[gmx-users] GROMACS vs GROMOS

[Erik Lindahl]

Naveen Michaud-Agrawal wrote:

> Hello,
>
> What are the differences between GROMOS and GROMACS? They both seem to
> come from the same group. (And I mean more than just the fact that GROMACS
> is open-source. What are the differences in capabilities?) Thanks.

Hi,

Actually the programs are not really from the same group, although both of
them have been
developed partly in Groningen; to make a long story short (I'm sure David can
fill in here :-),
GROMOS is mainly written by Wilfred van Gunsteren who is now in Zurich. The
original
idea behind GROMACS was to write a parallell md code, but many of the
algorithms were
borrowed from GROMOS. In the last 10 years or so both programs have developed
considerably;
Wilfred's group has put a lot of effort into better forcefields while the
work on GROMACS in
Herman Berendsens group was more oriented towards making the code faster.

Since I'm quite partial in this I won't even try to make an objective
comparison of the merits
of GROMOS and GROMACS. The main reason we are putting more work in the code
is that
GROMACS is 3-10 times faster than any other MD code we know of, but I'm sure
there are
cases where other features are more important.

GROMOS costs money, but it's only a nominal fee for academic departments. It
can also do 4-dimensional
simulations, which we cannot. On the other hand, GROMACS can for instance use
PME electrostatics,
it runs in parallel (there are special parallel versions of GROMOS, though),
includes Nose-Hoover and
Parinello-Rahman temperature & pressure coupling, and uses general triclinic
boxes.


Cheers,

Erik