Umbrella sampling

Umbrella sampling could be useful to obtain a PMF (and all it's derivatives) for systems where direct PMF calculation is time consuming (for example in the case of deep PMF minimum for the particular configuration).
Very nice tutorial is represented at http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/umbrella/01_pdb2gmx.html

Parallel simulation

If a system for example is sensitive to the initial parameters configuration, it is convenient to calculate a few systems in parallel. There are a few ways to create initial configuration:
  • to assign different initial velocities for particles in the system (different random generator number);
  • to start with different initial configuration (ex. Packmol or produce a short trajectory at high temperature and take different frames)
  • to combine different velocities and initial coordinates.
It is usually convenient to keep all files for one system in one folder, so that every system has its own folder. One could find a template for python script (template.py) that takes care of starting every simulation (or performing analysis) from its own folder.

Compare Gromacs and Desmond


Gromacs (4.5.2)
Desmond (2.4)
comment
speed of calculation(~10 000 particles)
on Halimede: 1.3 - 5 ns/day (10 CPU)
on Halimede: 8-15 ns/day (8 CPU)
depends on the network
support of electrostatics for slab geometry
+
-

documentation
a lot, but unstructured time to time
often insufficient, Desmond user group is rather slow

analysis tools
well documented, big amount
poorly documented => hard to find the proper tool for 'THE' task

trajectory size
3X
X

'A' molecular topology
-
+
mind 'Convert.py' script by Kathleen
maximum system size, particles
99999
-