- Posts: 54
Martinize with side chain fix
- flaviyan
-
Topic Author
- Offline
- Senior Boarder
Less
More
7 years 11 months ago - 7 years 11 months ago #5538
by flaviyan
Martinize with side chain fix was created by flaviyan
Hi,
Regarding the side chain fix proposed in the paper:Improved side chain dynamics in Martini simulations of protein-lipid interfaces.
My question is with respect to other salt bridge interaction which can be found in alpha helical parts and coils that if I am right martinize currently will not take into account while creating the CG topology.Or for that matter any other native contacts other than those that keep the secondary structure together and elastic bonds to maintain the tertiary structure.
I assume that the force field automatically should take care of these side chain interactions? But I am not sure if it does, I checked this for a protein I am working with and had both the atomic and CG simulations:
The atomistic simulation was carried out with CHARMM36 force field for 150 ns and the equilibrated structure was used as the input for Martinize and Elendyn22 force field was used to build the topology. The CG simulation was run for 1 microseconds. For the CG simulation the new-rf.mdp was used.
The salt bridge I am interested in is between the charged residues ARG and GLU in the atomistic simulation the salt bridge is maintained throughout the trajectory but the same sidechain interaction is not observed in the CG simulation.
The distribution plots of distances between the com of ARG guanidinium group [-C-(NH2)2] and com of GLU side chain [-C-C-COO-] in atomistic [] and CG [] simulations. In case of the CG plot the distance is between the beads SC2[Qd] of ARG and SC1[Qa] of GLU.
This image is hidden for guests.
As can be seen in the atomistic simulation the distance distribution has a sharp peak at 0.4 nm but he CG system has a peak at 1.5 nm which is beyond the interaction cut-off.
Regarding the side chain fix proposed in the paper:Improved side chain dynamics in Martini simulations of protein-lipid interfaces.
My question is with respect to other salt bridge interaction which can be found in alpha helical parts and coils that if I am right martinize currently will not take into account while creating the CG topology.Or for that matter any other native contacts other than those that keep the secondary structure together and elastic bonds to maintain the tertiary structure.
I assume that the force field automatically should take care of these side chain interactions? But I am not sure if it does, I checked this for a protein I am working with and had both the atomic and CG simulations:
The atomistic simulation was carried out with CHARMM36 force field for 150 ns and the equilibrated structure was used as the input for Martinize and Elendyn22 force field was used to build the topology. The CG simulation was run for 1 microseconds. For the CG simulation the new-rf.mdp was used.
The salt bridge I am interested in is between the charged residues ARG and GLU in the atomistic simulation the salt bridge is maintained throughout the trajectory but the same sidechain interaction is not observed in the CG simulation.
The distribution plots of distances between the com of ARG guanidinium group [-C-(NH2)2] and com of GLU side chain [-C-C-COO-] in atomistic [] and CG [] simulations. In case of the CG plot the distance is between the beads SC2[Qd] of ARG and SC1[Qa] of GLU.
This image is hidden for guests.
Please log in or register to see it.
As can be seen in the atomistic simulation the distance distribution has a sharp peak at 0.4 nm but he CG system has a peak at 1.5 nm which is beyond the interaction cut-off.
Last edit: 7 years 11 months ago by flaviyan. Reason: errors
Please Log in or Create an account to join the conversation.
- djurre
-
- Offline
- Admin
Less
More
- Posts: 272
7 years 11 months ago #5544
by djurre
Replied by djurre on topic Martinize with side chain fix
Interesting data!
I'm not very surprised that the salt bridge is last. In the work for the Martini 2.2 I did see that Positive/Negative charge interactions are often to week, even in water.
You could try to run the CG simulation again with the Martini 2.2P (or Elnedyn 2.2P) parameters. It might be that the salt bridge is better stabilized in that case.
