normal How to assess the quality of the reverse transformation?

  • z.sichaib
  • z.sichaib's Avatar Topic Author
  • Offline
  • Senior Boarder
More
5 years 5 months ago #7861 by z.sichaib
Dear all,

I run CG simulations using Martini force field (martini 22) on a transmembrane protein. The protein was embedded in a membrane containing POPC, POPE and CHOL molecules, solvated with standard water model and ions were added to neutralize the system with a concentration of 0.15mol/L.

Before the production run, the system was minimized and equilibrated (NVT equilibration followed by NPT). The system has been simulated for ~ 4μs.

In order to do a deep analysis of my system, I.e. interactions between the protein and CHOL molecules I backmapped a part of the trajectory (not all the frames and only Protein, Water, CHOL, and Ions).

I noticed local unfolding of helical structures in some of the TM regions.

Now I'm having questions regarding the reverse transformation: How can I be sure that the loss of secondary structures that I'm observing is something that really happened during the simulation and not something that I lost during the backmapping?

To investigate this I:
1- Took the initial PDB structure,
2- run DSSP analysis,
3- then coarse-grained it using Martini22 FF,
4- I converted the output from PDB to gro, using gmx editconf (with the options -c -d 1.0 -bt cubic).
5- I backmapped the resulted CG structure to AA using initram (to amber FF).
6- Run DSSP analysis on the new AA structure and compared it with the original one.

To my great surprise, the DSSP results were not the same: I.e. one residue was in an alpha Helix in the PDB structure and now it is in a 3-Helix in the backmapped structure. Another one was in an alpha Helix and after the backmapping, it is in a turn. Another one was in a bend in the original structure but then it's in a coil...etc.

My questions are:
1- How can I assess the quality of the backmapping?
2- How can I be sure that what I'm observing after the backmapping is something that happened during the simulation?
3- I did not backmapped POPC and POPE molecules but only protein, Water, CHOL, and ions. I don't think that It can affect the final confirmation of the system. What do you think about this?

Thank you in advance for your help.

Please Log in or Create an account to join the conversation.

More
5 years 5 months ago #7863 by peterkroon
Martini proteins can not and will not change secondary structure because there are different parameters for the SS segments.
So what you're seeing must therefor be an artefact of backwards. Not having the membrane there can influence the structure you get out.

Please Log in or Create an account to join the conversation.

  • z.sichaib
  • z.sichaib's Avatar Topic Author
  • Offline
  • Senior Boarder
More
5 years 5 months ago #7864 by z.sichaib
Dear Peter,

Thank you for your reply.

I checked Martini literature to better understand your comments and now I get it! I don't know how I missed this.

Also, I checked the backward script; 'initram.sh script for version 5' that is available in Martini Website and I noticed something contradictory with the Temperature value:
The program is supposed to run a series of Position restrained NVT simulation at 300K, with different time steps but when I check the step 4 in the script for the used mdp parameters, this is what is written:
ref_t = 200
gen_temp = 300

ref_t is supposed to be equal to 300... Or maybe I misunderstood something?

Please Log in or Create an account to join the conversation.

More
5 years 5 months ago #7865 by riccardo
Yes, you are right. I checked all the scripts which can be downloaded from:

www.cgmartini.nl/index.php/back

and indeed
4. Position restrained NVT simulation (300K), with time step 0.2 fs
5. Position restrained NVT simulation (300K), with time step 0.5 fs
6. Position restrained NVT simulation (300K), with time step 1.0 fs
7. Position restrained NVT simulation (300K), with time step 2.0 fs

while:
ref_t = 200

then sets the temp to 200K. Good catch!

I'd say it does not make a huge difference but it's always good to be consistent. Want to make a pull request on Github ( github.com/ricalessandri/Backward/blob/master/initram.sh )? Otherwise I suppose I will do it / tell directly to Tsjerk.

Please Log in or Create an account to join the conversation.

