AW: Re: AW: AW: AW: why does the transition seem to be irreversible in aMD

From: Norman Geist (norman.geist_at_uni-greifswald.de)
Date: Tue May 24 2016 - 00:57:09 CDT

Oops, forgot you tell you why. It’s because of the fixed atoms. As you maybe know, during NPT when the box grows and shrinks periodically, the atom coordinates are rescaled in order to fit into the box. Since the fixed atoms are not rescaled, free atoms are moved towards the fixed ones, which causes high repulsion forces. The higher settings for decay and period of langevinpiston pressure control will cause a slower and smoother NPT run, allowing the system to react better on the appearing forces and so should crash.

 

Norman Geist

 

Von: owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] Im Auftrag von Norman Geist
Gesendet: Montag, 23. Mai 2016 15:42
An: namd-l_at_ks.uiuc.edu; 'sunyeping' <sunyeping_at_aliyun.com>
Betreff: AW: namd-l: Re: AW: AW: AW: namd-l: why does the transition seem to be irreversible in aMD

 

Are you running an NPT ensamble? If yes, try increasing langevinpistonperiod and decay by about 10x times.

 

Norman Geist

 

Von: owner-namd-l_at_ks.uiuc.edu <mailto:owner-namd-l_at_ks.uiuc.edu> [mailto:owner-namd-l_at_ks.uiuc.edu] Im Auftrag von sunyeping
Gesendet: Montag, 23. Mai 2016 10:54
An: Norman Geist < <mailto:norman.geist_at_uni-greifswald.de> norman.geist_at_uni-greifswald.de>
Cc: namd-l < <mailto:namd-l_at_ks.uiuc.edu> namd-l_at_ks.uiuc.edu>
Betreff: namd-l: Re: AW: AW: AW: namd-l: why does the transition seem to be irreversible in aMD

 

Hi,

I am trying to fix the CA atoms of the protein whem applying boost energy, but constraint failure occurs

ERROR: Constraint failure in RATTLE algorithm for atom 7874!
ERROR: Constraint failure; simulation has become unstable.
ERROR: Constraint failure in RATTLE algorithm for atom 7926!
ERROR: Constraint failure; simulation has become unstable.
ERROR: Constraint failure in RATTLE algorithm for atom 3117!
ERROR: Constraint failure; simulation has become unstable.
ERROR: Constraint failure in RATTLE algorithm for atom 3175!
ERROR: Constraint failure in RATTLE algorithm for atom 8734!
ERROR: Constraint failure; simulation has become unstable.
ERROR: Constraint failure in RATTLE algorithm for atom 3590!
ERROR: Constraint failure in RATTLE algorithm for atom 11621!
ERROR: Exiting prematurely; see error messages above.

 

Is there a way to address this problem?
====================================================

WallClock: 3.514848 CPUTime: 3.252506 Memory: 528.042969 MB
[Partition 0][Node 0] End of program
ERROR: Constraint failure; simulation has become unstable.
ERROR: Constraint failure; simulation has become unstable.
ERROR: Constraint failure; simulation has become unstable.

 

------------------------------------------------------------------

From:Norman Geist <norman.geist_at_uni-greifswald.de <mailto:norman.geist_at_uni-greifswald.de> >

Time:2016 Apr 5 (Tue) 17:31

To:孙业平 < <mailto:sunyeping_at_aliyun.com> sunyeping_at_aliyun.com>

Cc:namd-l <namd-l_at_ks.uiuc.edu <mailto:namd-l_at_ks.uiuc.edu> >

Subject:AW: AW: AW: namd-l: why does the transition seem to be irreversible in aMD

 

Ahh, that’s a different thing. As the boost potential will also affect the protein, you might need to fix its structure (maybe CA atoms) so the boost is only active for your ligand. Also, you could prevent the short peptide from dissociation using a weak constraint.

 

 

Norman Geist

 

Von: sunyeping [mailto:sunyeping_at_aliyun.com]
Gesendet: Dienstag, 5. April 2016 11:11
An: namd-l < <mailto:namd-l_at_ks.uiuc.edu> namd-l_at_ks.uiuc.edu>; Norman Geist < <mailto:norman.geist_at_uni-greifswald.de> norman.geist_at_uni-greifswald.de>
Cc: namd-l < <mailto:namd-l_at_ks.uiuc.edu> namd-l_at_ks.uiuc.edu>
Betreff: Re: AW: AW: namd-l: why does the transition seem to be irreversible in aMD

 

Dear Norman,

 

    Maybe you are right, but I am not sure.

    The conformation 1 I mentioned corresponds to the state that both termini of a short peptide bind to a protein of two long chains, and the conformation 2 corresponds to the state that one terminus of the short peptide dissociates with and strecthes away from the protein.

