Re: [AMBER] Can i combine the result given by sander and pmemd together

From: David A Case <case.biomaps.rutgers.edu>
Date: Sat, 6 Dec 2014 09:14:30 -0500

On Sat, Dec 06, 2014, wangmt wrote:
>
> I want to calculate the free energy using TI. I have finished the
> calculation corresponding to lambda = 0.1~0.9 by sander. Since i have to
> use the soft core potential which is invalid in sander when lambda=0 and
> 1, so when lambda = 0 and 1, i calculated that by pmemd . Then the
> question is that can I combine the two together to give a whole result
> from lambda=0 to 1.

To add to the previous reply: If you use numerical integration rule (Gaussian
integration, Simpson's rule, etc) you can get the "whole results" just from
the sander numbers. If you want to get a bit more precision, you can add
endpoint information from pmemd. But be sure that you look visually at the
integrand, and that it is smooth throughout the whole region (i.e. that the
endpoints from pmemd are where you would expect them to be, based on the
sander results.)

Of course, if the pmemd points are not in the expected locations, you can't
combine the two data sets. And if they are in the expected locations, you
haven't learned much by doing the extra calculations.

[Note: it *can* be helpful to consider lambda = 0 and 1 if you are trying to
get very precise free energy estimates, e.g. to compare different sampling
methods or force fields. This is usually not needed for comparisons to
experiments, where force field and sampling errors, and not integration
errors, generally dominate such comparisons.]

...good luck....dac


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Dec 06 2014 - 06:30:03 PST
Custom Search