A few weeks back I promised that our group would do detailed tests of
sleap's prmtop file generation. This is a report.
The overall recommendation:
Upgrade to AmberTools 1.5 patched at least with bugfix.3. For now,
use tleap to generate files for use with the GLYCAM06g force field.
Sleap should be patched soon, too.
More information:
Here is how we tested. We wrote a separate program to insert SCEE &
SCNB entries into pre-existing tleap-generated prmtop files. We also
modified tleap so that it writes topology files with SCEE & SCNB
entries. We compared prmtops generated by these to each other and to
prmtop files generated by sleap. We made the comparison by
post-processing an existing trajectory using all three prmtop files.
The system from which the trajectory was produced contained a protein
and an oligosaccharide that were in complex but not chemically bonded.
We processed the trajectory by running a single-point minimization in
sander on each frame. The results from the three prmtop files should
be identical if they contain equivalent information.
Here is what we found: The modified tleap and the program for
modifying existing prmtop files gave identical simulation results.
The results from the sleap-generated file were nearly identical to the
other two. After close inspection, we found one difference: an
improper dihedral was being double-counted by sleap.
How to interpret this: The sleap-generated files will not produce
terrible errors. The result is that certain planes will be
over-stabilized. In most cases, this will not cause trouble in a
simulation. But, it is a difference, and in some cases will matter.
Recommendation: Upgrade your AmberTools to at least version 1.5
patched at least through bugfix.3. Use tleap instead of sleap for
generating files that use the GLYCAM force field until the sleap
behavior changes. If you find bugs in the new tleap prmtop
generation, please contact me as soon as possible (on this list or
separately).
Other: In future releases of the force field, we will take steps to
ensure that similar mistakes are less likely to happen. The program
we used to add SCEE & SCNB entries to an existing prmtop will be made
available on the GLYCAM-Web site within a few weeks.
--
:-) Lachele
Lachele Foley
CCRC/UGA
Athens, GA USA
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri May 13 2011 - 12:00:03 PDT
