VMD-L Mailing List

From: Peter Freddolino (petefred_at_umich.edu)
Date: Mon Sep 07 2020 - 07:31:01 CDT

In *most* cases you should get compatible results because the physics are
*mostly* the same, but there may be some edge cases where differences would
arise -- for example, I'm pretty sure that a bug in long-range vdW
corrections was fixed between 2.9 and 2.13, which would affect the physics
for large systems if you have vdW tail corrections on (you should be able
to see other changes by browsing the release notes). But even if you can't
identify a particular reason why the results would be different, if it were
me, I'd be going out of my way to use a consistent version for all
calculations going into a single study. This might mean a conversation with
the folks administering your cluster to see if they can make the older
version available again as well, or setting up your own installation of 2.9.
Best,
Peter

On Mon, Sep 7, 2020 at 5:48 AM Krishna Vkm <krishnatifr_at_gmail.com> wrote:

> Hi Peter
> Yes, you pointed out the main issue. I have to write that I have used
> different versions of NAMD in the method section. So, I think the reviewer
> can ask that there can be an issue with using different NAMD versions
> regarding reproducibility.
> The main thing is that I am studying a protein complex system consisting
> of two proteins (let's say A and B). I am looking at the distance between
> the center of mass of A and B. I ran multiple independent simulations and
> I got different distances every time. Now I want to add more number
> independent simulations for better statistics because the distances between
> A and B are not consistent and so require good statistics. Now if different
> versions of NAMD differ in some aspects ( which I am not aware of) which
> can affect the distances between A and B, then I think it will be a problem
> if I mix the simulations from NAMD2.9 and NAMD2.13 to know about the
> distances between A and B.
>
> Thank you
>
> On Mon, Sep 7, 2020 at 5:30 AM Peter Freddolino <petefred_at_umich.edu>
> wrote:
>
>> With that said... do you really want to write in your methods section
>> that you used different versions of namd for your different simulations?
>> Best,
>> Peter
>>
>> On Sun, Sep 6, 2020 at 7:44 PM Josh Vermaas <joshua.vermaas_at_gmail.com>
>> wrote:
>>
>>> I would expect both versions to sample the same ensemble. Is that not
>>> what you are seeing? I'd also recommend 2.14 while you are at it.
>>> Josh
>>>
>>> On Sun, Sep 6, 2020, 4:03 PM Krishna Vkm <krishnatifr_at_gmail.com> wrote:
>>>
>>>> Dear NAMD User
>>>> I did multiple all-atom MD simulations of my protein system two years
>>>> ago with *NAMD2.9*. Now I want to do more simulations of the same
>>>> system but my cluster has *NAMD2.13* instead of *NAMD2.9*. I would
>>>> like to have your suggestions on whether the results from NAMD2.9 and
>>>> NAMD2.13 will be different or the same.
>>>>
>>>>
>>>> *Thanks and regards*
>>>> *Krishna Kant*
>>>> *TIFR Mumbai*
>>>>
>>>>
>>>>