    The transition from conformation 1 (the initial crystal structure) to conformation 2 can occur at a relatively smaller boost energy with the whole protein structure seeming to be normal. Increasing the boost energy leads to the distortion of the whole protein together with the transition from the conformation 1 to 2. It seems to me that the distortion or disorder is irrelevant to the transition, and it does not predict the transition paths, neither. So I can hardly believe that adding a larger boost energy can lead to the transition from the conformation 2 to 1. Instead, I tend to guess that a larger boost energy will lead to more serious distortion or even collapse of the protein (dissociation of the two long chains), or dissociation of the whole short peptide from the protein. I may presume that in a extremely entended aMD this collapse process could reverse, but I cannot afford such a long-time simulation. So it is important to design the simulation rationally.

 

Best regards.

 

Yeping

------------------------------------------------------------------

From:Norman Geist < <mailto:norman.geist_at_uni-greifswald.de> norman.geist_at_uni-greifswald.de>

Time:2016 Apr 5 (Tue) 16:07

To:孙业平 < <mailto:sunyeping_at_aliyun.com> sunyeping_at_aliyun.com>

Cc:namd-l < <mailto:namd-l_at_ks.uiuc.edu> namd-l_at_ks.uiuc.edu>

Subject:AW: AW: namd-l: why does the transition seem to be irreversible in aMD

 

Usually that’s how conformational sampling is about to work. The structures that you call disordered or distorted are part of the conformational space and also part of the transition paths between intermediate local minima. You will have to sample some amount of data and analyze the probability of states e.g. by creating histograms of proper reaction coordinates like rgyr and mainchain hbonds and compute free energies from it. This will also tell you the amount of energy required to do the transition you mentioned.

 

 

Norman Geist

 

Von: sunyeping [ <mailto:sunyeping_at_aliyun.com> mailto:sunyeping_at_aliyun.com]
Gesendet: Dienstag, 5. April 2016 09:48
An: namd-l < <mailto:namd-l_at_ks.uiuc.edu> namd-l_at_ks.uiuc.edu>; Norman Geist < <mailto:norman.geist_at_uni-greifswald.de> norman.geist_at_uni-greifswald.de>
Betreff: Re: AW: namd-l: why does the transition seem to be irreversible in aMD

 

Dear Norman,

 

Thank you for your replay. You are right that the aMD is aimed at conformational sampling but I wish to see that transitions between the two conformation occur alternately. It seems that the transition from conformation 1 to conformation 2 is easy in my aMD, but not vice versa. However, if I keep increasing the boost energy, the whole protein structure becomes distorted and many structural motifs become disordered in addition to the conformational transition I am studying. So I guess simply increase the boost energy may be not a wise strategy.

 

Best regards.

 

Yeping

 

From:Norman Geist < <mailto:norman.geist_at_uni-greifswald.de> norman.geist_at_uni-greifswald.de>

Time:2016 Apr 5 (Tue) 15:30

To:namd-l < <mailto:namd-l_at_ks.uiuc.edu> namd-l_at_ks.uiuc.edu>, 孙业平 < <mailto:sunyeping_at_aliyun.com> sunyeping_at_aliyun.com>

Subject:AW: namd-l: why does the transition seem to be irreversible in aMD

 

Well If I’m correct, you don’t actually want your structure to get stucked in an aMD, as the main goal is to perfrom conformational sampling. So your boost potential might be too small to leave the rather stable structure your system ran into.

 

 

Norman Geist

 

Von: <mailto:owner-namd-l_at_ks.uiuc.edu> owner-namd-l_at_ks.uiuc.edu [ <mailto:owner-namd-l_at_ks.uiuc.edu> mailto:owner-namd-l_at_ks.uiuc.edu] Im Auftrag von sunyeping
Gesendet: Dienstag, 5. April 2016 05:21
An: namd-l < <mailto:namd-l_at_ks.uiuc.edu> namd-l_at_ks.uiuc.edu>
Betreff: namd-l: why does the transition seem to be irreversible in aMD

 

Dear NAMD users,

 

I ran an aMD with NAMD and observed a transition in my structure from conformation 1 to conformation 2 within 10 ns. However, when extending the simulation, I couldn't observe the reversion from conformation 2 to conformation 1. Since the aMD is done by adding an extra boost energy to the simulation system, does that mean that conformation 2 is located in the high energy contour in the potential surface? Is the aMD biased to sample the conformations in the high energy contour in the potential surface rather than those in low energy basins? What can be done to biasly sample conformations in low energy basins? To be specific, how could I realize the reversion from conformation 2 to conformation 1?

 

Yeping

 

 

 

This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:20:29 CST