Hi Oliver,
> Now I wonder: If I start a calculation with say 4 lamda-steps and then
> want to produce data for some more lambda steps later on, shouldn't I
> have planned the further steps when doing the 4 steps?
Yes, you should have. In TI you can add more points iteratively, AFAIK,
FEP is not suited to this.
> My idea would be: Principially setting the lowest lamda to 0.05 and
> highest to 0.95 and interval to 0.05 to write the potential values. (This
> would most probably be the finest resolution of intermediates I would like
> to do).
For FEP you need to sample from l=0 to l=1, unless you have some good
scheme for extrapolation from 0.05 and 0.95 outwards. At the moment,
softcore potentials do not work at 0 and 1, so right now MBAR in Amber is
only practical for electrostatic changes. I am working to change that...
> Will this slow down the calculations much?
not much, you need uncorrelated samples anyway, so doing the BAR data
collection every 100th step or so is enough.
Also, I wrote a small script to extract Amber energies from the BAR module
into a format that fits for Shirts pyMBAR tools. It isnt included in
Amber11 yet. Let me know if you would like to test that for your system.
Kind Regards,
Thomas
Dr. Thomas Steinbrecher
BioMaps Institute
Rutgers University
610 Taylor Rd.
Piscataway, NJ 08854
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed May 05 2010 - 00:00:03 PDT
