[28-I-2013]
Hi Thomas,
Thank you very much for your help. Actually I was aware of the meaningless -in a physical sense- of the intermediate steps in a TI / FEP calculation. However, the unfavorable difference in energy (for the expected solvating stabilization) was so strange that I thought that maybe something was wrong. Also, I didn't expected that the unbalanced internal energies during the process might somehow compensate the solvating term. I will run the calculation again activating the DVDL energy decompostion to check the breakdown.
Anyway, I still have a couple of doubts:
1) According to the amber manual, the "idecomp" options to switch on the DVDL breakdown are "=1" and "=2". The first option adds the 1-4 non bonded energies to the internal energies, and the second option to the EEL and VDW. In my particular case, it seems that the best option would be selecting "idecomp=2" (not sure anyway). Actually, I'm wondering in what situations is better so select option 1 or 2. Could you please give any advice/reference about this?
2) At the beginning I was running my tests in two different ways: In the first one, I switched off/on ALL the charges in the glucose molecule, and in the second one (like the example given in the tutorial), I only switched off/on the charges on the specific mutated atoms. This meaning, in the second case, that removing the charges only in certain atoms produced a formal charge in the residue creating an ion during the intermediate steps of the TI transformation. I have read that ions in solvent MD calculations under periodic boundary conditions may lead to some artifacts when lattice-sum methods are employed (J. Phys. Chem. B 2004, 108, 774-788). In my case the situation is even worse because I am creating an unbalanced net formal charge in the unit cell, therefore the artifacts in the energy and also the conformational exploration can be really non-negligible. So, do you think it is better to switch off/on all the charges in the molecule as I do rather
than only the charges in the mutated atoms as in the tutorial?
Any help will be very welcome.
Kepa K.
**************************************************
________________________________
De: "steinbrt.rci.rutgers.edu" <steinbrt.rci.rutgers.edu>
Para: Kepa K. Burusco <kekoburgo.yahoo.es>; AMBER Mailing List <amber.ambermd.org>
Enviado: Viernes 25 de enero de 2013 16:30
Asunto: Re: [AMBER] TI alpha-beta glucose Glycam06h
Hi Kepa,
what you see is not necessary a problem. Bear in mind that TI
transformation results are not physically meaningful without embedding
them into a thermodynamic cycle, especially individual simulation steps
with non-existent end states.
Furthermore, in this case, note that you are changing forcefield
intramolecular energies, the absolute values of which are completely
non-physical. The 1-4-EEL interaction are particularly important here and
easily amount to 100 kcal/mol or so. Look at the DVDL breakdown to find
out which terms contribute most to your dvdl.
For the system you study, would you expect MD free energy calculation to
produce a good result? The different glucose isomers should differ in ring
strain and 1-4 interactions, terms that might need to be determined
quantum mechanically.
Kind Regards,
Thomas
On Fri, January 25, 2013 8:42 am, Kepa K. Burusco wrote:
> [25-I-2013]
>
> Hi everybody,
>
> I'm a newcomer to Thermodynamic Integration calculations and I have been
> doing some tests running a TI to transform the alpha anomer of glucose
> into the beta anomer to determine the Difference in Free Energy. I am
> using Amber11 with AmberTools12 and Glycam06h force field. I am
> following the tutorial A9 (http://ambermd.org/tutorials/):
>
> 1) First of all I created the input files for the alpha and beta anomers
> and solvated them with TIP3P (glaWAT glbWAT) taking care of all the
> requierments needed to avoid problems with common/special atoms.
> 2) Then I divided TI process into 3 Legs: switching off all the charges in
> the glucose, mutating atoms (employing soft-core potentials), and then
> again switching on the charges (files *.in).
> 3) I extracted the lines "DV/DL average" and "DV/DL fluctuations" from
> output files to create the 3 columns (Lambda, Average, Fluctuation) text
> file as recommended by the tutorial ("stepX.TIresults.txt").
> 4) Finally, I processed the lines with the perl script in tutorial
> ("TotalTI.stepA.txt").
>
> After checking the results I have discovered something strange, and I
> would like to ask you if I have been anything wrong.
>
> In the first leg, I switch off the charges of the glucose molecule. Since
> it is a polar one, interaction with water should be favorable and
> therefore going from the molecule with charges (polar) to the molecule
> without charges (apolar) should give a positive difference in energy.
> Instead, the difference in free energy is about -91.8 kcal/mol (???).
> We find exactly the same strange result in the third leg when I apply the
> charges. Going from the apolar molecule in water to the polar molecule
> should be favoured, however, I again find the wrong sign in the energy
> (about +92.5).
>
> I send you attached a tar file with the text documents (submitting scrips,
> input files, structures, results...)
> Does anyone have any clue about what I am doing wrong?
>
>
> Than you very much.
>
> Kepa K.
> ***************************************************************_______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
Dr. Thomas Steinbrecher
formerly at the
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 Mon Jan 28 2013 - 04:00:02 PST
