Re: Binding free energy tutorial

From: jing liang (jingliang2015_at_gmail.com)
Date: Wed Oct 31 2018 - 04:31:13 CDT

Hi everyone,

I ran the binding free energy tutorial case (alchemical route) with the
suggested modifications. For the unbound simulation
I got the BAR estimator of -54.12 and for the bound case this estimator was
-36.41. That BAR estimator was obtained through
the parseFEP plugin from VMD. By looking at the summary.png plot from
parseFEP, I noticed that the forward and backward
pathways (in both bound and unbound cases) show a larger hysteresis
(~10kcal/mol difference) than the one you presented
in the Fig. 5 of the tutorial. I got some questions, any comment is
appreciated:

* Are all energy restraints included in the BAR estimator? I mean, for
instance the free energy contribution due to the restraining
potential as in Fig. 7 of the tutorial. Or should I get the energy
contribution by integrating some extra terms?

* In the tutorial it says that one can do "grep dA/DLambda rest-01.log" to
get the restraints contributions, but I couldn't find that
output file.

* Should the binding free energy difference be equal to the (BAR estimator
unbound) - (BAR estimator bound)?

Thanks.

El dom., 28 oct. 2018 a las 17:47, David Hardy (<dhardy_at_ks.uiuc.edu>)
escribió:

> Hi Giacomo,
>
> Thanks for clarifying. Sorry for my misunderstanding about how Colvars
> was now handling this configuration.
>
> I will get in touch off-list with the tutorial authors.
>
> Best regards,
> Dave
>
>
> On Oct 28, 2018, at 11:20 PM, Giacomo Fiorin <giacomo.fiorin_at_gmail.com>
> wrote:
>
> Hi everyone, I can confirm that the message:
> colvars: Error: lower wall and upper wall are equal in the domain of the
> variable "Psi".
> is obtained correctly, because -180° and 180° are the same angle. In much
> older versions, the Colvars walls code was simpler and it was possible to
> resolve the error internally by disabling the walls.
>
> The improved version does not allow this as easily, and most importantly
> we agreed during earlier private conversations that an error message is
> more appropriate than a warning about incorrect input. The tutorial's
> input is wrong most likely due to inadvertently pasting the input for a Psi
> interval shorter than -180°:180°: this is a type of mistake that we all do
> often, but it is easy to fix by disabling the walls.
>
> Jing, please disable the wall potentials for that example, and that will
> resolve your error. For more information about the very few example of
> syntax changes, this is the relevant section:
>
> https://colvars.github.io/colvars-refman-namd/colvars-refman-namd.html#sec:colvars_config_changes
>
> Giacomo
>
> On Sat, Oct 27, 2018 at 4:35 PM David Hardy <dhardy_at_ks.uiuc.edu> wrote:
>
>> Let's please keep our replies on the list.
>>
>> I thought that Giacomo and Jerome had updated Colvars to still maintain
>> backwards compatibility with the tutorial's older syntax for defining a
>> harmonic collective variable, but maybe I'm mistaken. (Hey, if either of
>> you see this, maybe respond?)
>>
>> At any rate, the issue is then not a bug in the code (it seems to be
>> giving a correct error message) but instead a bug in the tutorial. For
>> now, you can solve this one yourself by taking a look at the Colvars
>> chapter in the NAMD user guide to find out how to correctly define a
>> "harmonicwalls" collective variable and then fixing the config file.
>>
>> You'd be doing everyone involved a favor if you reply to namd-l with the
>> fixed syntax.
>>
>> Thanks.
>>
>> --
>> David J. Hardy, Ph.D.
>> Beckman Institute
>> University of Illinois at Urbana-Champaign
>> 405 N. Mathews Ave., Urbana, IL 61801
>> dhardy_at_ks.uiuc.edu, http://www.ks.uiuc.edu/~dhardy/
>>
>> On Oct 28, 2018, at 3:35 AM, jing liang <jingliang2015_at_gmail.com> wrote:
>>
>> Hi,
>>
>> thanks for your comments. I downloaded the 2.13b2 KNL version from
>> ks.uiuc.edu website. It doesn't show the
>> message I wrote before. Instead it says that the lower and upper wall
>> values are the same for the angles:
>>
>> colvars: Initializing a new "harmonicwalls" instance.
>> colvars: # name = "Psiw"
>> colvars: # colvars = { Psi }
>> colvars: # outputEnergy = off [default]
>> colvars: # timeStepFactor = 1
>> colvars: # writeTISamples = off [default]
>> colvars: # writeTIPMF = off [default]
>> colvars: # forceConstant = 1 [default]
>> colvars: # targetForceConstant = -1 [default]
>> colvars: # lowerWalls = { -180 }
>> colvars: # upperWalls = { 180 }
>> colvars: # lowerWallConstant = 0.5
>> colvars: # upperWallConstant = 0.5
>> colvars: Error: lower wall and upper wall are equal in the domain of the
>> variable "Psi".
>> FATAL ERROR: Error in the collective variables module: exiting.
>>
>> I don't get that error message if I use instead of -180 the value of
>> -179.999 but I guess that could be a bug from
>> the code which might have other consequences?
>>
>> Thank you.
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> El sáb., 27 oct. 2018 a las 10:15, David Hardy (<dhardy_at_ks.uiuc.edu>)
>> escribió:
>>
>>> Dear Jing Liang,
>>>
>>> Please try running that part of the tutorial again with the most recent
>>> NAMD 2.13b2. The "misunderstanding" between the tutorial configuration
>>> file and the Colvars module has recently been fixed.
>>>
>>> Best regards,
>>> Dave
>>>
>>> --
>>> David J. Hardy, Ph.D.
>>> Beckman Institute
>>> University of Illinois at Urbana-Champaign
>>> 405 N. Mathews Ave., Urbana, IL 61801
>>> dhardy_at_ks.uiuc.edu, http://www.ks.uiuc.edu/~dhardy/
>>>
>>>
>>> On Oct 27, 2018, at 3:37 AM, jing liang <jingliang2015_at_gmail.com> wrote:
>>>
>>> Hi,
>>>
>>> I downloaded the tutorial "Protein:ligand Standard Binding Free
>>> Energies: A Tutorial for Alchemical and Geometrical Transformations" in the
>>> site: https://www.ks.uiuc.edu/Training/Tutorials/
>>> The Alchemical Bound state simulation only ran fine for the backwards
>>> direction. In the forwards
>>> direction it showed the following error: jacobi too many iterations. I
>>> am working with the
>>> version 2.12 shared memory version. Have you experienced this issue in
>>> the present test case?
>>>
>>> Thanks.
>>>
>>>
>>>
>>>
>>
>
> --
> Giacomo Fiorin
> Associate Professor of Research, Temple University, Philadelphia, PA
> Contractor, National Institutes of Health, Bethesda, MD
> http://goo.gl/Q3TBQU
> https://github.com/giacomofiorin
>
>
>

This archive was generated by hypermail 2.1.6 : Sat Sep 21 2019 - 23:19:59 CDT