[AMBER-Developers] News from the CHARMM/Amber developers meeting

From: David A Case <case.biomaps.rutgers.edu>
Date: Mon, 16 Jul 2012 16:32:46 -0400

Hi everyone:

We had a useful CHARMM/Amber developers' meeting July 10-13. Thanks to those
who attended and contributed. The meeting was more scientific (and less
code-oriented) than is typical of Amber meetings, but we got to hear a lot in
general about what people are working on.

Some notes for our community:

1. Holger and his group have prepared "FEW" (free energy workflow), which
automates several types of free energy, including LIE, MM/PBSA and TI
calculations. A version of this will be uploaded to the git repo with a few
months (as I understand it).

2. Lachele and friends have made "gmml" and "gems", which allow a lot of
building to be done, and have most of the functionality of tleap. One key
point is that prmtop files made with tleap and gems can be directly compared.
You can look at this now (google on "gems glycam"), but a more polished
version will show up in the git repo, again "soon".

3. The Amber group agreed that we need some better controls on what goes into
pmemd. The general ideas were these:
    a. new ideas and capabilities should be implemented first in sander (so
       that pmemd continues to be a subset of sander);
    b. those who wish to add something to pmemd (certainly in the "master"
       branch) should circulate a one- or two-paragraph description to the
       amber-developers list, in order to get feedback. Ideally, we want the
       "best" and most useful things in sander to be ported.
    c. we need to periodically make sure that changes in pmemd don't adversely
       affect either performance of other things, or the memory footprint.
       Right now there are memory issues for both CPU and GPU versions of
       pmemd that need to be addressed.
It is possible that there may be exceptions to rule "a", but requests for such
would need a strong justification.

4. We agreed to create a new web page that will explain the ins and outs
of Amber force fields more clearly, to help users understand what they
are getting, to make clearer comparisons between alternatives, and to
understand how we name things. I will make an initial stab at this, with
Lachele to do a second edit, and we'll then ask for general input. I'm
thinking of this as a much-expanded version of Chapter 2 of the AmberTools
reference manual, which will have tar files of parameters as well as
up-to-date information. The format should allow people to dig down to
underlying detailed information if they wish.

The "gaff2" and "pol12" parameterizations will be made available to Amber
developers soon (probably in a 2-6 month time frame, no promises), before any
release to a wider public. We didn't decide anything about a coordinated
release, but the web page described above would be a logical place to collect
any announcements about new parameter sets.

5. Rumors are that CHARMM may soon get a big performance boost; so we need to
continue to work on that area ourselves, as well as to (continue to) have
force fields of the highest quality.


David A. Case                     | case.biomaps.rutgers.edu   
BioMaPS Institute and Dept. of    |                             
    Chemistry & Chemical Biology  |    fax:      +1-732-445-4320
Rutgers University                |    office:   +1-848-445-5885
174 Frelinghuysen Road, Rm. 208b  |    cell:     +1-609-651-6817
Piscataway, NJ 08854        USA   | http://casegroup.rutgers.edu
AMBER-Developers mailing list
Received on Mon Jul 16 2012 - 14:00:03 PDT
Custom Search