Hi,
I've noticed a common practice in publications where multiple Molecular Dynamics (MD) simulations are conducted with varying initial velocities. However, I'm uncertain about the appropriate method for analyzing the resulting data. Could someone please provide guidance on this matter?
To clarify, let's say I have files named md1.xtc, md2.xtc, and md3.xtc. I understand that I can perform individual analyses such as Root Mean Square Deviation (RMSD) and Root Mean Square Fluctuation (RMSF) for each simulation and display them on the same graph. Alternatively, I could calculate the average RMSD values along with standard deviation and represent them on the graph.
Moving forward, I intend to conduct MD simulations involving ligands. My plan is to derive an average structure from these simulations and apply docking. My question is: does it make a difference which simulation I choose to continue with for further analysis?
Moreover, I'm curious about the expected presentation of these analyses when publishing an MD simulation study. Could anyone share their experiences and insights on these matters?
Thank you in advance for your assistance.
I would recommend trying out different analysis techniques to figure out which one gives you the appropriate time and spatial resolution. For my part I often feel like raw RMS values at too coarse grained and should be accompanied by structural views of representative conformations. Really depends on the thing you like to get across though.
One thing to keep in mind is that a „calculated average“ structure is typically not something that you would encounter in nature. it might not even be a conformation that’s working with your force field. If you want to do docking with a calculated average structure I would find the frame that holds the structure that is closest to this calculated average. good luck :)
Martin Rosellen
Thank you for your helpfull comment.
Similar to the rationale expressed in the quote from the website, which emphasizes "the necessity of running multiple independent copies for robust data in MD simulations", I've conducted three simulations from the energy minimization (em) stage. If I understand correctly, this is the expectation for computational studies to be published.
To elaborate, I'm seeking to trying the procedures outlined for AMBER in the provided link, I could not yet adapting them to GROMACS due to its complexity in implementation. They answered this question "How can we extract a representative structure from our simulations data? ( like getting an average structure from 3 .xtcs)" but for AMBER. Is there you any advise me about apply these methodologies to GROMACS?
https://amberhub.chpc.utah.edu/clustering-a-protein-trajectory/
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Atomic, Molecular and Optical Physics
- Atoms