Re: Which conformation is more proper for the SMD

From: Eric H. Lee (
Date: Mon Jun 09 2008 - 13:47:27 CDT

Shulin -

Your RMSD values do suggest that the protein is most likely
equilibrated. I have done quite a few SMD simulations myself and have
published procedures on equilibrations before the SMD run. In many
cases today, I will equilibrate for much longer than 1ns, but then I
also run rather large flexible systems that often require 20-50ns for
accurate sampling. Feel free to do a pubmed search for "Lee and
Schulten" and you'll find a few references.

I think what is clear from your brief equilibration simulation is that
the protein you are studying is fairly flexible/mobile, and this is
leading to the dynamic behavior of the hydrogen bonds you observe.
What you might look for then, are the hbonds which are conserved
throughout your entire equilibration simulation (or towards the last
few 100ps of it), using a much more stringent criteria (closer
distance, narrower angle). These are likely the critical hbonds
involved in enforcing stability for your protein.


On Jun 9, 2008, at 1:21 PM, Shulin Zhuang wrote:

> Hi, Eric,
> Great thanks for your so rapid reply. Just as what you suggest, I
> measure the Hbond with the criterion: the distance between acceptor
> and donor pairs cutoff :3.5 A; donor-H-acceptor angle cutoff is 30 .
> My protein has 90 residues. For the whole Hbonds within the protein,
> the initial x-ray strucute has 60 Hbond; the minimized strucutre has
> 71 Hbonds; after heating the system to 300K, the final heated
> conformation has 56 Hbond; for the NPT equlibiration, the first
> 500ps has 55 Hbond; the 1ns conformation has 49 Hbonds; 1.1 ns
> conformation has 48 Hbonds; 1.2 ns conformation has 50 Hbonds; 1.3
> ns conformation has 44 Hbonds; 1.4 ns conformation has 39 Hbonds;
> for 1.5 ns conformation, it have 56 H bonds in total. Some times the
> Hbonds may not be the same in above Hbonds.
> I do the equlibration just according to some SMD papers, they often
> do 1ns equilibration before starting SMD. Generally compared with
> the initial structure, if the backbone RMSD is below 2.0 A, it is
> ok. For my case, compared with the x-ray structure, all the
> equilibrated conformation has a backbone RMSD belwo 1.2 A, several
> conformation has a backbone ofaround 0.9 A. Here I have a question
> that how long is enough for the equilibration before SMD. How to
> effectively judge whether the equilibrated conformation is right or
> not. Great thanks!
> Best regards
> Shulin
> On 6/9/08, Eric H. Lee <> wrote:
> How are you measuring whether a hydrogen bond is present? Are you
> using VMD to visualize them? Keep in mind that one can adjust
> distance and angle cutoffs, and it is best to measure these values
> manually to confirm that the Hbond is actually "broken." If the
> bond is obviously not viable (such as the acceptor donor pairs being
> separated by say, >3.5A), then it can be considered "broken."
> I would actually equilibrate with free dynamics longer (say double
> the time to 3ns total) and observe how this hydrogen bond behaves,
> in order to determine whether it may significantly impact your SMD
> simulation or not.
> Eric H. Lee
> Medical Scholars Program
> Theoretical and Computational Biophysics Group, UIUC
> On Jun 9, 2008, at 11:43 AM, Shulin Zhuang wrote:
>> Dear All,
>> Before I perform SMD on my protein, I first equilibrate it for 1ns.
>> Based on this 1ns equilibrated conformation, I run another 500 ps.
>> I use the same config parameter to do the equilibration on the same
>> linux cluster. For the 1ns equilibrated conformation and 1.5 ns
>> conformation, compared with x-crystal structure, these two
>> conformation have a backbone RMSD of 0.9. The backbone RMSD
>> between this two conformation is 0.7. However, compared with the x-
>> crystal structureI find that in 1.5 ns equilibrated conformation,
>> one backbone hydrogen bond is lost, while in the 1ns equilibrated
>> conformation, no backbone hydrogen bond is lost. Here I wonder if
>> the 1.5ns equilibrated conformation is right or not, should I only
>> use the 1ns equilibrated conformation for the SMD? Thanks a lot!
>> Best regards
>> Shulin

Eric H. Lee
Medical Scholars Program
Theoretical and Computational Biophysics Group, UIUC

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:49:33 CST