[gmx-users] Another questions about Encad inside GROMACS

[Erik Lindahl]

Hi,

On Oct 18, 2005, at 6:21 PM, Leonardo Sepulveda Durán wrote:

> Thanks  Erik for your fast response. My reference of Encad FF is  
> "Computer Physics Communications 91 (1995) 215-231"; i have some  
> questions about that. First, levitt used a Brook's modification of  
> Beeman's algorithm to mantain energy. I have read from daggett's  
> works this method is very conservative, better than normal CHARMM  
> shift function, and very similar to FSHIFT, using only cutoff  
> without PME for electrostatics (Biochemistry 2005, 44, 609-616).  
> Using Beeman and other considerations, levitt showed a 2fs can be  
> used without SHAKE. My questions about that are:
>
> 1.  Beeman can be used for integration with encadff, to avoid using  
> SHAKE inside GROMACS?

No. Your best bet for long timesteps in general is to use LINCS,  
possibly in combination with dummy hydrogens.

> 2.  Which is better for energy conservation? Leap frog (Which is  
> the default GROMACS integrator, isn't?) Velocity verlet, verlet or  
> Beeman? I found a reference about VER and SHAKED VER (J. Mol. Biol  
> (1983) 168:595-620) but not comparisons with LEAP. Do you have some  
> proper reference?

All verlet class integrators yield the same trajectory, but the  
velocity accuracy depends on how you do the averaging.

> 3.  In pdb2gmx two encadff are available. Which is the one  
> described in Levitt1995??
>
"s" is the solvent one, "v" the vacuum version (scaled down charges).

> 4. In levitt1995, a Asc scaling factor is applied to repulsive LJ  
> potential in order to compensate loss of dispersive interactions  
> due to cutoff. Does encadshift uses that scaling?? If that is the  
> case, Is required to define it explicitly, or the program decides  
> based on encadshift value???
>

You can use true integrated dispersion corrections in Gromacs.

I would not recommend Encad for accurate solvent simulations at this  
point - the primary use is the vacuum version for rapid  
discrimination. Some trials we've done indicate that OPLS-AA/L  
combined with TIP4p produce cRMS values for some small test proteins  
around 1.1-1.2A, while Encad & F3C gives >=2A.

Cheers,

Erik
---------------------------------
Erik Lindahl  <lindahl at sbc.su.se>
Assistant Professor, Stockholm Bioinformatics Center
Stockholm University, SE 106 91 Stockholm
Phone: +46 8 5537 8564     Fax: +46 8 5537 8214




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20051018/1be353e0/attachment.html>