On Fri, Feb 17, 2012, Ilyas Yildirim wrote:
>
> Niel is right. The atom types of atoms C6 and C8 (pyrimidine and purine)
> in parmCHI_YIL are C3/C4 (Cytidine/Uridine) and C1/C2
> (Adenosine/Guanosine).
>
> [Check $AMBERHOME/dat/leap/parm/frcmod.parmCHI_YIL].
This is a symptom of a larger problem with Amber parameter development: we
have no official procedure to prevent atom types from being used
inconsistently in different force fields. In the present case, the Amber
united atom people add atom types C1, C2 and C3 to represent united atom
carbons:
C1 13.018 0.878 sp3 aliphatic C(UA model CH )
C2 14.026 0.878 sp3 aliphatic C(UA model CH2)
C3 15.034 0.878 sp3 aliphatic C(UA model CH3)
Independently, Ilyas used C1,C2,C3 and C4 for all-atom carbons in nucleic
acids. This might not have caused problems, except that GB radii rules
were modified for the united atom case, then applied to the nucleic acid
case, where they are not appropriate.
Lachele (thanks!) has started a doc/AtomTypesTableWorkspace.lyx (in git),
which is not yet complete (the last three columns seem to only be
partially filled in.) Once this is complete, we can get rid of unused
atom types in many of the learpc files, and identify (then correct)
the ones with conflicts. Going forward, we need to ensure that anyone
adding new atom types enter their choices in this table (or some similar
database), and ensure that they are not getting conflicts.
We also need a careful audit of the routines to compute atomic Born radii:
these are computed in leap (in unitio.c) and in ParmEd, maybe elsewhere.
I'm cc-ing this to amber-developers. Thanks to Niel and Ilyas for alterting
us to the problem.
....dac
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Mon Feb 20 2012 - 06:30:02 PST
