[gmx-users] RE: Re: Wierd results from Umbrella sampling (Justin A. Lemkul)

[Justin A. Lemkul]

On 5/21/12 10:05 AM, Du Jiangfeng (BIOCH) wrote:

Please don't reply to the entire digest message; copy and paste only the 
relevant parts.

> On 5/16/12 4:08 AM, Du Jiangfeng (BIOCH) wrote:
>> Dear Sir/Madam,
>>
>> I have performed umbrella pulling and umbrella sampling my protein from a
>> DOPC/DOPS membrane. Unfortunately, the results are really bad (Energy curve
>> suddenly turns to zero at the last 1 nm) and the histograph does not show any
>> overlap. Actually, I did it strictly based on Justin's tutorial, with the
>> sample spacing of 0.2 nm.
>>

Can you provide images of what your histograms and PMF profile look like?

> What energetic term is this?  A summation can decrease if the values being added
> are all negative.
>
> The energetic term should be binding energy.
> Even there are some samples with the negative energies but the energy curve shouldn't go to zero (I calculated them by hand).
>

I don't understand what this means.  Binding energy is not stored in the .edr 
file, and I'm not clear what you calculated yourself.

>>
>> Followings are some lines from the end of histograph file:
>>
>> Distance(nm) 5.455761 0       0       0       0       0       0       0       0       0       0       0       0       0       0       0       0       0       0       8 5.483663      0       0       0       0
>> 0     0       0       0       0       0       0       0       0       0       0       0       0       0       8 5.511565      0       0       0       0       0       0       0       0       0       0       0       0       0       0       0       0       0       0       12
>>   5.539467     0       0       0       0       0       0       0       0       0       0       0       0       0       0       0       0       0       0       4 5.567369      0       0       0       0       0       0       0       0       0       0
>> 0     0       0       0       0       0       0       0       2
>>
>>
>> I am really depressed because it took me quiet a long time to sampling but it
>> seems in vain... I really no idea to find out what went wrong.
>>
>
> Nor do we.  What is in your .mdp file?  How many windows are you using?  What is
> the total desired length of the reaction coordinate, and what are the initial
> and final COM distances that you are restraining?  What are your box dimensions?
>
> -Justin
>
> My box dimensions are  x: 7.94677  y: 7.94677  z: 13.48094 (nm)
> The initial COM of DOPC(lipids) is at z=3.5 nm;
> The initial COM of my protein is at z= 6.65nm.
> The desired length is 3.6 nm, but it moved 3.4 nm along z-dimension.
> So I used 19 windows for 3.4 nm movement (0.17 nm / window)
>
> In pulling.mdp:
> dt= 2 fs;  nsteps=1 ns; nstxout=10 ps; nstvout=10ps
> rlist/rcoulomb/ rvdw= 1.4
> Tcoupl= v-rescale;  Pcoupl= Parrinello-Rahman
> ; Pull code
> pull            = umbrella
> pull_geometry   = distance  ; simple distance increase
> pull_dim        = N N Y
> pull_start      = yes       ; define initial COM distance>  0
> pull_ngroups    = 1
> pull_group0     = DOPC
> pull_group1     = protein
> pull_rate1      = 0.0036      ; 0.0036 nm per ps = 3.6 nm per ns
> pull_k1         = 1300      ; kJ mol^-1 nm^-2
>
> In umbrella sampling.mdp:
> nsteps=9.5 ns; pull_rate1= 0.0; pull_k1= 1000
> other parameters are same as pulling.mdp.
>
>
> Thank you for your help, above are my parameters.
>

In principle, this setup should work fine.  What does grompp say about the 
restrained distances?  Does it agree with what you've calculated above in terms 
of COM distances?

-Justin

-- 
========================================

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================