Re: [AMBER] Steered dynamics related questions

From: Sergey Samsonov <sergeys.biotec.tu-dresden.de>
Date: Tue, 26 Jul 2011 12:49:50 +0200

Dear Adrian,

thanks you VERY MUCH for the answers!

On 07/26/2011 12:01 PM, Adrian Roitberg wrote:
>
>> 1. Force of pulling proteins or receptor/ligand apart. In AMBER 10
>> manual example the constant of 5000 kcal/(mol A^2) is used. In the
>> literature I've seen absolutely different ranges of these values (down
>> to 1kcal/(mol A^2)). The explanations are a bit controversial: in some
>> papers the authors claim that the constant should be comparable to kT/2,
>> in other that this constant should be essentially higher in order to
>> have less effects from thermal motions. What do you think is appropriate
>> and what is the best way to optimize this constant for each individual
>> system? For pulling ion apart from the protein I noticed quite a
>> significant influence of this constant on the calculated free energy.
>>
> First, kT/2 has units of kcal/mol while the force constant has units of
> 1kcal/(mol A^2) (for a distance), so the text 'The explanations are a
> bit controversial: in some papers the authors claim that the constant
> should be comparable to kT/2,in other that this constant should be
> essentially higher in order to have less effects from thermal motions.'
> makes no sense. Can you make it clearer what you mean ?
>
Sorry for the typo, surely I meant comparison to the spring constant
(force/squared amplitude), which is characteristic (in terms of
amplitude, ~A, for example) for kT energy value of the related
oscillator. For example, as it mentioned here:
http://enzyme.fbb.msu.ru/Tutorials/Tutorial_3/
where "constant /k/=150 pN/A allows 0.5 A thermal fluctuations of the
constrained coordinate". In AMBER Manual the value is 5000 kcal/(mol A^2)

> The value of the force constant AND the speed of pulling depend
> crucially in what you want to do. Are you using this data to get some
> general idea of the protein-protein association pathway or you really
> want the free energy? If you want the free energy, then you need to
> repeat the pulling MANY times, from an initially 'equilibrated' sampled,
> where all the snapshots have the same distance. Then, look at Schulten
> papers where he describes the so-called stiff spring approximation. The
> force constant CANNOT be very small, otherwise what the pulling spring
> and your molecules are doing is very different. Basically, you pull with
> a soft spring, and the molecule just does not care and does not follow
> the distance.
>
Thank you, I'll try several values. I'm looking for free energies, so
first I'll try to get the convergence of the free energy by running N
simulations with some initial force and speed from one of the papers
with a similar system and then I'll try to see which values do the
parameters (force and speed) have to reach the fastest convergence.
>
>
>> 2. Speed of pulling. Again, I found various values in the literature
>> with the range from 1 to 100 A/ns. Am I right if I consider an ability
>> of the simulation to keep the structures of pulled apart proteins
>> undistorted as one of the most important criteria for the selection of
>> this parameter?
>>
> The speed of pulling is extremely system dependent. For free energies,
> there are too many tricks to do to explain in this space. Take a look at
> the paper below for some ideas.
>
> Xiong, Hui; Crespo, Alejandro; Marti, Marcelo; Estrin, Dario; Roitberg,
> Adrian E.. Theoretical Chemistry Accounts. 116(1-3):338-346 (2006)
>
That's great. Thank you!
>
>
>> 3. Is there an explicit way in AMBER (I use AMBER 10) to apply the
>> pulling forces to the centers of masses of protein/protein pair or to
>> all backbone atoms instead of choosing only two atoms? Otherwise could
>> an alternative solution be just putting restraints on backbone or
>> C-alpha atoms of the proteins and applying the pulling force on two
>> atoms (though this sounds to me being physically more ambiguous)?
>>
> Yes, look at group restraints. You can set a distance between center of
> masses of group for instance. Look at the manual also for natural
> language restraints.
>
>
>> 4. Is there the way in Leap to create an assymetric water box for PBC
>> with an elongated side, along which the pulling is occuring? This would
>> save the computational expenses a lot in case of SMD.
>>
> I am pretty sure this is not needed ! Since you are using PBC, the box
> is never really assymetric anyways. If you have your molecule to one
> side of the box, then the molecular in the next box is also displaced,
> so you gain nothing.
>
Thank you very much, this all helps a lot.

Best regards,

Sergey
> Adrian
>
>
>> Thank you very much in advance!
>>
>> Best regards,
>>
>> Sergey
>>
>>
>


-- 
Sergey A. Samsonov
Postdoctoral researcher
Structural Bioinformatics
Biotechnology Center
Tatzberg 47-51
01307 Dresden, Germany
Tel: (+49) 351 463 400 83
Fax:   (+49) 351 463 402 87 
E-mail: sergey.samsonov.biotec.tu-dresden.de
Webpage: www.biotec.tu-dresden.de
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jul 26 2011 - 04:00:02 PDT
Custom Search