Minimum distance and trajectory convertation

User discussions
1

GROMACS version: 2023.3
GROMACS modification: No
The gmx mindist command is used to calculate the distance between the protein and its periodic image to see if the condition of distance >1.2 (as the cutoff for non-bonded interaction) is met.
Minimum distances calculations were applied for the same trajectory but in two modifications:

  1. processed by:
    gmx trjconv -pbc mol -center -ur compact (centering is based on the residues that are close to geometric center of protein but not exactly in the center)
  2. sequentially processed by:
    gmx trjconv -pbc mol -center -ur compact
    gmx trjconv -fit rot+trans.
    For the second case (fitted trajectory), the minimum distances were usually higher than in the first case.
    Which option displays the correct distances and should be used when checking for the “minimum image convention” effect?
    And if in the case (1) min distance is < 1.2 and in case (2) min distance > 1.2, is it correct to use fitted trajectory for analysis?
2

I may be wrong, but I’m quite certain that fitting the trajectory (with rotation) will not change the box vectors, so the distance to the periodic image will not be the same as when you performed the simulation. I would avoid the second alternative for these kinds of analyses.

3

Thank you very much for your reply. But I’m still struggling to fully understand the meaning.
You indicated that fitting would not change the box vectors. Then why the distances “will not be the same” as during simulation.
From my observations, the calculations of several values ​​for fitted and non-fitted trajectories (Rg, SASA etc) remain unchanged. However, min distances were higher for fitted trajectory then for non-fitted one (just even when I performed centering of 1 chain molecule by group 1 Protein).

If analysis of fitted trajectory should be avoided, what may be the reason for this?
And if I get a case when
for non-fitted trajectory min distance is <1.2
and for fitted trajectory min distance > 1.2
Should I re-run simulation with larger box or fit and perform further analysis?

4

Consider having your protein in a cube (as a simplification, but it will be the same for any box shape). If you rotate the protein in the cube, the distance to periodic images will change, depending on the shape of the protein relative to the periodic box.

Simplified 2D:

[         ]
[         ]
[    |    ]
[    |    ]
[    |    ]
[    |    ]
[         ]
[         ]

rotated to (something similar to):

[         ]
[         ]
[  \      ]
[   \     ]
[    \    ]
[     \   ]
[         ]
[         ]

or:

[         ]
[         ]
[         ]
[   ----  ]
[         ]
[         ]
[         ]
[         ]

The minimum distances to periodic images may increase or decrease upon fitting with rotation, depending on the initial configuration and shape of the molecule and the shape of the periodic box.

I don’t think any other observables, in the same trajectory, will be affected by the rotation itself.

I think you will have to ask yourself the question if you think the images of the protein affect each other enough to warrant re-running the simulation. If you publish the results make sure that you clearly state that you observed this after the simulation.

5

Are there any recommendations on achieving distances between periodic images (in order to talk about significant/insignificant influence on the simulation)?
I can say that often when creating a dodecahedron box with distance to box edge 1.0 or 1.2 or 1.4 nm (-d option in gmx editconf), some frames during simulation would have min distance lower than 1.2 nm (that is a cut-off for non-bonded interactions in my case).
Should the simulation be carried out to achieve “no frame with lower than cut-off 1.2 distance” or to “achieve the cut-off predominantly” or “some percentage” or maybe there are other accepted standards in the field?