MD

I recently helped a friend compile an old version of CHARMM (c32b2) on his Ubuntu 10.04 64-bit machine. Quite a bit of troubleshooting was necessary, so here are some instructions for what worked out in the end.

1. Install the Intel compilers for both Fortran and C++. There are some good instructions on this page: http://hawknotes.blogspot.com/2010/05/ubuntu-1004-installing-64-bit-inte...
2. Go to your extracted CHARMM directory and edit build/UNX/Makefile_gnu and insert the Vaxlib compile flag for ifort:
   

If relatively early in a simulation you run into a Gromacs mdrun error with output like this:

step 300, will finish Mon Mar 28 07:57:40 2011
A list of missing interactions:
            G96Angle of   2341 missing      1

Molecule type 'POPC'
the first 10 missing interactions, except for exclusions:
            G96Angle atoms   10   11   12      global   647   648   649

-------------------------------------------------------
Program mdrun, VERSION 4.5.3
Source code file: /Users/admin/gromacs-4.5.3/src/mdlib/domdec_top.c, line: 173

Software inconsistency error:
Some interactions seem to be assigned multiple times
For more information and tips for troubleshooting, please check the GROMACS
website at http://www.gromacs.org/Documentation/Errors
-------------------------------------------------------

realize that the error text is a bit misleading.

Short answer: it can't yet be done.

I spent a bunch of time configuring my system to be able to compile a GPU-accelerated (via OpenMM) Gromacs binary. I should have read the latest Gromacs v4.5.3 release notes: "Force fields: Supported FF are Amber, CHARMM."

So despite the fact that many of the other system parameters for which the MARTINI forcefield is designed are not yet supported by OpenMM e.g.

WARNING: OpenMM does not support leap-frog, will use velocity-verlet integrator.
WARNING: OpenMM supports only Andersen thermostat with the md/md-vv/md-vv-avek integrators.

Note: newer (>4.0) versions of Gromacs will often automatically detect multiple cores or processors. So the following method may not be necessary.

As you might expect, the base Gromacs package for Ubuntu 10.04 (Lucid Lynx) does not contain libraries for running your simulation on multiple processors or clusters. To run multiprocessor MD simulations you have several options for packages:

* gromacs-openmpi
* gromacs-mpich
* gromacs-lam

It seems like most Gromacs fans use the utilities included with Gromacs for analysis of MD trajectories. This Gromacs documentation page lists external tools like MATLAB at the bottom of the list. In addition, it recommends the use of gmxdump for exporting coordinates (which is plain ugly).