Can you see in the simulations "what" causes the distance the increase? Is it a change in the overall shape of the protein (e.g. the alpha helices moving a way from each other) or is one of the side chains rotating to the other side of the backbone (as it was observed in "Improved side chain dynamics in Martini simulations of protein-lipid interfaces")?
I'm not very surprised that the salt bridge is last. In the work for the Martini 2.2 I did see that Positive/Negative charge interactions are often to week, even in water.
You could try to run the CG simulation again with the Martini 2.2P (or Elnedyn 2.2P) parameters. It might be that the salt bridge is better stabilized in that case.
Can you see in the simulations "what" causes the distance the increase? Is it a change in the overall shape of the protein (e.g. the alpha helices moving a way from each other) or is one of the side chains rotating to the other side of the backbone (as it was observed in "Improved side chain dynamics in Martini simulations of protein-lipid interfaces")?
Please Log in or Create an account to join the conversation.
- flaviyan
-
Topic Author
- Offline
- Senior Boarder
Less
More
- Posts: 54
7 years 11 months ago #5548
by flaviyan
Replied by flaviyan on topic Martinize with side chain fix
Hi Djurre,
I will test this with both martini2.2P and Elnedyn 2.2P and let you know.
I will also look in to the backbone angles and BB-SC1-SC2 angles in both the atomistic and CG system,
to make a proper comparison,
I will test this with both martini2.2P and Elnedyn 2.2P and let you know.
I will also look in to the backbone angles and BB-SC1-SC2 angles in both the atomistic and CG system,
to make a proper comparison,
Please Log in or Create an account to join the conversation.
- flaviyan
-
Topic Author
- Offline
- Senior Boarder
Less
More
- Posts: 54
7 years 11 months ago #5571
by flaviyan
Replied by flaviyan on topic Martinize with side chain fix
Hi Djurre,
Sorry for the delay in replying.
I tested the same system with Martini22, Martini22P, elnedyn22 and elnedyn22P. All four system did not show the salt bridge interaction seen in the atomistic system.
As to why:
In the case of both the martini22 and Martini22P the helix in which these residues are present(they are in two different helix) were pushed father apart due to structural distortion of the whole protein. This is expected due to the lack of elnedyn22 bonds.
In the case of elnedyn22 the tertiary structure was maintained very well but the residues never came within interacting distance, although the side chains beads did not rotate away from one another.
I tried removing all the elastic bonds that connected both these residues back bone beads, as well as the elastic bonds connecting the backbone beads of preceding and succeeding two residues, this some what made the helical region flexible enough to bend a little bit and the forcefield was able bring the beads to a salt bridge like interaction, the distribution of distance shifted from 1.5nm to 0.5-0.6nm. This happened in both the elnedyn22 as well as the elnedyn22P.
Sorry for the delay in replying.
I tested the same system with Martini22, Martini22P, elnedyn22 and elnedyn22P. All four system did not show the salt bridge interaction seen in the atomistic system.
As to why:
In the case of both the martini22 and Martini22P the helix in which these residues are present(they are in two different helix) were pushed father apart due to structural distortion of the whole protein. This is expected due to the lack of elnedyn22 bonds.
In the case of elnedyn22 the tertiary structure was maintained very well but the residues never came within interacting distance, although the side chains beads did not rotate away from one another.
I tried removing all the elastic bonds that connected both these residues back bone beads, as well as the elastic bonds connecting the backbone beads of preceding and succeeding two residues, this some what made the helical region flexible enough to bend a little bit and the forcefield was able bring the beads to a salt bridge like interaction, the distribution of distance shifted from 1.5nm to 0.5-0.6nm. This happened in both the elnedyn22 as well as the elnedyn22P.
Please Log in or Create an account to join the conversation.
- djurre
-
- Offline
- Admin
Less
More
- Posts: 272
7 years 11 months ago #5573
by djurre
Replied by djurre on topic Martinize with side chain fix
Very interesting. Summing-up one could say that without elastic network the ternary structure changes too much, while with elastic network the ternary structure changes too little to allow the formation of the salt bridge.
