:min_genkic_hbonds 1
-cyclic_peptide:min_final_hbonds 1
-cyclic_peptide:fast_relax_rounds 20
-cyclic_peptide:rama_cutoff 10.0
-ex1
-ex2
-extrachi_cutoff 5
-mute all
-unmute protocols.cyclic_peptide_predict.SimpleCycpepPredictApplication
-out:file:silent outputs/output.silent
-extra_patch_fa /Computation/Rosetta-371/main/database/chemical/residue_type_sets/fa_standard/residue_types/inhouse_ncaa/PHO.params

design_ncaa.txt
DLEU LEU DPRO DLEU LEU DALA PRO PHO

I encountered the following error during scoring:

ERROR: Error in core::scoring::RamaPrePro::random_mainchain_torsions(): No mainchain score table for residue type PHO exists.

I would appreciate any suggestions on how to address this problem.

Thank you so much in advance!
Dammar

[roccomoretti]

I think it should be -extra_res_fa .../PHO.params rather than -extra_patch_fa

If that doesn't work, you may need to add RAMA_PREPRO_RESNAME PHE to the params file to get it to properly use the PHE version of the RamaPrePro settings.

[dnbadu24]

Thank you so much @roccomoretti for quick response. Adding RAMA_PREPRO_RESNAME PHE worked perfectly. This is very helpful and has definitely moved me one step forward. To take it further, if I’m designing a completely new NCAA, how would I go about using this line or defining similar settings for a custom residue?

[roccomoretti]

This line is to specify which canonical amino acid should be used for the Ramachandran backbone potential, taken from the tables specified by the RAMA_PREPRO_FILENAME file(s). (If you have a NCAA where you have a Ramachandran distribution substantially different from a canonical amino acid, you'd have to create your own version of those files, which is not something many people do. Alternatively, you can just omit the rama scoring for that residue.)

[dnbadu24]

Thanks again @roccomoretti, Could you please clarify that where I need to make changes to omit the rama scoring for a new residue?

[roccomoretti]

I think just omitting the RAMA_* lines from the params file should do it.

[dnbadu24]

Adding or removing the RAMA_PREPO_FILENAME line, for the NCAA I am currently working on, cycloleucine (abbreviated as AC5), lead to crash during prediction. I have attached sdf file used as input for fake_rotlib.py and generated .param file. Let me know if I am missing something. Thanks.
AC5_sdf.txt
AC5_params.txt