search

chrys-athome / sire

Automatically exported from code.google.com/p/sire
0 stars 1 forks source link

Inconsistent evaluation of coulombic energies between vacuum and solvated systems. #10

Open GoogleCodeExporter opened 8 years ago

GoogleCodeExporter commented 8 years ago 
        I think there is a bug in the SireOpenMM free energy single topology code. The problem is that 1,5 or greater pair-pair interactions in the same solute are computed using a reaction field but only if using a periodic cutoff. As a result the intramolecular coulombic of a completely rigid solute differs between a run in vacuum that uses no cutoff (and no RF) and one in a water box (that uses RF). This creates spurious contributions from intramolecular terms that do not cancel out in thermodynamic cycles as the energy function between vacuum and periodic systems differ.  
        This would explain why the TI hydration free energies results I got so far are in good agreement with the literature for very small solutes, but deviate substantially for solutes with intramolecular 1,5 or more pair interactions.

 Gaetano probably didn't spotted this effect because the solutes he has done his tests on in a water box did not have >1,4 interactions.   

   A possible solution could be to use a PeriodicCutoff for vacuum simulations (and perhaps adjust the value of the dielectric constant)

   Another one could be to not apply RF to 1,5 or more pair-pair interactions.

Original issue reported on code.google.com by julienm...@googlemail.com on 17 Jul 2013 at 3:40

GoogleCodeExporter commented 8 years ago 
Gaetano indicated the problem is  solved by using the keyword cutoffnonperiodic 
for the setup of vacuum simulations. So the issue can be closed.

Original comment by julienm...@googlemail.com on 18 Jul 2013 at 10:02