GROMACS version: 2018.4
GROMACS modification: No
Hi there,
I am trying to calculate the polypeptide entropy of a solvated protein system. I am using the gmx covar tool to obtain the covariance matrix first, and then the gmx anaeig to obtain the entropies (the quasi harmonic and Schlitter ones).
My system is made up of a protein of 1705 atoms solvated in a big truncated dodecahedron box (~63,000 water molecules). I am simulating the system in NVT using the velocity verlet algorithm over 2ns, 1fs timestep (printing frequency 10, 20, 50 timesteps). This printing frequencies are more than enough to overcome the minimum number of eigenvalues required to proceed accurately with the calculation (according to the theory): the number of atoms x 3 DoF.
The point is that the calculated values reported for the QH and the Schlitter methods (quite similar between them for each setup) change when the printing frequency is changed, something that I was not expecting at all. ¿Is it normal? From what I read in the literature about these methods, there is no term indicating this variable can be affecting the calculations if I am right.
Hope I can have some help from some of you.
Best.
JuanJo