[gmx-users] Tips and tricks to make your own coarse-grained model?

ms devicerandom at gmail.com
Fri Oct 9 14:45:06 CEST 2009

Justin A. Lemkul ha scritto:
>> My aim is to re-implement a formerly standalone CG model by mean of gmx.
>> It seems that gmx is well suited to the task; however I understand that
>> there is an incredible number of pitfalls where a newbie fall in.
>> I am currently reading and re-reading the manual, looking at the online
>> docs and tutorials. But most of information, understandably, refer to
>> all-atom modeling and there is little on how to build your own model
>> on gmx.
>> Maybe my googling skills are not as good as once they were -in this case
>> can anyone point me to any page where this kind of task is documented?
>> If such a page there isn't (and even if there is), I would appreciate a
>> lot if anyone of you can share your wisdom on best practices, pitfalls,
>> things to know etc. before diving head-on in such an endeavour.
> So, to clarify, this is an already-existing CG model that you simply
> want to use for simulations in Gromacs?  It is then rather easy.  You
> need to define all appropriate force field files (nb.itp, bon.itp, .atp,
> etc), and perhaps .rtp files if you can also come up with a way to
> define an appropriate CG input for pdb2gmx.

Yes. I know the specs of the cg model, and I want to re-implement it
within gromacs. Actually it is a class of related models. I'd start from
a very minimal one and then improve gradually.

I understand I have to define all appropriate FF files, but I wonder
what is the best way to do it, what documentation to learn by heart,
what sanity checks... You make me optimist in saying "it is rather easy"
, but science is always full of traps, especially for the newcomer...


More information about the gromacs.org_gmx-users mailing list