I want to predict the protonation states of Histidine in my protein and I don't have any ligand. As long as I try Propka 3 (only protein) and Propka 3.1 (protein and ligand), I receive different results and I'm wondering which one is trustworthy. Can anybody help? Thanks.
I would suggest that you start with a true comparison. You've got two variables - software version and presence of a ligand. If the two versions produce the same protein pKa values in the absence of the ligand, then you've resolved the version issue. It is also perfectly reasonable that the binding of a ligand affects the pKa values of the protein residues; the local environment has changed.
Hi Justin, thanks for your answer, but the point is that I tried both
version for my protein in the absence of ligand and I get
significant difference for the results obtained from two versions.
This is strange as they wrote in the site that PROPKA3.1
implements the same force field as PROPKA3.0 but also includes
ligands in the pKa prediction". So I would expect to receive the
same results when I don't have any ligand.
OK, that makes more sense. I would suggest contacting the PROPKA developers to see what the root cause might be. It would seem something has changed in their algorithm.
Hi, the difference between PROPKA 3.0 and 3.1 is that histidine residues are treated as ligands in version 3.1, but residues in 3.0. The consequence of doing this is that the center-of-interest (the charge center) is the ring in version 3.0, but one of the nitrogen atoms in 3.1. We tested the two approaches and it didn't give any significant change for the set we looked at (very minor difference). Some descriptors can of course change because of this ~1 A change, e.g. how buried the residue is, desolvation, electrostatic interaction with neighbours etc. Hydrogen bonding is always counted from the nitrogen atoms in both versions.
So which is the one to trust? As always one has to judge case-by-case if there are significant discrepancies, but a priori I tend to be a bit biased and favour version 3.0 because of the delocalised charge. One should, however, also ask oneself if the discrepancy is really significant. Typically both experiments and predictions are more difficult for histidines than say aspartic acids. Participating in the 'pka cooperative' blind prediction a few years ago was a bit of an eye-opener for me and others participating (see for instance Olsson PROTEINS 12/2011; 79(12):3333-45). The residues predicted in this cooperative are however among the most difficult to predict to date.
Dear Mats,
Do you know when the Propka website will become available for interested users? Are there any alternatives?
Thanks,
Dmitri
You can use the ProteinPrepare web application (www.playmolecule.org) for protonation prediction and structure optimization. We use PROPKA3.1 as well.
Dear Gerard,
I am working with a dimer of a GPCR and wanted to optimize the pKa @ pH 7.2 . I used ProteinPrepare and must say it is a very interactive app and great work there.
However, I cannot locate a dependable README or usage document, esp with respect to forcefields (rtp) used. My protein ended up having different names for some residues like: LYS -> LYN, CYS -> CYX, GLU -> GLH. I am not sure which forcefield nomenclature does it use?
Also, my structural model have an evolutionary conserved disulfide bridge and to my surprise, the corrected structures removed one of the participating Cysteine.
I request you to please enlighten me on these issues by providing a user manual or reference document as I cannot access the JCIM publication through my university library.
Thanks.
- Badges
- Science topic