I talked to Junmei, Piotr and Ray. We propose,
1) Glycam will use "CY" atom type. Changes in other parts will have to be
made for consistency.
To avoid conflict:
2) Atom type of amino acids and nucleic acids will be named using two
upper case letters (e.g.: CA).
3) Atom types in GAFF will be named using two lower case letters, as is
the case now (e.g.: ca).
4) New atom types in Glycam will be named using upper case first letter
and lower case second letter (e.g.: Ca).
5) Atom types in lipids will be named using lower case first letter, upper
case second letter (e.g.: cA), the same as what Ross has done already.
In the long run, however, we need to develop a robust fragment-based
scheme. There are numerous advantages in the atom type based schemes. A
good example is the OPLS force field which is almost entirely based on
atom types. However, as the need of force field accuracy increases, the
number of atom types will inevitably grow. More importantly, some of the
critical parameters, such as partial charges, can not be reliably
parameterized using atom types. In fact, AMBER has been using a hybrid
atom type + fragment scheme. For example, the charges of nucleic acids and
amino acids are fragment-based whereas the bond parameters are atom type
based. This is also the case for glycam and lipids. We think it is time to
move fully to the fragment based approach. One clear advantage of the
fragment-based approach is the potential to develop accurate partial
charges for GAFF. However, this will take considerable effort to develop,
not least of which is the need to improve leap/gleap and the need to
develop the fragment databases. We envision that atom types will remain an
important part of the new scheme but their role will be mainly in the area
of connecting the fragments.
We also hope that this move will solve the long-standing issue of
chirality.
Let us know if you have any comments.
--
Yong Duan, Ph.D, Professor
UC Davis Genome Center and
Department of Biomedical Engineering
University of California at Davis
Davis, CA 95616
530-754-7632
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Wed Dec 14 2011 - 14:30:03 PST
