Re: [AMBER] and rms fit: Results depend on reference frame

From: Lennart Gundelach <>
Date: Thu, 27 May 2021 09:07:15 +0000


I had sent over a zip with files some time ago but I now noticed that I can only send tiny file sizes, and that the message was likely never delivered. I have instead hosted the files on guthub at

It contains two identical trajectories of 2 frames each, one autoimaged and rms fit, the other without. Produce different energies.

Best wishes,


From: David A Case <>
Date: Wednesday, 12 May 2021 at 15:01
To: AMBER Mailing List <>
Subject: Re: [AMBER] and rms fit: Results depend on reference frame
CAUTION: This e-mail originated outside the University of Southampton.

On Wed, May 12, 2021, Lennart Gundelach wrote:
>This was because the script does
>an autoimage and rms fit on the whole trajectory before calculation
>energies or entropies (which it obviously cant in the snapshot by snapshot
>I have naively been assuming, that the energies obtained would be invariant
>under rotations and translations of the system and that the final energies
>would not depend on the choice or reference frame or ordering of the
>trajectory. This is not the case.

>My fix is simply to autoimage and rms fit the trajectory before slicing it
>into individual snapshots,

Can you find a snapshot that gives a different energy before the
autoimage/rms and afterwards? I think that will be needed to figure out
what is going on. [Since MMPBSA is interpreting part of a periodic snapshot
in a non-periodic fashion, it's certainly possible that autoimage might not
be doing the desired thing here: autoimage is coded for periodic systems.]

As an aside, I don't (immediately) see how you are going to average energies
that arise from snapshots with different numbers of waters. But I haven't
given that a lot of thought....


AMBER mailing list;;sdata=%2Fo%2F%2FNSm9Oh3eEUOeB3OwssrRzhjgi99aMqtSJ2AHFTw%3D&amp;reserved=0
AMBER mailing list
Received on Thu May 27 2021 - 02:30:02 PDT
Custom Search