[gmx-users] PMF with the pulll code

[Justin MacCallum]

Hi George,

in principal you should be able to extract a potential of mean force from an
AFM run by simply integrating the average force along the reaction
coordinate. Ie. if you have the average force as a function of position, you
could simply integrate that to get the PMF. In practice however, you will be
pulling orders of magnitude too fast for this to work. Instead you will
simply see the huge amount of irreversible work that you are doing.

In this respect, umbrella sampling is much more advantageous. A number of
people, myself included, have been able to get very reasonable PMF's using
umbrella sampling. It does require substantially more computer power because
you need many seperate simulations, but they are all independent and are a
perfect application for a cluster of Intel/AMD based nodes.

The pull code that is in the released versions of GROMACS has been in
various states of disrepair. I have actually made a substantial number of
changes and am hoping to get them integrated into the CVS tree. I'm hoping
to have that done in the next few weeks.

Cheers,
Justin

> Sorry if you got this message twice. I am using the GROMACS pull code
> trying to reconstruct the PMF of a small molecule through a membrane
> transporter. I have data from an AFM type of calculation but I am
> wondering how could I get the PMF out of them? Is there any relevent
> gromacs analysisi program?
>
> Is the umbrella type of calculation of the pulling code any more
> advantageous in this respect?
>
> Thanks in advance
> George Patargias