Re: A problem in ABF simulation

From: Jérôme Hénin (jerome.henin_at_ibpc.fr)
Date: Mon Mar 13 2017 - 17:04:00 CDT

Hi,

Please keep the conversation on the mailing list.

Most likely, that simulation is not converging because of large
non-equilibrium effects, due to many slow degrees of freedom coupling to
your chosen coordinate. A single RMSD is rarely a good coordinate to
describe conformational change.

Jerome

On 12 March 2017 at 14:29, 宋瑞珩 <141130085_at_smail.nju.edu.cn> wrote:

> Thank you for your answer.
> Maybe you misunderstood what I meant. In the previous simulations, I found
> the initial state have a lower energy than the final state(about
> 100kcal/mol). However, in the ABF calculation, the initial state have a
> much higher energy than the final state(about 300 kcal/mol). I also checked
> the colvars.traj files, and found the sampling was just around the low
> boundary during the 150ns simulation. This is quite confusing.
> Besides, I choose 448 atoms for the definition of the coordinate.
>
> ------------------ Original ------------------
> *From: * "Jérôme Hénin"<jerome.henin_at_ibpc.fr>;
> *Date: * Thu, Dec 15, 2016 00:29 AM
> *To: * "Namd Mailing List"<namd-l_at_ks.uiuc.edu>; "宋瑞珩"<
> 141130085_at_smail.nju.edu.cn>;
> *Subject: * Re: namd-l: A problem in ABF simulation
>
> Hi,
>
> The Jacobian term in that PMF is approximately (RT*(3N-4) / x), so the
> difference of 100 kcal/mol you see is entirely expected if about 50 atoms
> are involved in the definition of the coordinate.
>
> Jerome
>
> On 14 December 2016 at 12:45, 宋瑞珩 <141130085_at_smail.nju.edu.cn> wrote:
>
>> Dear all
>>
>> I used the RMSD as a reaction coordinate to calculate the potential surface of a conformation change. I found the free energy decreased steeply with the reaction coordinate(the gradient is about -100kcal/A, this is quite a large value). However, the previous MD simulation in NPT ensemble revealed that the initial state(rmsd=1) had a lower energy than the final state(rmsd=4), and the energy difference between the two states was about 100kcal/mol. I have no idea about the paradox. Can anyone give some suggestions? Here is my input file:
>>
>>
>> Colvarstrajfrequency 100
>>
>> Colvarsrestartfrequency 1000
>>
>>
>> colvar {
>>
>> name RMSD
>>
>> width 0.01
>>
>> lowerboundary 1.0
>>
>> upperboundary 4.0
>>
>> lowerwallconstant 100.0
>>
>> upperwallconstant 100.0
>>
>> rmsd {
>>
>> atoms {
>>
>> atomsFile 3.5.pdb
>>
>> atomsCol B
>>
>> atomsColvalue 1.00
>>
>> }
>>
>> refPositionsFile ref.pdb
>>
>> refPositionsCol B
>>
>> refPositionsColValue 1.00
>>
>> }
>>
>> }
>>
>>
>> abf {
>>
>> colvars RMSD
>>
>> fullSamples 500
>>
>> }
>>
>>
>

This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:21:09 CST