Re: ABF vs Umbrella Sampling entrance barrier heights discrepancy

From: Jérôme Hénin (jerome.henin_at_ibpc.fr)
Date: Thu Aug 10 2017 - 14:11:42 CDT

Here by wall I meant the part of the protein surrounding the pore.
Something similar might happen inside the pore, but only if it is
pathologically-shaped. If its border is smooth and it doesn't bend hugely,
I wouldn't expect it.

The classic solution for the pore opening is to restrain lateral diffusion
to a cylinder.

Jerome

On 10 August 2017 at 19:45, Olya Kravchenko <ovkrav_at_gmail.com> wrote:

> Hi Jerome,
>
> thank you for replying to my question. By "wall" you mean the boundary
> potential or the physical walls of the pore? Would there be a similar
> effect inside the pore, if the pore is wide?
>
> Olga
>
>
>
> On Sun, Aug 6, 2017 at 11:17 AM, Jérôme Hénin <jerome.henin_at_ibpc.fr>
> wrote:
>
>> Hi Olga,
>>
>> my best guess is that your ABF simulation is hurt by slow lateral
>> diffusion around the pore entrance. Essentially, the ion bumps into the
>> wall, giving the high barrier you are seeing. In umbrella sampling, the ion
>> doesn't do that - or anything else, for that matter. Therein lies the
>> beauty of that method.
>>
>> Jerome
>>
>> On 2 August 2017 at 20:39, Olya Kravchenko <ovkrav_at_gmail.com> wrote:
>>
>>> Hi all,
>>>
>>> I ran a number of ABF and umbrella simulations to obtain pmf for the ion
>>> permeation through the pores of a spherical protein and in all of my
>>> simulations I see discrepancy in the estimate of the barrier heights at the
>>> entrance of the pore (i.e. where ions enter the protein from the cytosol
>>> region). For example, umbrella sampling may give me ~20 kcal/mol, while ABF
>>> is ~100kcal/mol (I have more precise numbers, but the difference is more or
>>> less in this order). My reaction coordinate is the distance between the ion
>>> and the center of mass of the protein, i.e. the central axis of the pore)
>>> and my ABF windows are 2-4 A wide.
>>>
>>> I am wondering if anyone else sees this when comparing ABF with US? I
>>> have not found anything specific in literature yet. Essentially I am trying
>>> to figure out if there are known problems with ABF method near the entrance
>>> of the channel that I should have taken into an account while setting up
>>> simulations?
>>>
>>> I will appreciate any input.
>>>
>>> Olga
>>>
>>>
>>>
>>>
>>
>>
>

This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:21:33 CST