Re: Umbrella sampling tilt with respect to Z-axis

From: Aron Broom (broomsday_at_gmail.com)
Date: Wed Oct 08 2014 - 18:54:44 CDT

I may be wrong, but it should be doable in WHAM. I thought some of the
tutorials are done using dihedral angles between two alanines, and in that
case you also go from -180 to +180. But I've never myself used dihedrals
for WHAM (I'm thinking a dihedral angle is more or less what you are
talking about here, yes?).

In the worst case scenario (if some WHAM oddity did make a high barrier),
you could build the PMF from -180 to +180, and then copy your data,
post-process it turn -179 to +181 and -1 to +359 (and everything
inbetween), then make another PMF and align the two to get a nice
continuous PMF from -180 to +360. I don't think that will be needed
though, have you looked at the manual for Alan Grossfield's WHAM code? (
http://membrane.urmc.rochester.edu/content/wham) it might have something
about this.

On Wed, Oct 8, 2014 at 7:07 PM, Mitchell Gleed <aliigleed16_at_gmail.com>
wrote:

> I will look into it that, thank you! That seems a reasonable solution, and
> one that would still allow for the ligand to explore all psi's and phi's
> over the distribution of umbrella windows. I'm still trying to grasp using
> either in US-REMD, but I think I'm getting a better understanding--my
> mentor and I have been debating about it a fair amount this afternoon.
>
> I could very well be on the wrong track here so please correct me if I'm
> wrong, but I've since had an idea that I could simply place the reference
> atoms so their tilt is along the Z axis--perhaps simplifying things a bit
> and still allowing me to use the tilt colvar. Any tilt deviation would then
> be away from 0 degrees and toward 180, eliminating the need to have a sign.
>
> Feeding the data into WHAM with this method, however, seems a bit
> problematic to me. For example, if the ligand is oriented toward zero
> degrees so cos(0deg)=1, then all of the colvar data will be less than 1.
> Would this example and it's opposite result in extreme barriers in the PMF
> at cos(theta)=1 and -1? Is this where the periodic flags are useful? I
> suppose this may be the wrong forum for WHAM questions, but if anyone has
> experience in that area, any suggestions would be welcome.
>
> Thank you!
>
> On Wed, Oct 8, 2014 at 2:45 PM, Giacomo Fiorin <giacomo.fiorin_at_gmail.com>
> wrote:
>
>> Hello Mitchell, if the "tilt" cosine function does not work for you
>> because of its symmetry, how about defining a "spinAngle" around an axis
>> orthogonal to Z? That should give you the tilt angle, but with a sign (if
>> I understand correctly this is what you are looking for).
>>
>> Giacomo
>>
>> On Wed, Oct 8, 2014 at 4:38 PM, Mitchell Gleed <aliigleed16_at_gmail.com>
>> wrote:
>>
>>> I am working on a procedure for 2D umbrella sampling of a ligand through
>>> a membrane protein. The two dimensions I hope to explore are Z-distance
>>> (using distanceZ from the colvars module) and tilt angle (hopefully using
>>> tilt from the colvars module) of the ligand to generate a 2-D PMF.
>>>
>>> I might not fully understand, but in my testing, and as explained in the
>>> User Guide, the cosine of the tilt angle is measured between the atoms of
>>> interest and reference atoms with 1 being parallel and -1 being
>>> antiparallel to the reference atoms. However, the protein is asymmetric,
>>> and a reported deviation in 10 degrees of the ligand, for example, doesn't
>>> tell me whether it tilted toward +Z or -Z space, where I would expect the
>>> ligand to behave differently. Furthermore, I'm not sure I could use tilt
>>> with 2D-US-REMD as I've planned (based on the
>>> NAMD_Source_2.10b1/replica/umbrella2d example), since replica bias centers
>>> would not distinguish between +Z/-Z tilt change.
>>>
>>> I'm considering post-processing the trajectories and measuring
>>> cos(theta) with respect to the Z axis and then using this data to generate
>>> the PMF, but disk space becomes a big issue, and it wouldn't solve the
>>> replica-exchange limitation. (At which point I could switch to 1D REMD for
>>> distance and employ constant tilt constraints.)
>>>
>>> Can the tilt colvar be adjusted to measure the cosine of the tilt angle
>>> with respect to an axis, rather than with respect to reference coordinates
>>> (such as Z)? Or is there another path I could take or colvar to use to
>>> study this dimension?
>>>
>>> Thank you,
>>>
>>> Mitch
>>>
>>
>>
>>
>> --
>> Giacomo Fiorin
>> Assistant Professor of Research
>> Institute for Computational Molecular Science (ICMS)
>> College of Science and Technology, Temple University
>> 1925 North 12th Street (035-07), Room 704D
>> Philadelphia, PA 19122-1801
>> Phone: +1-215-204-4213
>> https://icms.cst.temple.edu/members.html
>> http://giacomofiorin.github.io/
>>
>>
>

-- 
Aron Broom M.Sc
PhD Student
Department of Chemistry
University of Waterloo

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:22:55 CST