Project

General

Profile

Bug #1892

Alchemical change of charges biases Coulomb-14 evaluation in log-file

Added by Andreas Mecklenfeld almost 4 years ago. Updated almost 4 years ago.

Status:
Closed
Priority:
Normal
Assignee:
Category:
preprocessing (pdb2gmx,grompp)
Target version:
Affected version - extra info:
Affected version:
Difficulty:
uncategorized
Close

Description

I've performed an alchemical change of state of Aspirin in TIP3P within Gromacs 5.1 where I did change the partial charges of the atomtypes.
Although the two sets of partial charges differ significantly, the average Coulomb-14 interactions as provided in the log-files of the alchemical end states are nearly identical.

I repeated the alchemical change of state with fixed partial charges of typeA and typeB respectively. Those Coulomb-14 interactions differ significantly due to the influence of the charges.

The interactions of the simulation with a change of charges tend to represent the behaviour of the charge set of topology A (lambda = 0) in both end states. This seems odd, since in the alchemical end state B (lambda = 1) type B charges should be used.

PROD_logA_lambda1.log (58.3 KB) PROD_logA_lambda1.log type A charges (dissolved) Andreas Mecklenfeld, 01/26/2016 08:51 AM
PROD_topA_lambda0.log (58.4 KB) PROD_topA_lambda0.log type A charges (vacuum phase) Andreas Mecklenfeld, 01/26/2016 08:51 AM
PROD_topB_lambda0.log (58.4 KB) PROD_topB_lambda0.log type B charges (vacuum phase) Andreas Mecklenfeld, 01/26/2016 08:51 AM
PROD_topB_lambda1.log (58.3 KB) PROD_topB_lambda1.log type B charges (dissolved) Andreas Mecklenfeld, 01/26/2016 08:51 AM
PROD_lambda1.cpt (72.8 KB) PROD_lambda1.cpt change of charges with type B configuration (dissolved) Andreas Mecklenfeld, 01/26/2016 08:51 AM
PROD_lambda0.cpt (72.8 KB) PROD_lambda0.cpt change of charges with type A configuration (vacuum phase) Andreas Mecklenfeld, 01/26/2016 08:51 AM
PROD.mdp (6.25 KB) PROD.mdp Andreas Mecklenfeld, 01/26/2016 08:51 AM
PROD_lambda1.tpr (86.9 KB) PROD_lambda1.tpr change of charges with type B configuration (dissolved) Andreas Mecklenfeld, 01/26/2016 08:51 AM
PROD_lambda0.tpr (86.9 KB) PROD_lambda0.tpr change of charges with type A configuration (vacuum phase) Andreas Mecklenfeld, 01/26/2016 08:51 AM
Aspirin.itp (14.6 KB) Aspirin.itp Andreas Mecklenfeld, 01/26/2016 09:07 AM
PROD_lambda1.log (58.4 KB) PROD_lambda1.log change of charges with type B configuration (dissolved) Andreas Mecklenfeld, 01/26/2016 09:10 AM
PROD_lambda0.log (58.4 KB) PROD_lambda0.log change of charges with type A configuration (vacuum phase) Andreas Mecklenfeld, 01/26/2016 09:10 AM

Associated revisions

Revision 42cc1cd8 (diff)
Added by Berk Hess almost 4 years ago

Add check for double perturbation

Added a check for changes atom charges and/or types through both
the topology and the couple-moltype option. The couple-moltype
option would override the topology settins for decoupled states.

Fixes #1892.

Change-Id: Ia3d4c0b175af0ad12b189c1b728aee25603afd4f

History

#2 Updated by Berk Hess almost 4 years ago

  • Category set to preprocessing (pdb2gmx,grompp)
  • Status changed from New to Accepted

You are changes the charges both manually in the itp file and "automatically" by setting couple-moltype = ASP in the mdp file. I guess that removing the couple-moltype option will do what you want.
We should add a check that the top file doesn't perturb any atoms when using couple-moltype.

#3 Updated by Gerrit Code Review Bot almost 4 years ago

Gerrit received a related patchset '1' for Issue #1892.
Uploader: Berk Hess ()
Change-Id: Ia3d4c0b175af0ad12b189c1b728aee25603afd4f
Gerrit URL: https://gerrit.gromacs.org/5581

#4 Updated by Berk Hess almost 4 years ago

  • Status changed from Accepted to Fix uploaded
  • Assignee changed from Mark Abraham to Berk Hess
  • Target version set to 5.1.3

I uploaded a fix to gerrit that adds the check, but you don't need that to fix your issue.

#5 Updated by Andreas Mecklenfeld almost 4 years ago

Dear Berk,

thanks for your answer. When I remove the couple-moltype option, the free energy difference of the ligand in bulk with different charge sets will be calculated, correct?

My original idea was to treat the alchemical end states differently in order to describe the ESP of the ligand in bulk and vacuum with corresponding charges. Is this not possible?

#6 Updated by Berk Hess almost 4 years ago

These are user questions that should be posted to gmx-users, please post there if you want to continue this discussion.

Yes, you will calculate the free-energy difference of the different charge sets. But note that you are also changing intra-molecular interactions (1-4 and longer distance), so it's not obvious what the meaning of this free-energy difference is.

I don't see the point of using different charge sets for bulk and vacuum, since the force-field does not give proper energy differences for changing charges (this is related to my first comment). If you want to correct for changing charges, you should use solvation optimized charges and calculate the polarization correction for the change of polarization between vacuum and solvated using an electronic structure method. You should search the literature for this if you're interested in doing that.

#7 Updated by Berk Hess almost 4 years ago

  • Status changed from Fix uploaded to Resolved

#8 Updated by Mark Abraham almost 4 years ago

  • Status changed from Resolved to Closed

#9 Updated by Mark Abraham almost 4 years ago

  • Target version changed from 5.1.3 to 5.1.2

Also available in: Atom PDF