Hi,
I want to use TI calculations to estimate the effect of core variations on
ligand binding to a protein, i.e. I have a ligand with a small aromatic core
with which some substituents attached to it. Now I keep the substituents the
same, but vary the core by moving nitrogen atoms around (e.g. pyridine with
the N at the 1 position to a pyridine with N at position 3).
Now I am not quite sure how large I should make the part of the system which
is perturbed, I could think of several possibilities:
1) I only include positions 1 and 3 into the scmask and crgmask, as
these are the positions where the ligand changes (CH -> N and N-> CH). This
works and I do get reasonable results for the Free Energy change.
2) I include all atoms which are connected to positions 1 and 3.
3) I include all atoms of the aromatic core.
Does anybody have experience (or actually has studied) which would be the
appropriate size of the part of the system which is perturbed? Are there any
practical considerations (apart from that its bad/difficult to remove
everything and then let it re-appear :-), e.g. in terms of convergence?
Would there be a difference if I used a single-step procedure (ie changing
charge and vdw parameters simultaneously) versus the "historical" 3-step
procedure (remove charge -> change vdw -> recharge)?
Anybody aware of a publication which discusses this issue? I haven't found
anything neither in literature nor in the amber archives.
Thomas, if you read this: are there any limitations/features of the code you
are aware of that would strongly favor one scenario over the others?
Thanks a lot,
Th.
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Feb 24 2011 - 23:30:03 PST
