Re: [AMBER] Bug in CPPTRAJ following alignment

From: Carlos Simmerling via AMBER <amber.ambermd.org>
Date: Tue, 27 May 2025 12:38:19 -0400

I don't know the details of your system, but aligning a trajectory carried
out in a periodic system can shift it such that the system no longer fits
in the periodic cell that was originally present. the resulting energies
will indeed be far off, since the system does not fit in a periodic cell
but you are asking Amber to calculate the energies using that cell.
In my experience cpptraj warns about this when it processes that command.
Did you see anything in the output when you did the trajout step?

On Tue, May 27, 2025 at 12:34 PM Helmut Carter via AMBER <amber.ambermd.org>
wrote:

> Hello,
>
> I have found a bug in CPPTRAJ where aligning a trajectory, saving it, then
> loading it causes incorrect energies, possibly due to the placement of
> non-target atoms (that is, atoms in the mask for align). In a minimal
> example using a small molecule solvated in water, where the trajectory is 2
> frames, the following happens:
> # script 1
> parm example.parm7
> trajin example.nc
> align !(:WAT)
> trajout aligned.nc
> energy * etype pme cut 9
> go
> avg(ENE_00001[total]) # returns -17959.450391 as expected
> check
> go # no warnings from check
>
> # script 2
> parm example.parm7
> trajin aligned.nc
> energy * etype pme cut 9
> go
> avg(ENE_00001[total]) # returns 84244.631388
> check
> go # 10 close contact warnings from check
>
> By varying the energy mask, I have found that the incorrect energy stems
> from the solvent molecules. In the analysis of a 5 frame long trajectory,
> the energy after saving and loading was 35591055618150936.000000. This bug
> happens with V6.4.4 from AmberTools and V6.29.13 from GitHub. Please let me
> know if there's any more useful information I can provide.
>
> Thank you,
> Helmut Carter
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue May 27 2025 - 10:00:03 PDT
Custom Search