I conducted a 100ns simulation of my docked complex, obtaining a maximum RMSD of 4.2 Angstroms and an average of 3.1 Angstroms. Can someone help me how can I justify this high RMSD? I have attached a snap of the graph below.
As far as I have interpreted this graph. Initially, this is a gradual increase in RMSD but then it became stable with an average RMSD of 3.1. So basically this RMSD is high but is stable. But Why its value is high? Is it ok to have a high RMSD?
Aliza Naz Check other parameters also i.e., Radius of gyration, RMSF, as well as H-bonds/distance based calculations to know the actual cause of unstability/stability.
Radius of Gyration is parameter of compactness of the protein structure,
RMSF depicts which region/C-alpha/residues are contributing in major fluctuations,
H-bond/distance based calculation will clarify the cotribution of from ligand in stablization/destablization.
Harvinder Singh Thank you for your prompt response. I have checked these parameters as you said.
The radius of Gyration (Rg): mean is 41.06, maximum is 52.02 reflecting the higher-ordered and compact nature of the system.
RMSF of the complex: Mean is 1.71 Å with max of 6.27 Å at 285 which is a loop region.
β-factor: Average is 99.84 Å2 with max of 1034.27 Å2.
Are these values fine overall? Is the system stable or not?
Aliza Naz yes now you can state that the fluctuations belongs to loop region and other wise everything is fine/stable as compared to apo form (protein only) simulation.
Dear Aliza,
Interpreting RMSD (Root Mean Square Deviation) values from molecular dynamics simulations requires careful consideration and context. RMSD is a measure of the deviation of atomic positions between two structures, usually a reference (initial) structure and the simulated structure. A high RMSD doesn't necessarily imply a problem, but it does require investigation and justification.
You can check first, stability and convergence analysis, as RMSD (Root Mean Square Deviation) is crucial for assessing the reliability and quality of your molecular dynamics simulation results. This analysis helps you ensure that your simulation has equilibrated and reached a consistent, meaningful state. As you've correctly noted, a gradual increase in RMSD followed by stabilization suggests that your system has reached a relatively stable state during the simulation. This stabilization indicates that the complex has undergone equilibration and may have achieved a conformational equilibrium. RMSD analysis can highlight conformational changes in your system. Significant changes in RMSD could indicate structural rearrangements, transitions between different states, or unphysical behavior. Stability analysis allows you to compare different simulations or replicate runs to determine if their behaviors converge. Consistent RMSD patterns among different runs provide confidence in the reliability of the simulation outcomes.
Secondly, the docking quality, The high RMSD could be due to the initial positions of ligands and proteins in your docked complex. Docking tools, while powerful, may not always generate perfect predictions of binding orientations. A high RMSD could indicate that the initial docked pose had a significant deviation from the actual bound state. Consider analyzing the binding interactions and potential hydrogen bonds to ensure that the binding is reasonable.
The third reason may be the system's complexity. High RMSD might be expected in systems with flexible or large components, complex protein-ligand interactions, or significant conformational changes. If your system involves protein flexibility, loop movements, or side-chain rearrangements, these factors can contribute to higher RMSD values.
The stability of RMSD over a 100 ns simulation is a positive sign. However, longer simulations might be needed to capture more complex conformational changes or to ensure convergence. Short simulations might not fully explore the conformational landscape of the protein-ligand complex.
Another suggestion is that you may go through a deep-level trajectory analysis. Analyzing the simulation trajectory can provide insights into specific structural changes causing the RMSD fluctuations. Consider focusing on the portions of the trajectory where the RMSD is high to understand what conformational changes are occurring.
The next suggestion is if possible it's helpful to compare your results with experimental data, if available. Literature values or known crystal structures of similar complexes can provide context for the expected range of RMSD values.
Lastly, I can say that please evaluate whether the observed RMSD affects the biological relevance of your simulation. If the RMSD is relatively high but the binding mode and key interactions are preserved, the simulation might still provide valuable insights into the system.
In summary, a high but stable RMSD can be acceptable depending on the complexity of the system and the goals of the simulation. A thorough analysis of the trajectory, understanding the system's characteristics, and comparing it with experimental data can help you justify and interpret your results accurately.
Hope this answer will be helpful for you to justify your results.
I wrote an shorter answer before that I managed to accidentally delete again xD. The long answer by Upasana Hazarika covers most of the things I wanted to say though. Regarding longer simulations time if you can you should aim for 1000ns that seems to be kind of expected nowadays. Depending on where in the world you are you will have free access to super computers for research purposes you simply have to find out how to apply for computer power. And then learn how to do simulations on a cluster (but its absolutely doable).
Regarding stabiliy, although I have only seen this in like one paper so far, you can make a 2D histogram of RMSD vs radius of gyration, this can help you show if the system reached a stable state even more.
A short remark about your "high" RMSD: Your RMSD curve is below 4 Angstrom the entire time. There are models where we consider a cutoff distance of 4A up to even 8A as the distance between residues where those will be in contact. In this context I don't know if it is even fair to say, that the absolute value of your RMSD is highg. Besides that what Upasana Hazarika said ^^.
Also, if you have complexes, the RMSD might undergo stark deviations or be at the very least extremely rugged while the interface of the two molecules stay in place. There are other metrics for such scenarios like the interface RMSD (iRMSD) and iRG accordingly, or the percentage of native / initial contacts that your complex shows.
Benjamin E. Mayer Thank you for your valuable insights. Yes, you are right I am also thinking of extending simulations to get betters insights of the system. For now, due to time constraints, I have to submit the report so I am ending it here. But for the paper, I will definitely extend simulations.
Benjamin E. Mayer Upasana Hazarika Harvinder Singh Could anyone please recommend publicly available cloud supercomputers where I can run simulations for free? I lack the necessary resources to extend my simulations promptly. Otherwise, I have to wait for my turn to extend my simulations. Your suggestions would be greatly appreciated. Thank you!
Benjamin E. Mayer Thanks for adding more clearer text to support her.
Could you please refer to the paper title which "make a 2D histogram of RMSD vs radius of gyration"?
Upasana Hazarika I have to admit I can't find it anymore. There are a few papers where they use the (RMSD, RG) landscape as reduced coordinates for visualizing the free energy on it as color information, e.g., here Article Understanding Conformational Diversity of Heat Shock Protein...
. In the following article https://doi.org/10.1038/s41598-017-15930-4 they do something like the histogram, by using this to show the distribution of a certain cluster. If I see the paper again I was thinking about I will post it here. Indeed I haven't seen that many papers solely about the convergence of an MD simulation. Although this really recent talk from nvidia where they do drug design with MDs seems to suggest that multiple short MDs are superior to one large MD. This is an argument I've heard in the past in the context of replica exchange simulations, and that is certainly true but probably needs a lot of computer power xD
Benjamin E. Mayer yeah what you told me in the last part, I also read a similar discussion like that. And many people suggest doing multiple short MDs rather than one long MD for better interpretation .
Thanks for sharing the paper. I will have a happy read.
- Badges
- Science topic