Re: [AMBER] issues with MMPBSA and explicit waters

From: Jason Swails <>
Date: Mon, 2 Sep 2013 12:08:57 -0400

On Mon, Sep 2, 2013 at 10:39 AM, Alfredo Quevedo <>wrote:

> Dear Amber users, I am facing a problem with an MMPBSA calculation when
> keeping explicit water molecules. The procedure I am following to
> perform the decomposition analysis is:
> 1: center and image the explicit MD trajectory,
> 2: use cpptraj to identify "closest" water molecules, write stripped
> .prmtop file and resulting trajectory,
> 3: generate "complex", "receptor" and "ligand" .prmtop files using
> (without stripping waters),
> 4: run the module, settinf the "strip_mask" keyword to non
> default parameters so as to keep explicit waters.
> When I performed the above mentioned procedure keeping just 1 explicit
> water (closest 1: in step 2), the calculation finished without any
> problem and calculated components seemed to be ok.
> My problem originates when I want to keep 2 or more explicit water
> molecules in the trajectory, in which when running step 4, the
> calculation for the complex finishes ok, but when the calculation for
> the receptor starts I get the error "bad atom type: HW". This error
> originates from sander when trying to run the decomposition calculatioin
> from the files created by (_MMPBSA_gb_decomp_rec.mdin,
> _MMPBSA_dummyreceptor.inpcrd). I inspected the generated files with VMD
> using receptor.prmtop and everything seems to be ok. I cant understand
> why sander gets an "atom type" error if everything was ok for the
> simmulation corresponding to the complex.
> Has anyone faced this problem before?.

This is caused by the fact that there are no parameters for gbsa=2 code for
HW atom types. Look at the relevant code in mdread.F90 [or mdread.f]
($AMBERHOME/src/sander, around line 2215 or so, I think). You will need to
add code in order to assign a radius to the HW type (you should probably
make it the same as the other hydrogen type(s) that are defined there).

In the future, it would be helpful to tell us what version of Amber and
AmberTools you are using. I don't see how you could possibly be getting
this error with Amber 12, so I assume you're using Amber 11 (if you are
using Amber 10 or earlier, will not work with decomposition).

Good luck,

Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Mon Sep 02 2013 - 09:30:02 PDT
Custom Search