[gmx-users] QM/MM optimization in gromacs/gaussian (nikolaev at spbau.ru)

Groenhof, Gerrit ggroenh at gwdg.de
Fri Mar 16 13:06:29 CET 2018


Hi,

Is the potential energy still going down, despite the rupture of the QM region?

Input is same for EM and MD, and the fact that you can perform QM/MM dynamics looks promising. Are you using constraints ?

What happens if you do 100 steps of CG, using jus MM,but all other parameters the same?

GErrit

Well, the QM/MM dynamics for this protein goes well (except the fact that
one CH3 group in the QM part rotates, i.e. is not equilibrated well).
I have optimized this structure in gaussian using quite high-level QM/MM.
So there must not be any problems with this structure, I used it for a
number of purposes.

Should be there something different in the QM/MM files for optimization
and dynamics? Maybe something is just calculated wrongly, because in the
end (after 100 steps of cg minimzation) the whole protein just tears
apart, so it is definitely not the problems with the structure.



> Hi,
>
> Could the rupture suggest that perhaps the underlying QM/MM model is
> somewhat flawed?
>
> What happens when you run the minimisation without the constraint between
> the QM and MM atoms? Do you have convergence problems?
>
> Gerrit
>
>
>
>
> Message: 2
> Date: Fri, 2 Mar 2018 13:01:09 +0300
> From: nikolaev at spbau.ru
> To: gmx-users at gromacs.org
> Subject: Re: [gmx-users] QM/MM optimization in gromacs/gaussian
>         (nikolaev at spbau.ru)
> Message-ID: <40fa81b002d9ded18e6f471f19a1ada7.squirrel at mail.spbau.ru>
> Content-Type: text/plain;charset=UTF-8
>
> And which one (and with what parameters) is better for the QM/MM
> optimization?
>
> Steep almost didn't change anything and I couldn't relax the system as I
> wanted.
> CG seemed to work fine, but in the end it teared an H-atom from my QM
> subsystem.
> BFGS could not work with the constraints which are necessary for the QM/MM
> calculations in gromacs...
>
>
>
>
>>
>> Hi,
>>
>> It is possible, but only using gromacs' internal optizers: SD(steep),
>> CG,
>> or BFGS.
>> And you can only optimise minima, not transition states.
>>
>> Best,
>> Gerrit
>>
>>
>>
>>
>>
>>
>>
>> Message: 3
>> Date: Thu, 1 Mar 2018 12:44:53 +0300
>> From: nikolaev at spbau.ru
>> To: gmx-users at gromacs.org
>> Subject: [gmx-users] QM/MM optimization in gromacs/gaussian
>> Message-ID: <ff117f222d3d76fdce88a36282ae90b6.squirrel at mail.spbau.ru>
>> Content-Type: text/plain;charset=UTF-8
>>
>> Dear all!
>>
>> I need to perform the QM/MM optimization in the Gromacs/Gaussian
>> interface. However, I know that in 2015 this was not possible.
>>
>> The question: is there such an opportunity nowadays (I use gromacs
>> 5.1.2)
>> and which kind of parameters I need to write in the .mdp file in order
>> to
>> obtain such an optimization?
>>
>> Thank you in advance,
>> Dmitrii
>>


More information about the gromacs.org_gmx-users mailing list