[gmx-users] Simulation parameter problem about protein unfolding

ms devicerandom at gmail.com
Mon Oct 18 15:16:31 CEST 2010


On 18/10/10 13:12, Chen wrote:
>
>> At 2010-10-18 16:38:30£¬"David van der Spoel"<spoel at xray.bmc.uu.se>  wrote:
>
>>> On 2010-10-18 06.56, chris.neale at utoronto.ca wrote:
>>> Generally, forcefields are not parameterized for temperatures other than
>>> 298K, so simulations are not expected to reproduce the expected
>>> properties (like boiling water and the correct temperature denaturation
>>> of proteins).
>>>
>>> There's almost certainly other issues here (including the fact that I'm
>>> entirely sure that you can get a lot more than 24 ns of simulation on a
>>> 54 aa protein; and 26 atom of pressure seems pretty arbitrary) but it
>>> will come down to this eventually.
>>>
>>> Just because you found a paper in which they get a denatured state does
>>> not imply that they got the correct denatured state.
>>>
>> There is no correct denatured state. There are infinitely many. Check
>> out recent work on NMR of "unfolded" proteins.
>
>
> I thought about the unfolding state space is huge. I just wonder
> whether the relative low radius of gyration value space sampled by us is caused
> by some error setting in MD parameters. If no one here finds there's problem in
> my MD parameters, then I can keep going. Thanks!

In theory, you should be able to calculate the expected gyration radius 
of a purely non-interacting random chain (which one can take as the 
limit of an unfolded protein). given the distance between monomeric 
units (i.e. C-alpha for example here). It is a three-dimensional random 
walk.
http://en.wikipedia.org/wiki/Radius_of_gyration#Molecular_applications

Doing the calculation for the equation of the three-dimensional random 
walk of your 54 aa protein, with N=54 and a=0.35 nm, it turns out that 
the expected gyration radius is:

 >>> (np.sqrt(54) * 0.35) * (1/np.sqrt(6))
1.0500000000000003 nm

So, yes, your average protein gyration radius is consistent with the 
protein being a random coil -that is, unfolded.

In truth your random coil ball should be a bit larger because in the 
calculation we ignore the chain self-avoidance, but on the other hand we 
have that a protein chain is not completely non-self-interacting, for 
obvious reasons. Therefore it seems you can conclude that your protein 
is reasonably close to being unfolded.

Hope it helps,
Massimo




More information about the gromacs.org_gmx-users mailing list