  • z.sichaib
  • z.sichaib's Avatar Topic Author
  • Offline
  • Senior Boarder
More
5 years 5 months ago #7867 by z.sichaib
I was going to make a pull request on Github but since you can tell it directly to Tsjerk, that would be better/fast I think :)

I have another question, please. Concerning my system, before running the CG simulation, I run AA one and we choose a Temperature of 315K.

When I'll run the backmapping, I should choose the temperature based on the physical properties of the lipid, most notably the phase transition temperature. So in my case, 315K and not 300K? Or is 300K the standard value that we must use, as you specified in the backward protocol?

Maybe it will not make a big difference but in the end, I would like to be able to compare the results of both simulations and also for consistency.

Please Log in or Create an account to join the conversation.

More
5 years 5 months ago #7868 by tsjerk
Hey :)

1. I checked the temperature thingy and it is a bit odd. The reason for using 200K in the PR-NVT is that it equilibrates more easily, while a start temperature of 300 gives a bit more of a kick. Of course, with position restraints it doesn't matter much anyhow. However, the help should be a proper reflection of what happens. Note that this pertains to the relaxation after backmapping; it is not necessary to think much about the transition temperature of lipids and whatnot. The time scales are so small that little can happen in that respect.

2. There seems to be a problem in the backbone reconstruction routine, causing errors in the backbone trace, which will be the major factor in the secondary structure assignment. I think I've found a solution, but I don't yet fully understand the origin of the problem. I'll try to push the change tomorrow.

Cheers,

T.

Please Log in or Create an account to join the conversation.

  • z.sichaib
  • z.sichaib's Avatar Topic Author
  • Offline
  • Senior Boarder
More
5 years 4 months ago #7887 by z.sichaib
Dear Tsjerk,

Please excuse me for my impatience but,

Did you investigate the Temperature thing? Finally, what is the correct Temperature that we should use in the backmapping protocol (NVT equilibration steps: 200 or 300K?)

Also, what about the backbone reconstruction routine and secondary structures assignment? Did you find the origin of the problem?

Best,

Zeineb

Please Log in or Create an account to join the conversation.

More
5 years 4 months ago #7888 by tsjerk
Hi Zeineb,

My github has these things corrected (maybe Riccardo is a bit behind): www.github.com/Tsjerk/Backward

As I mentioned before, the 200K is to just be a bit careful during the equilibration, but it probably won't matter whether it's 200 or 300. I haven't tested that extensively. But please note that at that point the system is still relaxing from a not entirely physical state to something that's close enough to be part of the conformational ensemble. The temperature is not the biggest concern, except as a means of keeping the system from going about too wildly. The part after, the subsequent MD run, is where you really need to bring the rest of the physics back in order.

Cheers,

T.

Please Log in or Create an account to join the conversation.

  • z.sichaib
  • z.sichaib's Avatar Topic Author
  • Offline
  • Senior Boarder
More
5 years 4 months ago #7889 by z.sichaib
Dear Tsjerk,

Thank you for your fast reply. Now it's clear for me for the Temperature.

What about the secondary structures thing? Is it normal to have different secondary structures assignment (and the Ramachandran plot) when I take an AA structure ---> coarse-grain it ---> and backmap it again to AA, as I detailed in the first message?

Best,

Zeineb

Please Log in or Create an account to join the conversation.

More
5 years 4 months ago #7890 by tsjerk
If you take the latest version from github, the difference shouldn't be very large. The routine you sketch is the exact same procedure we used to check the quality of the backmapping for the paper.

T.

Please Log in or Create an account to join the conversation.

  • z.sichaib
  • z.sichaib's Avatar Topic Author
  • Offline
  • Senior Boarder
More
5 years 4 months ago #7896 by z.sichaib
Ok, great! Thanks a lot, Tsjerk :)

Cheers,

Zeineb

Please Log in or Create an account to join the conversation.

Time to create page: 0.120 seconds