But that also suggest that there was no salt bridge in the crystal structure right? To some extend it then makes sense: the elastic network does not allow changes and thus no salt bridge forms.
If you still feel like trying more: what happens if you start with a structure with salt bridge (e.g. taken from the atomistic simulation) and start a Martini simulation (without elastic network, 2.2 or 2.2P) from there. Is the salt bridge enough to keep the structure stable?
In any way, good to know and thanks for letting us know!
But that also suggest that there was no salt bridge in the crystal structure right? To some extend it then makes sense: the elastic network does not allow changes and thus no salt bridge forms.
If you still feel like trying more: what happens if you start with a structure with salt bridge (e.g. taken from the atomistic simulation) and start a Martini simulation (without elastic network, 2.2 or 2.2P) from there. Is the salt bridge enough to keep the structure stable?
In any way, good to know and thanks for letting us know!
Please Log in or Create an account to join the conversation.
- flaviyan
-
Topic Author
- Offline
- Senior Boarder
Less
More
- Posts: 54
7 years 11 months ago #5574
by flaviyan
Replied by flaviyan on topic Martinize with side chain fix
Hi Djurre,
The initial structure that I used as input for martinize.py had the proper salt bridge interaction, it is a well relaxed structure from the atomistic simulation.So no the martini22 and 22P is unable to maintain either the salt bridge or the tertiary structure, even starting from a structure that has salt bridge. I collected the data after a series of EM, NVT and NPT runs where the backbone was position restrained before starting the MD run so I don't know when these beads moved farther apart. I can check that.
In the case of elnedyn your statement sums up my observations even stating from the structure with salt bridges the lack of flexibility prevents them being formed in CG simulations, this could be because the CA position is more rigid and the side chain bead move accordingly before they get a chance to interact to form the salt bridge when undergoing NVT and NPT where the backbone is positional restrained
The initial structure that I used as input for martinize.py had the proper salt bridge interaction, it is a well relaxed structure from the atomistic simulation.So no the martini22 and 22P is unable to maintain either the salt bridge or the tertiary structure, even starting from a structure that has salt bridge. I collected the data after a series of EM, NVT and NPT runs where the backbone was position restrained before starting the MD run so I don't know when these beads moved farther apart. I can check that.
In the case of elnedyn your statement sums up my observations even stating from the structure with salt bridges the lack of flexibility prevents them being formed in CG simulations, this could be because the CA position is more rigid and the side chain bead move accordingly before they get a chance to interact to form the salt bridge when undergoing NVT and NPT where the backbone is positional restrained
Please Log in or Create an account to join the conversation.
- flaviyan
-
Topic Author
- Offline
- Senior Boarder
Less
More
- Posts: 54
7 years 10 months ago #5620
by flaviyan
Replied by flaviyan on topic Martinize with side chain fix
Hi Djurre,
Just a quick question here, I know the martinize can read the trajectory and take the secondary structure assignment based on the trajectory (-ssc option), would it be also possible to detect the native contacts and construct the Elnedyn network based on the native contacts? I am pretty sure the default 0.9 cut off between BB ( i -> i+4) will essentially indirectly include all the native contacts, but this will not distinguish between side-chain mediated contact?
Just a quick question here, I know the martinize can read the trajectory and take the secondary structure assignment based on the trajectory (-ssc option), would it be also possible to detect the native contacts and construct the Elnedyn network based on the native contacts? I am pretty sure the default 0.9 cut off between BB ( i -> i+4) will essentially indirectly include all the native contacts, but this will not distinguish between side-chain mediated contact?
Please Log in or Create an account to join the conversation.
