RE: Free energy did not converge by using ABF

From: Nielsen, Steven (steven.nielsen_at_utdallas.edu)
Date: Fri Dec 08 2017 - 15:01:55 CST

Jeff,

Out of curiosity, how did you combine the ABF windows together, in other words, how did
you determine the offset of each free energy profile so that they line up correctly?

-Thanks, Steve
________________________________________
From: owner-namd-l_at_ks.uiuc.edu [owner-namd-l_at_ks.uiuc.edu] on behalf of Jeffrey Potoff [jpotoff_at_wayne.edu]
Sent: Friday, December 8, 2017 1:22 PM
To: namd-l_at_ks.uiuc.edu
Subject: RE: namd-l: Free energy did not converge by using ABF

We've done these kinds of calculations before.
Direct calculation of 1-octanol–water partition coefficients from adaptive biasing force molecular dynamics simulations The Journal of Chemical Physics 137, 014502 (2012); https://doi.org/10.1063/1.4730040
Navendu Bhatnagar, Ganesh Kamath, Issac Chelst, and Jeffrey J. Potoff

Pretty much any solute going into water works ok. 50ns per window is a lot. For dunking stuff into water 4-8ns seems to work ok. Take a look at your gradients and counts files and see what is going on there. Maybe your interface is shifting during the simulation?

We have found ABF has some advantages for calculating free energies of transfer for certain molecules going into organic solvents, but as a general rule, if all you want is the free energy of solvation FEP/TI is easier and faster once you get a protocol that works for your solute/solvent combination.
Regards,
Jeff

-----Original Message-----
From: owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] On Behalf Of Yaxin An
Sent: Friday, December 08, 2017 9:57 AM
To: Jérôme Hénin <jerome.henin_at_ibpc.fr>
Cc: Namd Mailing List <namd-l_at_ks.uiuc.edu>
Subject: Re: namd-l: Free energy did not converge by using ABF

Hi Jérôme;

Thanks for the kind reply.

I actually saw larger fluctuation of sampling at the interface. Here, I should use one window across the complete interface? That is to say, if the interface was at 20A, I should use 15-25 A in this window, right?

When you say ABF is not the method of choice to compute free energies of hydration, you mean other methods like TI are better choices? In some papers, TI was used to calculate the free energy of hydration.

Thanks very much.

Bests;

Yaxin

On Fri, Dec 8, 2017 at 7:39 AM, Jérôme Hénin <jerome.henin_at_ibpc.fr> wrote:

> Hi,
>
> First, ABF is not the method of choice to compute free energies of
> hydration, unless you are interested in the hydration mechanism when
> the solute crosses the interface. But it should work anyway.
>
> Second, fluctuations of the shape and structure of the interface are
> going to be the main obstacle to convergence here. Actually splitting
> into several windows may do more harm than good here, because it may
> trap the system in local minima. I would use a single window spanning
> the complete interface.
>
> Best,
> Jerome
>
> On 8 December 2017 at 00:22, Yaxin An <yxan2016_at_gmail.com> wrote:
>
>> Hi All:
>>
>> I am running ABF simulations for the free energy of hydration of
>> alkanes

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:20:44 CST