[gmx-users] Protein Solvent Dynamics - Coordinates for docking

[Mark Abraham]

ram bio wrote:
> Dear Mark,
> 
> Thanks and you are right, that when a docked complex (protein +
> ligand) is simulated, the favorable ligand binding poses can be
> predicted using MD (longer runs). What I am trying to do here
> presently is not to simulate a docked complex, but to generate a
> modelled protein lowest energy configuration (P.E. surface
> exploration) using MD and further use this configuration for the
> flexible docking, and you are right as I think MD follows the law of
> conservation of energy as the P.E decreases the K.E increases or vice

Sure, under NVE, and even approximately so under other ensembles.

> versa for a configuration at an instance, as we are exploring the P.E
> surface by changing the coordinates very effectively using M.D, i want
> use the configuration (coordinates) with the lowest P.E produced  from
> MD and further minimize it (as i dont know whether the configuration
> obtained after gromacs MD can go further into a local/global minima by
> minimization) so that can be the input for flexible docking.

MD *as a sampling tool* is all very well, but the structures you want to 
get from the sample are not necessarily the low PE ones. You want the 
ones representative of the *free energy* minima, which need not 
correlate at all with low PE :-) Constructing such a free energy surface 
is a non-trivial undertaking, which is why people historically clustered 
trajectories, etc.

This comes back to how the docking methodology was derived. If their 
ligand binding targets were energy minimized to a certain degree, then 
you should probably reproduce that method on whatever you take for your 
set of target conformations. Hopefully their parameterization can deal 
with the intrinsic paradoxes of the situation. However if the target has 
some other structure, or range of structures, in the real system, then 
you have a (garbage in -> garbage out) docking study. Hence wanting a 
range of targets chosen by a useful criterion - or at least a study 
demonstrating that a small range is representative.

Either way, long MD (maybe REMD) and something like clustering of the 
output trajectories is a much sounder approach. But perhaps the docking 
forcefield is so random that it doesn't matter much :-) Literature is 
your friend (and I make no claim of being current with docking literature!)

Mark

> On Thu, Oct 22, 2009 at 4:59 PM, Mark Abraham <Mark.Abraham at anu.edu.au> wrote:
>> ram bio wrote:
>>> Dear Mark,
>>>
>>> You mean that I should consider a configuration with the lowest total
>>> energy for docking studies..Please clarify and suggest me.
>> If you read your docking program's documentation, they are using their
>> "energy" as a some kind of approximation to a free energy of ligand binding
>> (or some such). There could be all kinds of ad-hoc contributions that they
>> may have been able to demonstrate worked usefully enough on their test set
>> to be worth including.
>>
>> There's no reason to assume an MD potential "energy" (which itself is a
>> combination of more-or-less ad-hoc parameters) would correlate with the
>> above, since the MD potential wasn't parameterized to do that. It'd be
>> especially flawed to suppose that the fact that energy in MD is distributed
>> over PE and KE is immaterial. Low PE just means high KE.
>>
>> To identify favourable ligand-binding orientations with MD requires an
>> immense amount of sampling. In effect, you need to do enough MD to allow the
>> ligand to come in and out of the site many times in many different ways and
>> to compare the relative frequency so that you can also estimate the entropy
>> component of the free energy change associated with various binding modes
>> relative to each other. Even for implicit solvent calculation models, the
>> history of computing probably doesn't provide enough cycles to do this with
>> decent accuracy. Hence, docking.
>>
>> MD can be useful in more limited "docking" studies - a highly unfavourable
>> binding conformation will fly apart rapidly - but you'd hope the docking
>> force field could tell you about those cases!
>>
>> It sounds like you should really be doing some more background reading about
>> docking and MD, to better understand the methods you're using.
>>
>> Mark
>>
>>> On Thu, Oct 22, 2009 at 3:56 AM, Mark Abraham <Mark.Abraham at anu.edu.au>
>>> wrote:
>>>> ram bio wrote:
>>>>> Dear Justin,
>>>>>
>>>>> Thanks for the suggestion and advice.
>>>>> As i have used a modelled protein and want to obtain the lowest energy
>>>>> configuration of the protein by doing dynamics,
>>>> That's all very well, but what will that give you other than a set where
>>>> the
>>>> partition of total energy into potential and kinetic was skewed one way?
>>>>
>>>> Mark
>>>> _______________________________________________
>>>> gmx-users mailing list    gmx-users at gromacs.org
>>>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>>>> Please search the archive at http://www.gromacs.org/search before
>>>> posting!
>>>> Please don't post (un)subscribe requests to the list. Use the www
>>>> interface
>>>> or send it to gmx-users-request at gromacs.org.
>>>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>>>
>>> _______________________________________________
>>> gmx-users mailing list    gmx-users at gromacs.org
>>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>>> Please search the archive at http://www.gromacs.org/search before posting!
>>> Please don't post (un)subscribe requests to the list. Use the www
>>> interface or send it to gmx-users-request at gromacs.org.
>>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>>
>> _______________________________________________
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>> Please search the archive at http://www.gromacs.org/search before posting!
>> Please don't post (un)subscribe requests to the list. Use the www interface
>> or send it to gmx-users-request at gromacs.org.
>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the 
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>