[gmx-users] CHARMM-GUI to GROMACS

[Justin Lemkul]

On 10/21/13 11:14 AM, Zhi Yue wrote:
> Dear all,
>
> I built up my protein in lipid bilayer by CHARMM-GUI and now I want to do
> equilibration by GROMACS. I know GROMACS can build the system, but my
> subsequent production run must be conducted by CHARMM. I want to use
> GROMACS to do equilibration cuz it may take long to get it equilibrated.
> Now I get stuck in some problems.
>
> 1) I'm not quite sure about how to make conversion. I searched the archive
> of the forum. Someone recommended I split the pdb generated by CHARMM-GUI
> into different pdb files for protein, lipid, water and ions, and use
> pdb2gmx to convert them to gro files individually. Then combine those gro
> files and made *.top file manually. Someone also suggested I convert
> original pdb directly to gro. Which one is better?
>

Whichever one makes sense to you.  I prefer a "clean" approach where you deal 
with one molecule at a time, but others do it differently.  There's really no 
need to have pdb2gmx run through all the water, ions, etc. when really their 
topologies can easily be handled with simple #include statements.

> 2) I tried both protocols. When I checked the robustness of my structure by
> grompp, problems came. As CHARMM-GUI uses TIP3P and it would be better to
> use TIPS3P with lipid (we have tested them with CHARMM36 force field), I'm
> not sure how to change water model. Another problem, the most troublesome,

You change the water model by #including the right topology, hence my approach 
above.  The CHARMM27 distribution that is built into Gromacs uses tip3p.itp for 
the "classic" model and tips3p.itp for the CHARMM-specific model.  Our lab 
recently released a full CHARMM36 force field, with tip3p.itp containing both 
models, toggled by using #ifdef blocks.

> is that all water molecules whose residue ID's after 9999 could not be
> properly designated (say waters 10000TIP to 10009TIP become 1000TIP which
> has 30 atoms). I don't know why it happened. Is there anything wrong with
> my converting logic in question 1)? I mean, I should not convert water or
> ions, should I?
>

The residue numbering is irrelevant, and is simply an outcome of PDB format, 
which only allows for 5 digits in residue numbers.  It has no implications for 
the soundness of the topology or the simulation.

> 3) CHARMM-GUI uses rectangular or hexagonal boxes but they are supported by
> GROMACS. I'm wondering whether there is any direct way to match the box
> type in GROMACS.
>

Both are supported natively.  See manual section 3.2 and figure 3.2.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441

==================================================