[gmx-users] CHARMM ff cutoffs

[Justin Lemkul]

On 10/19/17 10:46 AM, João Henriques wrote:
> Dear Micholas,
>
> First of all, thank you for your input. I understand the whole cutoff
> problematic and it is my desire to stick to the standard. However (and
> maybe I didn't explain this part properly), the CHARMM36m publication
> reports different cutoffs (if you check the SI) depending on which protein
> is simulated. This publication *IS* the standard, right? I am well aware
> that other CHARMM implementations use 1.2 nm, but if you go further back in
> time, the cutoffs have taken other values. But I digress, the main question
> here is, how can I justify my choice when the official publication reports
> these two values and does not provide an explanation for it? I am well
> aware that the 1.2 nm value is the *de facto* value, but that is not an
> acceptable justification in the eyes of a journal referee, as we all can
> understand.

Disclaimer: I was not involved in designing the methods for the paper 
you mention, so you may want to email Alex about it for absolute certainty.

But what I can say is that we understand that the protein and nucleic 
acid force fields are more tolerant to changes in cutoff (as compared to 
the lipids, for example, which turn into an absolute disaster when 
changing things), so sometimes it is OK to reduce them especially in the 
case of a huge system or big REMD job to speed things up. I generally DO 
NOT advocate for this, because if you tell people you can toy with 
cutoffs, they start doing it without understanding the underlying 
physics or knowing how to check if things are actually OK.

If you want to be safe, use what I have posted on the GROMACS website. 
That is what is used in 99.9% of all CHARMM protein and nucleic acid 
simulations (at least, the good ones :). Lipid cutoffs can vary and 
there's lots of literature about that (and some open debate - see work 
by Jeff Klauda, Tom Piggot, etc).

As soon as the force fields are reparametrized for use with LJ-PME, this 
whole debate will be rendered moot.  In time...

-Justin

> Thanks!
> J
>
>
>
> On Thu, Oct 19, 2017 at 4:33 PM, Smith, Micholas D. <smithmd at ornl.gov>
> wrote:
>
>> João,
>>
>> If you use the CHARMM-GUI (charmm-gui.org) to build your systems, they
>> set their "standard" cut-offs to be 1.2nm, unless there is a membrane, then
>> they set it to 1.4nm (at least that's what it use to do).
>>
>> Changing cut-offs is kind of a mess. A lot of people will argue that the
>> cut-offs are explicitly part of the force-field, and so whatever cut-offs
>> were used for the parameterization are what you absolutely must use. But
>> then you'll find papers (lake you did) that uses 0.95nm cut-offs and get
>> reasonable results.
>>
>> I would stick to the standard unless you have a really good reason (not
>> performance) for toying with the cut-offs.
>>
>> -Micholas
>>
>> ===================
>> Micholas Dean Smith, PhD.
>> Post-doctoral Research Associate
>> University of Tennessee/Oak Ridge National Laboratory
>> Center for Molecular Biophysics
>>
>> ________________________________________
>> From: gromacs.org_gmx-users-bounces at maillist.sys.kth.se <
>> gromacs.org_gmx-users-bounces at maillist.sys.kth.se> on behalf of João
>> Henriques <joao.m.a.henriques at gmail.com>
>> Sent: Thursday, October 19, 2017 10:03 AM
>> To: gmx-users at gromacs.org
>> Subject: [gmx-users] CHARMM ff cutoffs
>>
>> Dear all,
>>
>> Is there any piece of literature that explicitly states what is the *de
>> facto* cutoff for CHARMM FFs, i.e., for the LJ and electrostatic
>> interactions? The old gromacs site states it's 1.2 nm:
>>
>> http://www.gromacs.org/Documentation/Terminology/Force_Fields/CHARMM
>>
>> However, I've read Robert Best's C36 and Huang's C36m papers, along with
>> older CHARMM ff publications by MacKerell and Co., and I'm rather confused
>> at this point. E.g., on the C36m paper, the RS peptide is simulated with
>> 0.95 nm cutoffs and other proteins with 1.2 nm. Robert is consistent and
>> uses 1.2 nm all the way. Older publications use even shorter cutoffs.
>>
>> I know the cutoffs can probably be toyed with to a certain extent (Stefano
>> Piana and Kresten Lindorff-Larsen showed that on their 2012 paper), but a
>> recent paper by Davide Mercadante on JCTC seems to show that it is not so
>> simple for IDPs, and shortening the cutoffs is a no-no (even though he did
>> it for his KBFF and AMBER FFs, not CHARMM).
>>
>> In my work I use CHARMM36m with 1.2 nm, but, looking at the literature,
>> there's no way I can justify my choice when the original C36m does not
>> stick to a single cutoff selection...
>>
>> I know Justin is/was affiliated with the MacKerell lab, so maybe he can
>> shed some light on this subject. Anyone else is encouraged to give their
>> input as well.
>>
>> Thank you in advance,
>> Best regards,
>> João
>> --
>> Gromacs Users mailing list
>>
>> * Please search the archive at http://www.gromacs.org/
>> Support/Mailing_Lists/GMX-Users_List before posting!
>>
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>
>> * For (un)subscribe requests visit
>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>> send a mail to gmx-users-request at gromacs.org.
>> --
>> Gromacs Users mailing list
>>
>> * Please search the archive at http://www.gromacs.org/
>> Support/Mailing_Lists/GMX-Users_List before posting!
>>
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>
>> * For (un)subscribe requests visit
>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>> send a mail to gmx-users-request at gromacs.org.
>>

-- 
==================================================

Justin A. Lemkul, Ph.D.
Assistant Professor
Virginia Tech Department of Biochemistry

303 Engel Hall
340 West Campus Dr.
Blacksburg, VA 24061

jalemkul at vt.edu | (540) 231-3129
http://www.biochem.vt.edu/people/faculty/JustinLemkul.html

==================================================