@Francesco, I would be curious to know what the specific problem is that you are/were having when comparing GROMACS and CHARMM. If bugs go unreported, we can't fix them! We recently released the CHARMM36 force field (including CGenFF 2b8) and the energies of all molecules and combinations thereof (we tested dozens, if not hundreds ranging from single molecules to fully hydrated protein-ligand complexes) matched perfectly between the two packages.

It has been commonly stated elsewhere that GROMACS seems particularly buggy, and that claim may be, in some part, due to the very transparent bug-reporting and code-reviewing processes. It would be incorrect to think that any software is perfect, and all the major packages have had serious errors in the past, and may still.

Gromacs combined with other softwares like APBS, ElNemo, VMD and PyMol becomes very powerful software for molecular simulation studies... more that that, its free and u can still publish papers... what more you want!!!!!!!!!

Gromacs combined with other softwares like APBS, ElNemo, VMD and PyMol becomes very powerful software for molecular simulation studies... more that that, its free and u can still publish papers... what more you want!!!!!!!!!

If you're dealing with large complexes, I would say it's potentially faster/more efficient than other packages.

It's specifically designed to be an open-source software package, and it's easier to re-modify than the other packages. So if you're CS-savy and want to do some 'hacking', it's more convenient, than say NAMD, IMO.

+ (Bias: I personally knew a couple of GROMACS developers when I worked @ ORNL, who did multi-million atom simulations w/ it.)

The Schrodinger library accomplishes all functions of GROMACS with better efficiency.

However, Giovanni is right, Gromacs software is free for good.

Gromacs is free, can manage large size systems and is very fast.

Moreover, it can use charmm AND amber forcefield.

It also includes some analysis tools.

Bon soir Andre. Connaissez vous la bibliothèque Schrodinger?

Nope, I only know PyMOL from Schrodinger. In what ways it it better than GROMACS? speed? accuracy?

Optimized routines. Very inteligent solutions. Please, tell me, you dont speak French? It is unusual for a person that works or studies at Université d'Evry-Val-d'Essonne. Sorry for my curiosity.

Gromacs is great - most of the simulations in my group are done with Gromacs + some with NAMD2 - but NAMD2 and Amber are equally good choices. None of them offers a major advantage over the other ones. As for accuracy, see Sapay & Tieleman. "Combination of the CHARMM27 force field with united-atom lipid force fields", J Comput Chemi 32, 1400-10 (2011). In addition to the force field comparison, they also tested Gromacs, NAMD and CHARMM software using CHARMM27 force field (see Table 2).

Concerning speed: some benchmarks can be found in the gromacs4 paper http://www.mpibpc.mpg.de/home/grubmueller/publications/PDFs/Hess_2008_JCTC_4_435.pdf

bottom line is that gromacs is much faster on the cpu on a single machine (eg parallelization on 8 cores). Namd is faster when it comes to use hundreds of cores (clearly you need a low latency network for that).

Concerning reliability: gromacs is not always accurate, for example charmm energies are off by approx 1 kcal/mol (nobody explaining the reason). In some cases can be buggy. Namd is a much more solid platform.

Concerning force fields: gromacs can use basically all force fields available: charmm, amber, opls and gromos. Topology and parameter files are easy to understand (very good documentation btw). Namd offers charmm and amber (maybe also opls) but it is profiled to be 100% charmm compatible.

Both namd and gromacs are free but the latter is gpl

http://en.wikipedia.org/wiki/GNU_General_Public_License

Yep, on a small number of CPUs GMX is definitely faster. Amber and NAMD have great GPU versions, haven't tried GMX's GPU version (yet).

Thanks Mikko for the additional infos.

Next version of gromacs (due in summer/fall?) will implement a more mature GPU version of the program they said. At the moment, gromacs is not very good on the GPU both because the speedup is not great and few algorithms are implemented (eg no charmm cmap). Amber is very fast on the GPU. NAMD does not run 100% on the gpu, consequently it is fast only when coupled with the cpus. As today, fastest program on the GPU is ACEMD (single licence, free). If you plan to run on the GPUs I suggest to give it a try. We bought licenses and speedup is amazing (100ns/day on a 20K atoms system).

Great I am doing well now with this, I am trying out all of them yu all mentioned, though I am farmiliar with some

I gotta try ACEMD, thanks for the info Francesco!

Can any one send me the Gromacs installation procedure for Windows or linux. I have tried but i can not install successfully. please help me...... my email id: [email protected]

On linux you could try

sudo apt-get install gromacs

^ More specifically if you're on ubuntu. apt is only on ubuntu (possibly debian, though I am not sure). On linux, certainly try through your package manager first. If your package manager does not have the appropriate install then you can try compiling from source.

(I haven't tried to install GMX, just looking at the thread and thought I would slightly correct the above post).

In case you don't want to use the package manager (such synaptic on Ubuntu or apt-get [Ubuntu] on command line, or comparable on Fedora, Suse, etc), here are the instructions for compiling & dependencies: http://www.gromacs.org/Documentation/Installation_Instructions

@Francesco, I would be curious to know what the specific problem is that you are/were having when comparing GROMACS and CHARMM. If bugs go unreported, we can't fix them! We recently released the CHARMM36 force field (including CGenFF 2b8) and the energies of all molecules and combinations thereof (we tested dozens, if not hundreds ranging from single molecules to fully hydrated protein-ligand complexes) matched perfectly between the two packages.

It has been commonly stated elsewhere that GROMACS seems particularly buggy, and that claim may be, in some part, due to the very transparent bug-reporting and code-reviewing processes. It would be incorrect to think that any software is perfect, and all the major packages have had serious errors in the past, and may still.