Scoring Electrostatics

Member Site Forums Rosetta 3 Rosetta 3 – General Scoring Electrostatics

Viewing 3 reply threads
  • Author
    • #1369


        I know that there used to be a scoring potential called hack_elec. Does anyone know if this still exists in Rosetta 3.x ? If it does what are the command line flags to invoke it during scoring?

        Thanks in advance,

      • #7541

          hack_elec is alive and well.

          There isn’t a flag for it. You can activate it (like all other terms) by passing a custom weights file with -score::weights and -score::patch (for protocols that use getScoreFunction(), which is most of them). Take a look in rosetta_database/scoring/weights to see weights/patch files; some use hack_elec already (for example the docking.wts).

          Determining what weight it _should_ have is an empirical and nearly unanswerable question. (There is a paper coming on OptE, which is a systematic method for addressing scorefunction weights). At the least, consider de-activating the pair term (give it a weight of zero) to avoid doublecounting electrostatics.

        • #7561


            @Steven: Thank you for the instructions. I shall turn it on based on your suggestion.

            @Rocco: Thank you for providing an idea of the weight. I found that for docking it is 0.026, and for ligands it is 0.25. I am trying to stabilize a protein-protein interface by using charged residues at the periphery. Based on this do you think I should be turning the fa_pair off (by setting a weight of 0) and turn on hack_elec to 0.5?

            Thanks for your help,

          • #7564

              Thank you Rocco. I will try both.

            • #7559

                In preliminary testing, a value of hack_elec = 0.5 seems to be at about optimal for most things.

                As Steven mentions, if you include both hack_elec and fa_pair you are likely to double count electrostatics. (Although on some problems tests indicate it doesn’t matter.) There is several ways to handle it. One is to deactivate fa_pair entirely when you turn hack_elec on. Another (useful if you’re looking at protein-ligand and protein-DNA binding) is to turn off protein-protein hack_elec interactions. You can do this by adding “NO_PROTEIN_PROTEIN_HACK_ELEC” (all-caps but no quotes) on a line by itself in the weights file (should work in both 3.3 and 3.4).

              • #7563

                  For ligand docking it’s listed as 0.25 in the weights file, but under typical running conditions (i.e. with the flag -ligand:old_estat) there’s internal fiddling in the ligand docking application to bump that up to somewhere around 0.42 ish. (I’m not familiar with the docking weights, though 0.026 seems feebly small).

                  If you’re only going to do one run, I’d probably recommend trying it with fa_pair turned off. But if you have the resources, I’d certainly recommend trying it both ways – once with both fa_pair and hack_elec at 0.5, and once with fa_pair at zero and hack_elec at 0.5.

              Viewing 3 reply threads
              • You must be logged in to reply to this topic.