I'm curious as to why the 'ho' vDW parameters in gaff.dat were set to mimic
the OPLS hydrogens, with 0.00 vDW radii.
I found, that in neutralized phosphates and sulfates, this can lead to
disastrous effects, since 1-3 VDW repulsons are ignored in AMBER between the
two oh's, instead treated implicitly in the valence angle. If an 'ho' is
bumped close enough to its neighboring 'oh' (not the 'oh' that it is
covalently bound to), the attraction between 'oh' and 'ho' grows infinitely
large and causes LINMIN failures. This is because the oh-p5-oh angle term
is the only thing to deter this and maxes out at around +45 kcal/mol as does
the ho-oh-p5, which provides for only 90 kcal/mol of repulsive energy. Very
quickly, the 'oh' and 'ho' can be superimposed on top of one another and
their infinite electrostatic interaction dominates the 90 kcal/mol of
repulsive angle terms.
ho ho ho
/ / /
oh oh -----> oh-ho|oh
\ / | /
\ / | /
p5 p5
Since the vDW radius of oh is 1.721 and the bond distance between oh-ho is
0.96, a vDW radius for 'ho' of 0.6, like all the other h's in gaff.dat,
would still leave the 'ho' within the radius of the 'oh' *and* prevent the
atom superposition.
Is there a reason I missed for why the 'ho' radius should not be set to a
non-zero value?
I read a 1/2003 post by Case on the reflector discussin phosphates in
nucleic acid simulations, that talked about adding a covalent bond between
the P and the H. I can see that this would also help to deter the atom
superpositions and am wondering if that has any advantages over increasing
the vDW radius of the 'ho' to a non-zero value.
thanks,
--Matthew
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Fri Sep 05 2003 - 23:53:00 PDT
