[gmx-users] COM-COM distance: PBC problem

[Arnost Mladek]

Dear Plumed users,

I've encountered one problem I'm unable to resolve. I use combination of Gromacs 5.0.4 and Plumed 2.1.3 and what I want is to calculate restrained distance between two center of masses (COMs) corresponding to two chains within a dimeric protein. If I let Plumed to calculate COM-COM distances "on-the-fly" (ie. during the simulation), I get correct results, let say distances in the interval of (4,9 ; 5.3) with the average value of 5.1 nm corresponding to the minimum of the restraining potential. 

If I use Plumed as a stand-alone software and calculate the distances along the trajectory via "plumed driver --mf_xtc trajectory.xtc --plumed pl.dat", I get the same correct results. Up to this point, no problem.

However if I fit (rot+trans) the trajectory frames to chain-A using GROMACS trjconv command (trjconv -f trajectory.xtc -s reference.tpr -o fitted_trajectory.xtc -fit rot+trans -center) and then run the plumed driver to calculate the distances from the fitted trajectory there are huge jumps of several nm. It is obvious that the problem is somehow associated with PBC and "WHOLEMOLECULES STRIDE=1 ENTITY0=protein" command within the plumed script is unable to resolve it. I've tried various trajectory transformations as suggested on GROMACS web page (http://www.gromacs.org/Documentation/Terminology/Periodic_Boundary_Conditions), like "-pbc mol -ur compact", "-pbc whole", "-pbc nojumps", etc. but nothing helped. As I am unable to get the COM-COM distances using GROMACS alone, I am not sure whether the problem relates to GROMACS or Plumed.

Thank you so much for your help.

AM