- flaviyan
-
Topic Author
- Offline
- Senior Boarder
Less
More
- Posts: 54
7 years 10 months ago #5621
by flaviyan
Replied by flaviyan on topic Martinize with side chain fix
also I noticed that in the Martini2.2 FF paper( Journal of Chemical Theory and Computation 2013 9 (1), 687-697
DOI: 10.1021/ct300646g) my observations are already seen for the charged residues
This image is hidden for guests.
DOI: 10.1021/ct300646g) my observations are already seen for the charged residues
This image is hidden for guests.
Please log in or register to see it.
Please Log in or Create an account to join the conversation.
- tsjerk
-
- Offline
- Expert Boarder
Less
More
- Posts: 103
7 years 10 months ago #5625
by tsjerk
Replied by tsjerk on topic Martinize with side chain fix
Hi Flaviyan,
The -ssc option is not for determining the secondary structure based on a trajectory; martinize will always do that. The -ssc option allows to tune when a certain type of secondary structure is chosen. By default the secondary structure needs to be present in at least half of the frames (-ssc 0.5) to be set.
The elastic network is still based on a single structure, and not on all frames, and it does not take into account what native contacts there are. In a strict sense that would also not be an Elnedyn network anymore, but an alternative. It could have some merit, but would require some work in coding it in and in testing it. Interested in a project like that?
The -ssc option is not for determining the secondary structure based on a trajectory; martinize will always do that. The -ssc option allows to tune when a certain type of secondary structure is chosen. By default the secondary structure needs to be present in at least half of the frames (-ssc 0.5) to be set.
The elastic network is still based on a single structure, and not on all frames, and it does not take into account what native contacts there are. In a strict sense that would also not be an Elnedyn network anymore, but an alternative. It could have some merit, but would require some work in coding it in and in testing it. Interested in a project like that?
Please Log in or Create an account to join the conversation.
- flaviyan
-
Topic Author
- Offline
- Senior Boarder
Less
More
- Posts: 54
7 years 10 months ago #5627
by flaviyan
Replied by flaviyan on topic Martinize with side chain fix
Hi Tsjerk,
You are right about the -ssc option,that is also my understanding. As to your second question I am definitely interested in a model that includes native contacts, and have experience with python.
You are right about the -ssc option,that is also my understanding. As to your second question I am definitely interested in a model that includes native contacts, and have experience with python.
Please Log in or Create an account to join the conversation.
- flaviyan
-
Topic Author
- Offline
- Senior Boarder
Less
More
- Posts: 54
7 years 10 months ago #5647
by flaviyan
Replied by flaviyan on topic Martinize with side chain fix
Hi Tsjerk,
I came up with the following scheme, assuming one has a equilibrium trajectory from atomistic simulation (say 20-25 frames):
1. Iterate over the trajectory frame by frame
2. Identify the center of mass of BB and SC is the particular frame
3. Check for BB -> BB native contact (based on a cut-off, this part is actually taken care of by the ElneDyn)
4. Check for SC -> SC native contact in charges residues and hydrogen bond donor/acceptor residues
5. Build a network model based on this (the trajectory information can be used for assigning the bond length and force constants)
Kindly let me what you think.
I came up with the following scheme, assuming one has a equilibrium trajectory from atomistic simulation (say 20-25 frames):
1. Iterate over the trajectory frame by frame
2. Identify the center of mass of BB and SC is the particular frame
3. Check for BB -> BB native contact (based on a cut-off, this part is actually taken care of by the ElneDyn)
4. Check for SC -> SC native contact in charges residues and hydrogen bond donor/acceptor residues
5. Build a network model based on this (the trajectory information can be used for assigning the bond length and force constants)
Kindly let me what you think.
Please Log in or Create an account to join the conversation.
- flaviyan
-
Topic Author
- Offline
- Senior Boarder
Less
More
- Posts: 54
7 years 2 months ago #7137
by flaviyan
Replied by flaviyan on topic Martinize with side chain fix
Please Log in or Create an account to join the conversation.
Time to create page: 0.110 seconds