> 1.I played with the example of PKA-balanol system using the same pqr
> and input file downloaded from the website:
> I obtained the electrostatic binding energy at about 20 KJ/mol as
> print energy 3 (complex) - 2 (pka) - 1 (bal) end
> Local net energy (PE 0) = 1.996342352771E+01 kJ/mol
> Global net ELEC energy = 1.996342352771E+01 kJ/mol
> The result is larger than should-be result of 5.8 KJ/mol (shown in the
> webpage). What.s wrong with the data?
Well, this is an old tutorial that was made for a class (not for the
APBS user base) using examples from an older version of APBS. In
fact, the first page of the tutorial contains a disclaimer about it's
"draft" status (it includes several typos). With that in mind, the
major changes in this number are due to changes in the APBS surface
definitions and discretizations (e.g., the sdens keyword).
> 2. When I tried to calculate electrostatic salvation energy based on
> dG(solv, ele) = comp_solv - comp_ref - mol1_solv + mol1_ref -
> mol2_solv +
> mol2_ref. I used the same grid spacing and grid length as in the
> mentioned before with dG(solv, ele) = complex - mol1 - mol2. For the
> reference state, I just simply change sdie from 78.0 to 2.0 to
> insure the
> self-energy being removed. But APBS gives me the results like
> print energy 5 (complex) - 6 (complex_ref) end
> Local net energy (PE 0) = -1.272186142028E+04 kJ/mol
> Global net ELEC energy = -1.272186142028E+04 kJ/mol
> print energy 3 (pka) - 4 (pka_ref) end
> Local net energy (PE 0) = -1.282665662708E+04 kJ/mol
> Global net ELEC energy = -1.282665662708E+04 kJ/mol
> print energy 1 (bal) - 2 (bal_ref) end
> Local net energy (PE 0) = -3.262888645731E+02 kJ/mol
> Global net ELEC energy = -3.262888645731E+02 kJ/mol
> print energy 5 (complex) - 6 (complex_ref) - 3 (pka) + 4 (pka_ref) - 1
> (bal) + 2 (bal_ref) end
> Local net energy (PE 0) = 4.310840713807E+02 kJ/mol
> Global net ELEC energy = 4.310840713807E+02 kJ/mol
> Why is it so different with 19 KJ/mol obtained before? May I
These solvation energies aren't converged with respect to grid
spacing (this issue was recently discussed on this mailing list).
You should find a significant change in solvation energy for the
protein and protein-ligand complex as the grid spacing is decreased.
In general, solvation energies for large molecules are hard to
converge -- the approach outlined above (your question #1) masks
these problems by canceling many of the artifacts associated with
grid spacing, position, etc.
> 3.When we calculate the solvation energy with self-energy being
> removed, the solvated state should be like pdie = 2.0, sdie =78.0
> and ion
> = 0.1 M; while in the reference state, pdie = 2.0 = sdie, ion = 0.0
> M. Am
> I correct?
Yes, that's usually the most sensible approach.
> If so, in the examples of FKBP/1d7h-dmso, why the ionic
> strength in the reference state is nonzero?
For historical reasons: this example was initially used to compare
APBS against UHBD results. The person setting up the UHBD example
chose to have non-zero ionic strength for his reference state for
reasons which are unclear to me. However, since we use the APBS
examples as our test suite to validate new releases, I've been
hesitant to change any of the examples without very good reason...
this confusion might be a good enough reason, though. :)
I'm working on new additions to the official APBS tutorial (http://
apbs.sourceforge.net/doc/tutorial/index.html) which will hopefully
address some of these questions for future users.
Associate Professor, Dept. of Biochemistry and Molecular Biophysics
Center for Computational Biology, Washington University in St. Louis