[gmx-users] Re: polymer simulations in gromacs

[ms]

On 05/12/10 19:41, Justin A. Lemkul wrote:
>
> Please keep all Gromacs-related correspondence on the gmx-users list. I
> am not a private help service.
>
> I do not do polymer simulations, nor do I have any real clue as to how
> to help you, aside from suggesting the following, which may or may not
> be relevant:
>
> http://www.gromacs.org/Documentation/How-tos/Polymers
> http://www.gromacs.org/Documentation/How-tos/Tabulated_Potentials
>
> I am CC'ing this message to the list. Please keep all further posts
> there, as perhaps someone can actually help you.
>
> -Justin
>
> Björn Nadrowski wrote:
>> Hi Justin, I just found out that you are simulating polymers using
>> gromacs. I would like to do the same stuff, but from a much less
>> molecular angle.
>> I am the only one here that does that stuff, so I have to jump through
>> all the loops myself, and the gromacs loop is really tiny and hard to
>> jump through.
>> Ideally, I would just like to define my polymer as a succession of
>> identical nodes, that have a user(i.e. defined by me) interaction
>> potential (essentially hard core repulsion with an intermedeiate
>> attractive regime), some 3-node angle stiffness term, friction (no
>> water! Just the polymers with friction!), position restraints (all
>> bonds between the nodes should be constrained to have the same
>> length), and
>> some boundary conditions.
>> How would I go about that?
>> I have scanned the documentation, but I am beaten to death by its
>> volume, where most of it concerns stuff I am not
>> interested in. Ideally, I would like to represent my nodes as some
>> kind of idealistic atom with said interactions. However, I do not find
>> the part where I define such an atom, or how to do that. Can you point
>> me to some source where I might get a nice simple tutorial of how to
>> define such an elementary node, and its interactions?
>>

Hi Bjoern,
Your experience is much like mine -I am actually doing something very 
similar, a coarse grain to model proteins which has very very similar 
features (I even suspect we're starting from the same literature, given 
the description). My advice is:

- Read the documentation. Multiple times. Understand it. *All* of it 
concerns stuff you *will* be interested in (apart perhaps some specific 
analysis tool, but even there...). Yes, it's long, but there's no hope 
of going anywhere without knowing it.
- Look at how residues and force fields are defined. Learn to create 
your own force field. The easiest thing is probably taking an existent 
force field, duplicate it and then work on the copy customizing it to 
your liking (That's what I did).
- Also, look at how protein residues are defined for use with pdb2gmx 
etc. - even if you are designing a non-protein polymer, it all basically 
starts from there, you then edit the residues etc. until you get the 
"unit" you want.
- Learn about tables and tabulated potentials if VdW functions are not 
enough.
- Learn about charge groups, energy groups etc.
- Remember that hardcore repulsion doesn't play well with MD 
integration: you'll spend a lot of time tuning it to be stiff enough but 
not too much to make your system explode
- About friction, do you mean Langevin dynamics?
- Do a lot of tests :)

It's very general advice, I know, but there's no real tutorial for such 
a thing, every custom model is... custom, and everyone, wanting 
different things, is quite on its own.

Best of luck! It took me about 1 - 1.2 years to get the model begin to 
work, so be patient and don't despair!

cheers,
M.

-- 
Massimo Sandal, Ph.D.
http://devicerandom.org