Hi Tarak,

I think your simulation results are in relatively good agreement with the x-ray data. When making a comparative analysis, we should keep in mind several factors:

1) protein is more flexible during simulation than in the crystal because temperature may be higher, hidration may be better etc

2) the location of the highly-flexible regions is very important; most often, highly exposed loops are very flexible (and you may expect to see higher loop flexibility in MD simulations than in x-ray structure).

I published several papers including a similar type of analysis, but I would suggest you to have a look at two of them:

https://www.researchgate.net/publication/5989412_Substrate-induced_conformational_changes_and_dynamics_of_UDP-N-acetylgalactosaminepolypeptide_N-acetylgalactosaminyltransferase-2

https://www.researchgate.net/publication/47812191_Structural_models_of_TREK_channels_and_their_gating_mechanism

Good luck!

Article Substrate-induced Conformational Changes and Dynamics of UDP...

Article Structural models of TREK channels and their gating mechanism

Tarak, you really need to provide more details on your simulation setup. Is it all-atom/coarse-grained, explicit/implicit solvent?If you are using Langevin dynamics what is the friction coefficient?

Also the correct formula is: B=[(8*PI**2)/3] * (RMSF)**2

Finally the crystal structure is taken at lower temperature with different hydration than your simulations so you can expect fluctuations.

Hi Sam,

It is an all-atom simulation. I think I've written the same formula as that of yours.

I checked some papers where I saw this kind of plots.

Thanks for looking at the question.

Hi Tarak,

I think your simulation results are in relatively good agreement with the x-ray data. When making a comparative analysis, we should keep in mind several factors:

1) protein is more flexible during simulation than in the crystal because temperature may be higher, hidration may be better etc

2) the location of the highly-flexible regions is very important; most often, highly exposed loops are very flexible (and you may expect to see higher loop flexibility in MD simulations than in x-ray structure).

I published several papers including a similar type of analysis, but I would suggest you to have a look at two of them:

https://www.researchgate.net/publication/5989412_Substrate-induced_conformational_changes_and_dynamics_of_UDP-N-acetylgalactosaminepolypeptide_N-acetylgalactosaminyltransferase-2

https://www.researchgate.net/publication/47812191_Structural_models_of_TREK_channels_and_their_gating_mechanism

Good luck!

Article Substrate-induced Conformational Changes and Dynamics of UDP...

Article Structural models of TREK channels and their gating mechanism

B-factors are related to atom displacement parameters. Initially you can assign isotropic  parameters, and in small molecules crystal structures, for non-H atoms, we convert them to anisotropic parameters namely, Uii, Ujj, Ukk, Uij, Ujk, Uik, which are the parameters of thermal ellipsoids, Assuming 0.05 A2 isotropic parameter is ideal initially for non-Hydrogen atoms. You can convert to the Standard form, B-factors

Thanks Dr. Milac

I'll go through those papers.

Thank you Dr. Sarma.

Is there any equation relating Mean Square Fluctuations (MSF) and Temperature (T)?

Hi Tarak,

Kinetic energy is proportional to temperature, but I don't think there is a direct relation of MSF to T. That is because the fluctuations are dependent on the shape of the potenial energy surface. However, roughly speaking, I would expect B-factors to vary about linearly with T. 

Your B-factor formula is correct. What water model did you use in your simulations? How long did you equilibrate and how long was the production run?

Regarding your results I also think they do not look incorrect. In addition to Adina's helpful response, you may want to try a crystal state simulation. We have seen that crystal simulations are more accurate in reproducing the experimental quantities. (eg.https://www.researchgate.net/publication/236598220_Peptide_crystal_simulations_reveal_hidden_dynamics; https://www.researchgate.net/publication/266249543_All-atom_crystal_simulations_of_DNA_and_RNA_duplexes ).

Pawel

Article Peptide Crystal Simulations Reveal Hidden Dynamics

Article All-atom crystal simulations of DNA and RNA duplexes

Thanks Pawel for your comments and suggestions.

I used TIP3P water model, equilibration for 2 ns and then production run for 100 ns. I took the last 20 ns to analyse the MSF and thereby calculate B-factors.

Crystal simulation of my enzyme, I think is not a good choice. :(

I was wondering if we can scale the B-factors obtained from my simulations to the values in the crystal structure. Anyway, I think it's good to proceed with the current plots. :)

briefly I comment about fluctuation of B-factor in enzyme complexes, it is known that theses system are more flexible than their single counter part proteins. Recently, we we able to capture correctly the flexibily of enzymes within a CG model and realize that a shift in B-factor is common due to the temperature effect. Perhaps, you can try to study the effect of the temperature in your system, I mean the dependency of the B-factor vs Temperature. This in my case, enter in the plots as a shift and depending on the length of protein, this could be very dramatic. Another thing, you do not mention is the RMSD of you enzyme, how far from the crystallographics has gone: 2,4,... Armstrongs? Are you really in a structural equilibrated conformation? I leave some here these question, and perhaps a better understanding of the difference in flexibily and structural changes can help to clarify this issue. best 

Hi Dr. Poma,

Thanks for your answer.

We realised that B-factor depends linearly with T, from harmonic oscillator approximation. Accordingly, the values can be scaled roughly to the desired temperature, and it seems our simulation plots are in good agreement with the B-factors from the crystal structure.

In my simulation (100 ns), I see rmsd of the the backbone atoms within a range of 2-3 Ang., an indication of good structural integrity.

moreover I want to pursuit a clear understanding of this problem. Regions comprised between  ~ 90-100, 115-135 and 160-165 are located at flexible region such as loops? The beginning of your RMSF can be justify by assuming that in crystalline condition the N and C terminus are not moving due to low temperature setup. In water simulation, one expect this change in RMSF due to temperature or some missing structural part, how complete is your sequence? do you have all aa resolved experimentally and in addition what about the disorder part: how far are from each (in sequence) this can be crucial to understand why the large fluctuation at the central part in your RMSF, which seems to be close to one another and not seen, from my point of view, in the experimental data. best

https://www.researchgate.net/publication/309041540_Important_Factors_Influencing_Protein_Crystallization?ev=prf_high 

Article Important Factors Influencing Protein Crystallization