We wish to perform a molecular dynamics simulation of cell division protein from a hyperthermophile species at 37 deg cel to check for its stability. Is there a specific analysis in GROMACS for determining the protein stability following the simulation.
Think about the time scale of protein folding. If you want to say with any reasonable degree of confidence that your protein is "stable," e.g. the interactions persist such that it doesn't unfold, you need a lot of sampling. If you run a 20-ns MD simulation, a reviewer will laugh in your face because that's a couple orders of magnitude shorter than any known folding/unfolding process. 100 ns is sort of a ballpark (again, multiple trajectories so your total simulation time is larger!) but generally a good estimate in a situation like this. I have done simulations that were "stable" out to 100 ns but that changed and manifested much more interesting behavior after 400 ns. It all depends on what a relevant time is for the behavior at hand.
Most importantly, you need to demonstrate that your data set is adequately converged (not still systematically changing with time).
RMSD/RMSF are the two first observables one uses to check the flexibility/stability of the protein. But you need to run some 100 ns first, atleast three trajectories
Thank you but why 100 ns is required if the RMSD shows very little variations at 10 ns?
It's not necessary to perform molecular dynamics simulations for 100ns unless you are providing a structural insight into the activation/inactivation mechanism of Protein-Ligand complex.
For the evaluation of the protein stability, you can check protein RMSD and RMSF plot.
Getting lesser variation in the plot is ok but its range should be considered, like RMSD variation below 3 angstroms is acceptable for protein stability.
The theoretical prediction of protein stability is still an open question. It may be accepted by peer reviewer if you are performing MD from native structure by using GROMACS. In principal, I don't believe the result from MD simulation until you could fold it from sequence by MD.
Hi Shiyong could elaborate on the point and explain it in more detail what you mean by folding it from sequence. The structure we have used has been modeled using Phyre2 and subjected to energy minimization and then used for MD. Is that what you meant?
The reliability of your model depends on the Phyre2. Of course, Phyre2 is a good software. However, the accuracy of output depends on your target sequence. If no homologous are found for your target sequence, the final model need to be verified by additional evidences. What I meaning folding it from sequence is that the total protocol is MD without any knowledge-based or template-based modeling software. As far as I know, only few cases are folded in that way.
There is no magic recipe to demonstrate stability, but you can piece it together by performing several analyses. Neither RMSF nor RMSD definitively show stability. RMSF shows that the protein will fluctuate about a mean structure; changes in RMSF due to mutation, ligand, condition, etc. can be interesting but in your case I doubt it is useful. A stable RMSD trace is sort of a first indicator, but RMSD is a degenerate metric and gives you no meaningful insight beyond "the protein didn't totally fall apart." Does that mean it's stable? Not necessarily. But if you analyze contacts, hydrogen bonding, secondary structure evolution, salt bridges, etc. you can put together an argument about the atomistic details that govern secondary and tertiary interactions. This is the real value in a simulation.
You will need good sampling to convince a peer reviewer (and yourself!) of these outcomes. Multiple simulations of 100 ns or more are becoming standard. A single trajectory of that length is not necessarily representative of reality and may not even be converged.
thank you Justin, but this again brings us to the question i asked earlier.. is 100 ns necessary? Just because we can do should we do or is there a reason?
Think about the time scale of protein folding. If you want to say with any reasonable degree of confidence that your protein is "stable," e.g. the interactions persist such that it doesn't unfold, you need a lot of sampling. If you run a 20-ns MD simulation, a reviewer will laugh in your face because that's a couple orders of magnitude shorter than any known folding/unfolding process. 100 ns is sort of a ballpark (again, multiple trajectories so your total simulation time is larger!) but generally a good estimate in a situation like this. I have done simulations that were "stable" out to 100 ns but that changed and manifested much more interesting behavior after 400 ns. It all depends on what a relevant time is for the behavior at hand.
Most importantly, you need to demonstrate that your data set is adequately converged (not still systematically changing with time).
Well explained by J. Lemkul but I would like to add regarding the last line.
By the end of the simulations (time duration that you have decided), there should be no prominent changes. There are scores of papers where molecular dynamics simulations of 5 to 10ns is performed (in silico drug design) and they check for the stability of the system by monitoring the variation in the RMSD plot. Moreover, if your plot is not converged, it indicates that you need to increase the time duration of simulations until it showed no prominent deviation.
I think that is not too big, therefore you should simulate as long as possible, recent simulation time of published works has been extending more than 1 microsec. good luck :)
Is there is any significance for the absence of disulfide linkages and heatsensitive typtophan residues in protein?
Is this property of the protein helps in increasing the stability of the protein in industrial processing?
thanks Justin, such a useful explanation!
how about also calculation of radius gyration? specifically in case of mutated-protein in comparison with none-mutated protein?
Fahimeh Maghsoodi Rg is only useful for systems that involve differences in compaction. For well-folded (globular) proteins, this quantity is generally useless as it does not vary. Rg is useful for disordered proteins, tracking protein folding, etc.
- Badges
- Science topic