Protein rotating in triclinic box


I have simulated a protein inside a triclinic box since it was computationally less expensive. During MD I see the protein coming out of the box and rotating. What’s more important is that I also see that its periodic image at a distance of 17 angstrom in the beginning of MD but the distance between these two images becoming ~10 angstrom at the end of the simulation. Is this usual or the choice of the triclinic box is not correct?
Is the triclinic box so short in the z axis that the protein is interacting with its periodic image? If it is so, do you think this simulation will have artifacts? NB: I used -c -d 1.0 -bt triclinic option during gmx editconf.

Any suggestion will be appreciated.

Thank you in advance,
Soumadwip Ghosh

1 Like


Your system will most definitely have artifacts if your protein interacts with itself; I would suggest making sure that your keep a minimum distance of 1.2 nm as this is common setting for the vdW cut-off for common force-fields.

It’s useful if you check the distance to the next perodic image of the protein using gmx mindist -pi. The only option then is to run the simulation again and increase the -d parameter in editconf.

Regarding your setup:

If you use only -bt triclinic without any further input options, you will just get a rectangular box, because editconf will still use the default parameters for the box angles. Any box is triclinic, it’s the most general box-category which includes rectangular, octahedral, etc. boxes.

For all cases where a protein freely moves in water, you would like to use a box that is as spherical as possible, which is achieved by using the rhombic dodecahedron with -bt dodecahedron .

Find more information on how to set up the simulation boxes here:

The change in the minimum distance may be due to conformational change taking place in the protein during the simulation (e.i proteins can go from a more compact form to a more extended form, and that causes a reduction of the minimum distance). Did you try to visualise the protein before and after the simulation? Did you use GROMACS tools to calculate the minimum distance?

I do not see any issues with the options you use for gmx editconf.

Best regards