[gmx-users] Inconsistency in gmx sasa/g_sas buried SASA?

[Leandro Bortot]

Dear users,

     I'm trying to calculate how much of a given ligand surface area is
buried upon binding to a protein.
     I did it before for two proteins by doing three separate calls of gmx
sasa, each for the surface and output groups "System" and "System" of files
containing protein A, protein B and the AB complex. Then, I used to
calculate the buried area by doing "sasa_A + sasa_B - sasa_AB". However, I
found some strange values with a protein-ligand system. Can you please
point me to what I'm doing wrong?


     If I follow the previously stated protocol with pdb files of the
complex, isolated ligand and isolated receptor, I end with the following
values for the SASAs: 85.8060, 8.288 and 86.4280 nm^2, respectively. By
doing the "A+B-AB" calculation, I get a buried area of 8.9100 nm^2, which
is actually more than the whole isolated ligand.
     On the other hand, by using the -surface and -output groups "System"
and "Other", I have the value of 2.970 nm^2 of exposed area for the ligand
in the complex, which corresponds to a buried area of 5.3180 nm^2, 64% of
the ligand, which seems reasonable according to visual inspection.

     Why are the values calculated by the two approaches not consistent?
     My first thought was that gmx sasa was not "seeing" the whole ligand
in the complex structure, hence the very small increment in the complex
SASA in comparison with the isolated receptor. However, the results for
using different -surface and -output groups doesn't agree with that.
     I have the same results with GROMACS 4.6.7 with slightly different
numerical values, which I assume is due to the changes in the values of
vdwradii.dat. I get > 100% are buried by doing the "A+B-AB" calculation and
around 60% when I use "System" and "Other" in g_sas.

     I tried using .gro files created from the .pdb by editconf, but
nothing changed. I cannot afford to generate the tprs because I'm trying to
use GROMACS to analyse a large number of complexes which came from a
virtual screening essay.
     The .pdb files are typical, there is nothing strange in them. At least
as much as I can tell.


     Here are the commands I used, for clarity.

         gmx sasa -f complex.pdb -s complex.pdb -nopbc -o sasa_complex.xvg
-surface "System" -output "System" -xvg none -probe 0.14 -ndots 24
         gmx sasa -f ligand.pdb -s ligand.pdb -nopbc -o sasa_ligand.xvg
-surface "System" -output "System" -xvg none -probe 0.14 -ndots 24
         gmx sasa -f receptor.pdb -s receptor.pdb -nopbc -o
sasa_receptor.xvg -surface "System" -output "System" -xvg none -probe 0.14
-ndots 24
         gmx sasa -f complex.pdb -s complex.pdb -nopbc -o
sasa_difference.xvg -surface "System" -output "Other" -xvg none -probe 0.14
-ndots 24


     Any help is greatly appreciated.

Thank you for your attention,
Leandro.