Hi Paulius,
Thanks for your email.
> On Nov 12, 2019, at 5:58 AM, Paulius Kantakevicius <paulius.kantakevicius.gmail.com> wrote:
>
> Hello,
>
> I experienced problems to reproduce the HFE values of 2+ and 3+ ions cited
> in papers (https://pubs.acs.org/doi/10.1021/ct400751u
> https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4306492/) using 12-6-4LJ
> potentials on AMBER18 and AMBER16. I am using Sander engine and I added the
> C4 terms using Parmed from AmberTools19, however the desolvation energy
> values I obtained are off: Ca2+ I obtained desolvation energy of 348.9
> kcal/mol (350.1 from charge removal and -2.2 from VdW removal) -
> experimental is 359.7 kcal/mol, whereas for Mg2+ I got 410.7 kcal/mol
> (412.8 from charge removal and -2.1 from VdW removal) - experimental is
> 437.4 kcal/mol. I tried running Y3+ ion as well, for which I obtained 788.2
> kcal/mol whereas experimental value is 824.6 kcal/mol.
>
Can you send these input prmtop and inpcrd files (after adding C4 terms) to my email: ambermailpengfei.gmail.com <mailto:ambermailpengfei.gmail.com>? I can help to do a check.
> I tried running Ca2+ ion without adding C4 term and I got HFE values
> further from experimental ~337 kcal/mol indicating that Sander uses C4 term
> in TI calculations.
> Then I tried running Ca2+ and Mg2+ ions using different set of parameters
> from older paper (https://pubs.acs.org/doi/10.1021/ct400146w) that were
> made for 12-6LJ (frcmod.ions234lm_hfe_tip3p) and have to reproduce HFE
> values accurately but coordination numbers and IODs with errors to see
> whether they will produce accurate HFE values. And they indeed produced
> values, which were very close to experimental values: 360.4 kcal/mol for
> Ca2+ and 437.5 kcal/mol. Therefore, I suspect that there is an issue in
> AMBER16 and AMBER18 in implementation of C4 term in TI calculations for
> Sander engine.
>
> Also, I noticed that some 12-6LJ parameter files in AmberTools19 contain
> wrong parameters:
> in frcmod.ions234lm_hfe_tip3p file for 3+ and 4+ ions instead of HFE values
> from the https://pubs.acs.org/doi/10.1021/jp505875v paper IOD values are
> entered instead and in frcmod.ions234lm_iod_tip3p file for 3+ and 4+ ions
> instead of IOD values 12-6-4LJ values are shown instead.
>
I checked the parameter files, you are right.
I apologize for these stupid mistakes, and thank you very much for finding them! I checked the git log and found these mistakes were made when I merged the parameter files in 2016: the parameters in AMBER 2015 were correct for the +3 and +4 ions, but afterwards the following parameter files were messed up because of my stupid mistakes:
(1) the file frcmod.ions234lm_hfe_tip3p should have HFE parameter set, but has IOD parameter set for the +3 and +4 ions in the TIP3P water model
(2) the file frcmod.ions234lm_iod_tip3p should have IOD parameter set, but has 12-6-4 parameter set for the +3 and +4 ions in the TIP3P water model, note that only the Rmin/2 and epsilon parameters in the file, while the C4 terms are still need to be added by the ParmEd program.
(3) the file frcmod.ions234lm_126_tip3p should have IOD parameter set, but has 12-6-4 parameter set for the +3 and +4 ions in the TIP3P water model, note that only the Rmin/2 and epsilon parameters in the file, while the C4 terms are still need to be added by the ParmEd program.
(4) the file frcmod.ions234lm_hfe_tip4pew should have HFE parameter set, but has IOD parameter for the +3 and +4 ions in TIP4PEW water model
It seems the other parameter files about +3 and +4 ions are fine. The fortunate part out of all the misfortunes is that the VDW parameters in the IOD and 12-6-4 parameter sets are not too far away from each other (Rmin/2 usually within 0.1 Angstrom with each other), and the HFE parameters for +3 and +4 are usually not used in MD simulations because their big errors for simulating the structural properties like IOD.
I am very sorry for any inconveniences these mistakes brought to the community… I will create a git commit to correct these mistakes later this week.
Yours sincerely,
Pengfei
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Nov 15 2019 - 16:00:03 PST
