PBC/trajectory reconstruction problem for discontinuous multi-chain RNA fragment

User discussions
1

Hi,

I have a small RNA construct excised from a much larger RNA/protein complex. For confidentiality reasons I cannot share the exact biological identity of the construct, but the relevant technical details are:

  • The simulated solute is RNA only.
  • It consists of several disconnected RNA fragments/chains (because the excised construct was based on distance from a particular atom), not one covalently continuous RNA molecule.
  • The fragments are spatially close in the starting structure and together form the local receptor/pocket of interest.
  • I prepared the system using CHARMM36/TIP3P.
  • GROMACS version used for the run: 2023.4.
  • The production run was a short 10 ns pilot.
  • The output trajectory was later exported to a multi-model PDB ensemble for inspection.

The issue: When I visualize the exported ensemble, each frame does not appear as one assembled RNA construct. Instead, the RNA fragments appear scattered across the periodic box/in different images. At first glance it looked like multiple copies of the construct, but checking the PDB showed that each model has the expected atom count, so it is not actually four physical copies per frame. It seems more likely that the disconnected RNA fragments are being wrapped independently, or possibly that the fragments are genuinely drifting apart during the simulation. This is different from the standard case because my solute is not one continuous molecule. It is a multi-fragment RNA construct.

My questions are:

  1. What is the recommended GROMACS workflow for reconstructing/centering/fitting a trajectory when the solute is composed of several disconnected RNA fragments that should be kept together as one receptor construct?

  2. Is `gmx trjconv -pbc cluster` the right approach here, using a custom index group that contains all RNA fragments?

  3. Should I first apply something like `-pbc whole`, then `-pbc nojump`, then `-pbc cluster` or `-center`, or is there a better order of operations?

  4. How can I distinguish a pure PBC-imaging artifact from real physical dissociation of the RNA fragments?

  5. More generally, is an unrestrained apo MD simulation of a discontinuous RNA pocket excised from a larger complex likely to be unstable unless I include restraints or a larger surrounding scaffold?