[gmx-users] Space

[Erik Lindahl]

Chris Shaw wrote:

> Hi All,
>
> Hope all are well. I have got a problem regarding space. I am now 
> beginning to produce some complex runs which are taking up rather alot 
> of disk space on the system at work. Is there a way to save the md run 
> at several different points and "merge" them together at the end on a 
> different system to analyse the data.
>
> Many Thanks
>
> Chris
>
Sure.

First, don't write the full precision .trr files too often - you only 
need it to restart from a crashed run. To analyze the coordinates it is 
enough to save the compressed .xtc trajectory. You can also select to 
only write a subgroup (like protein, without the solvent) to the xtc file.

And, there's usually no point in writing coordinates to often. If you 
are studying motions on a picosecond scale it is better to perform  a 
short separate run (say 1 ns) and use this to study the fast motions. 
Then you don't need to save the coordinates very often for your 
long-timescale simulations.

If you still want to split your trajectories in smaller parts, the 
easiest way is to reduce the number of timesteps in the mdp file, and 
then use tpbconv to extend and restart your simulation.  You can then 
use the program 'trjcat' to concatenate the trajectories back.

Cheers,

Erik