normal How sensitive of a simulaiton code it is?

  • dujiangfeng
  • dujiangfeng's Avatar Topic Author
  • Offline
  • Fresh Boarder
More
11 years 6 months ago #1160 by dujiangfeng
How sensitive of a simulaiton code it is? was created by dujiangfeng
Dear All,

Maybe the title is a big ambiguous! I mean that how big difference between pme and shift for coulomb type. I tried my protein-membrane system by pme and shift respectively. As an initial configuration with a distance of 3.5 nm between protein and membrane, the 'PME' code made the protein went to membrane by 300 ps while the 'shift' code didn't drive the protein to membrane.

My question here is that do we trust the result from the 'PME' code? it is too fast (300 ps) to be true. I repeated the simulation eight times, and all of them went to membrane around 300 ps, although the protein-membrane complexes are different to each other.

Thank you guys,

Jiang.

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

More
11 years 6 months ago #1161 by xavier
Replied by xavier on topic How sensitive of a simulaiton code it is?
Dear Jiang,

You are facing an unfortunate dilemma ... there is no definitive answer to your question. The difference you observe between the PME and cutoff is large and reflects some effect we have seen earlier on systems. We have been investigating the reason for this difference and have not yet been able to identify the exact source. To our understanding the treatment of elec interactions with PME might introduce some artifactual forces in a system in which electrostatics are important; here I mean that there are many charges in the system that local dipole or high density charge might be high. We unfortunately have not been able to reach the point where we can predict the outcome of the artifact ...

Objectively you should follow the method that would reproduce the experimental/atomistic information that you have in hand. If the only discrepancy with your "understanding" of the system is the kinetics, that might not be so bad! You system might actually behave reasonably.

Have you tried to simulate a protein/lipid bilayer complex (formed using PME) with a shift setup? If the protein leaves the bilayer that might be a problem ...

What makes you think that the protein should be interacting with the lipid bilayer?

I hope this helps.
XAvier.

dujiangfeng wrote: Dear All,

Maybe the title is a big ambiguous! I mean that how big difference between pme and shift for coulomb type. I tried my protein-membrane system by pme and shift respectively. As an initial configuration with a distance of 3.5 nm between protein and membrane, the 'PME' code made the protein went to membrane by 300 ps while the 'shift' code didn't drive the protein to membrane.

My question here is that do we trust the result from the 'PME' code? it is too fast (300 ps) to be true. I repeated the simulation eight times, and all of them went to membrane around 300 ps, although the protein-membrane complexes are different to each other.

Thank you guys,

Jiang.

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

  • dujiangfeng
  • dujiangfeng's Avatar Topic Author
  • Offline
  • Fresh Boarder
More
11 years 6 months ago #1162 by dujiangfeng
Replied by dujiangfeng on topic How sensitive of a simulaiton code it is?
Dear Xavier,
Thanks for your quick reply.

xavier wrote: Have you tried to simulate a protein/lipid bilayer complex (formed using PME) with a shift setup? If the protein leaves the bilayer that might be a problem ...


Yes. I did that. After the protein binds to membrane, I turned it to shift scheme and the complex is stable and some residues from the protein got buried into membrane even.


xavier wrote: What makes you think that the protein should be interacting with the lipid bilayer?


A plenty of references demonstrate the protein's membrane binding ability. I am going to calculate the binding energy of the protein and membrane by using umbrella sampling as your work published this year (X Periole, 2012). It seems only "PME" combined with pulling code is successful otherwise an error of "out of rlist, rcoulomb cutoff" came up.

One more thing, in your publication, you calculated the solvent accessible area but I couldn't get how to calculate? does 'g_sas' work in coarse grained system? Actually I tried this program but it didn't work then.

Thank you again,

Jiang.

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

More
11 years 6 months ago #1166 by xavier
Replied by xavier on topic How sensitive of a simulaiton code it is?
g_sas should work! You need to use a larger probe though to avoid the entrance of small probe everywhere ... the size used is generaly 2.6 Ang.

dujiangfeng wrote: Dear Xavier,
Thanks for your quick reply.

xavier wrote: Have you tried to simulate a protein/lipid bilayer complex (formed using PME) with a shift setup? If the protein leaves the bilayer that might be a problem ...


Yes. I did that. After the protein binds to membrane, I turned it to shift scheme and the complex is stable and some residues from the protein got buried into membrane even.


xavier wrote: What makes you think that the protein should be interacting with the lipid bilayer?


A plenty of references demonstrate the protein's membrane binding ability. I am going to calculate the binding energy of the protein and membrane by using umbrella sampling as your work published this year (X Periole, 2012). It seems only "PME" combined with pulling code is successful otherwise an error of "out of rlist, rcoulomb cutoff" came up.

One more thing, in your publication, you calculated the solvent accessible area but I couldn't get how to calculate? does 'g_sas' work in coarse grained system? Actually I tried this program but it didn't work then.

Thank you again,

Jiang.

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

Time to create page: 0.093 seconds