Re: [Apbs-users] Reproducing solvation free energies (PARSE) with APBS
Biomolecular electrostatics software
Brought to you by:
sobolevnrm
From: Jessica MJ S. <jsw...@mc...> - 2004-10-25 23:15:39
|
Hi Marc, In response to your last e-mail: > - the calculation procedure adopted in the PARSE paper is > "non-standard" as from today's practices in PB calculation > - the calculation results are strongly depending on the geometry > (that is the forcefield used to obtain those) even for very > simple structures like methanol Yes, these two conclusions seem correct to me. > > This now leads to some serious general questions > > - what is the sense of using the PARSE parameter set if we don't > use the same protocol as was used for its parametrization ? I think it IS important to use the same protocol as they used in the paper. Most important is the solute dielectric and the solvated surface definition: pdie 2 srad 1.4 # in both solvated and reference OR srad 1.4 # in solvated srad 0.0 # in ref Choosing srad 1.4 vs. 0.0 doesn't seem to make TOO much of a difference in small molecule test cases. > > - does one have to minimize each structure withi the same forcefield > (Discover/Insight was used by the PARSE authors) to have geometries > "compatible" with these parameters ? If you want to reproduce their results exactly, sure. If you want a set of parameters that reproduce experimental solvation energies for small molecules, no; the PARSE params should work fairly well for this. > > On the other hand, I am not quite settled on the status of possible > alternatives. Are there other general parameter sets meant to reproduce > experimental solvation free energies ? > > The particular point I am worried about is that I want to combine > the calculated PB energies with molecular mechanics, surface area > and eventually entropic contributions in an MMPBSA-like approach. > There are PB parametrizations for Amber and Charmm to be used within > such a framework, but as my molecular mechanics part was done with > Gromacs (Gromos forcefield) I wonder what is the best attitude > (parameter set) to adopt ? > Any hints or suggestions are very welcome. > Since you are interested in combining MM and PB energies in an MM/PBSA-like approach, the PARSE params are not the best choice. Their compatability with the MM forcefield is suspect. The best choice for this are the Nina et al. radii combined with the CHARMM ff or a set of radii that I have optimized for the AMBER ff (should be published within a month or so). Since you are working with the Gromos ff, you may be in a bit of a pickle.;) The best option for you may depend on what you are trying to get out of your MM/PBSA analysis. If you are looking at relative affinities between several small molecule ligands, I would stick with the PARSE params as they are better suited to pick up these differences. If you are interested in the absolute affinities OR are looking at protein like systems, I would choose either the CHARMM or AMBER options. Are the Gromos partial charge assignments similar to AMBER or CHARMM? Again, I hope this is helpful. Best, Jessica > > jsw...@mc... said: > >> Hi Marc, > >> Sorry to take a while to get back to you. I was away on travel. I > >> tried your structures and came to very similar conclusions as you > >> did. I think that the main cause of disparities is different > >> structures. I was getting differences around 0.2 - 1.0 kcal/mol for > >> altered conformations. > > >> I agree with your interpretation of the PARSE paper. It seems that > >> they used a solvent probe of 1.4 for their solvated system and 0.0 > >> for their reference calculations. As you said, the results get worse > >> (i.e., further from the PARSE results) if one keeps the solvent probe > >> at 1.4 in both calculations. This is really an odd way to do a > >> continuum solvation calculation and I don't think that many have done > >> it this way since then. As Nathan pointed out, it is standard > >> practice to use the same surface definition in the solvated and > >> reference systems. > > >> Hope this helps, Jessica > > Marc Baaden > > > -- Jessica Swanson McCammon Group UCSD-Chem/Biochem 858-822-2771 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ "A man's ethical behavior should be based effectually on sympathy, education, and social ties..." --Einstein-- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ My grandmother started walking five miles a day when she was sixty. She's ninety five now, and we don't know where the hell she is. ---Ellen Degeneres--- |