Hi
We end up with a similar conclusion as you, i.e. we believe that the
simulations are ok (probably) but there is something system dependent
that we do not understand. But when I think about what I am doing more
carefully I get confused. You see, in the way I am doing the
calculations (identical to the tutorial), there is nothing really
special with a +2->+2 mutations compared to a +2->+1 mutation. In the
first step step I always remove the charge on the mutating atoms (a
C(NH2)2 group in my +2->+1 mutation) and in the third step I always
introduce a charge on the mutating group a (CONH2 group in my +2->+1
mutation). So the only thing that is "changed" from a +2->+2 mutation
is the amount of charge I remove/add. I am correct here? As far as I
understand it. it is in these two steps I should look for errors. So
then I think that we get trouble since we are trying to remove/add a
too large charge, but should not this be cured by taking smaller
steps, i.e. more lambda-values? But we have tried this and have not
got any better results.
/ Samuel
2010/1/27 <steinbrt.rci.rutgers.edu>:
> Hi,
>
> I believe with such charge mutations there are two possible sources of error:
>
> a) Is the model applicable, i.e. is it ok to simulate a transition in
> which the total system charge changes with Amber's TI routines. In my
> personal opinion, we can answer that with a yes, even if problems may
> exist in special cases. Others on the list will most likely have their own
> ideas and caveats to add. We got good results for the free energy of
> solvation for charged compounds in many cases, but of course the question
> what the 'right' answer is here is an interesting one by itself.
>
> b) is the simulation good enough, i.e. does a typical MD simulation
> capture the full system response for an electrostatic change? That is much
> more system dependent and in case of proteins, far from certain (Apart
> from conformational changes, think about things like protonation state
> changes or bound ions). van der Waals changes and directed interactions
> are much more local in the force field picture than electrostatics, so a
> mutation changing such interactions may converge a lot quicker than one
> involving charges.
>
> Without further knowledge of your calculations, I cant say for sure, but I
> wouldnt be suprised if your bad results were caused by type b) problems
> more than type a)
>
>> We have tried to mutate a +2 ligand to +1 ligand and got much worse
>> results, compared to when we mutated a +2 to +2 ligand. We have not
>> gone further with this investigation, and we do not know if we can
>> trust the results. Let me know if you get good results.
>
> Kind Regards,
>
> Thomas
>
> Dr. Thomas Steinbrecher
> BioMaps Institute
> Rutgers University
> 610 Taylor Rd.
> Piscataway, NJ 08854
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jan 28 2010 - 03:30:02 PST
