I want to test a docked structure to a solved PDB structure in order to test HADDOCK’s accuracy on my system.
Currently, I align the protein structures in pymol, and then using rms_cur I calculate the RMSD between the peptides. This value comes out high so I’d like to try another approach.
The information I am looking for is how well haddock finds the interactions and contacts between the protein and peptide so FCC can be a more suitable choice (the peptide is highly flexible).
Is there a way to check the difference between FCC of both complexes? Or is there another suitable approach?
I use Haddock local version 2.4, but I can use other platforms as well
I would recommend trying HADDOCK3 in this context, using this workflow. You just have to change the molecules field, which will became your HADDOCK models and the reference_fname field, which will be your PDB structure. Check the fnat field in the run_directory/2_caprieval/capri_ss.tsv file: this will be the fraction of native contacts that are preserved between each HADDOCK model and the reference structure.
Otherwise there are multiple resources you can also try, such as CAPRI-Q.
PS: rms_cur does not perform the structural alignment between structures, align would be a better choice in pymol
Hope it helps!
Marco
I understand that rms_cur doesn’t perform structural alignment. However, if it did, wouldn’t that compromise the fairness of comparing the accuracy of the results? Or am I misunderstanding its role here?
Also, will the HADDOCK3 workflow you suggested produce the same results as CAPRI-Q, considering it’s based on it?
when you have two models you first have to align them…if for some reason they are randomly positioned in space you will get a high RMSD value because you didn’t remove the translations and rotations…no fairness problem here!
as for the comparison, we never benchmarked one method against the other, but the results should be very similar.
