Happy New Year!!
In leap, in addition to the traditional attributes associated with atoms,
one attribute was added, also called "atom types", for hybridization. If the
hybridization is not specified, it is assumed as sp0. In which case, leap
gives a series of friendly messages which is great. What is not so great is
that it also tries to figure out the best way to connect atoms by modifying
the coordinates using the unspecified hybridization. This is usually a
harmless exercise. However, occasionally, it has funny consequence. If the
atom missed in the hybridization list is a chiral center, it is possible
that the resulting molecule may have a flipped chirality.
This is really not a bug because the code behaves exactly as intended. But,
I feel that the behavior is somewhat idiosyncratic and improvements could be
made. For example, we had to struggle for several days to figure out just
exactly what went wrong. It should not surprise me if somebody else also had
similar interesting experience, albeit, perhaps not during holidays :).
We can of course change the code so such it aborts automatically whenever it
finds an atom with undefined hybridization (or crash the computer ...).
Alternatively, which I happen to think is a sensible approach, the input
coordinates should be used.
If, however, one must guess and change the input coordinates, can a more
sensible hybridization be used instead? How often do we see sp0? What is
sp0? By the way, it actually tells me that sp0 is an unknown hybridization
if I specify it (but why sp0 is the default ?!). Also, if I specify wrong
hybridization, things are ok too.
Thanks!
Yong
For those who want to see how does an amino acid with a flipped chirality
look like, try,
************
1) remove those "addAtomTypes" from any of the leaprc.ffxx files
2) tleap -f leaprc.ffxx
3) aa=sequence{ACE ALA NME}
4) savepdb aa aa.pdb
5) quit
***********
Put it on your favorite PDB viewer.
Received on Sun Dec 30 2007 - 06:07:36 PST
