Re: relax-users Digest, Vol 116, Issue 38 -- September 30, 2016 - 23:00

Hi Edward,

The protein itself is a monomer/dimer mix, normally it is a monomer;
however, the concentrations at which we observe it at (NMR concentrations
are around 1mM, the protein forms a dimer (primarily). Using titration
experiments we have found what looks to be an interface (using CSP), and
have used a docking program to show what the dimer would look like in
regards to the dimer interface and what is geometrically/energetically
possible. So we aren't looking for proof of a dimer, but my PI had informed
me that our S^2 would not be accurate if we used the pdb of the monomer
(due to slower tumbling effecting our relaxation data, basically having
data for a dimer, and thus the pdb file must also account for the larger
size/slower tumbling, basically since the data is for a dimer, the pdb
file/structure should also be that of a dimer so they correlate). However,
if the S^2 data doesn't get effected too much whether it is a dimer or
monomer, then I guess it doesn't matter too much in this case.

Also, could you tell me the exact modification I need to make to my script?
This is what it was before
structure.read_pdb('cluster1_12.pdb',set_mol_name='hRGS4')
Is this what I should modify it to?
structure.read_pdb('cluster1_12.pdb',set_mol_name='hRGS4',read_mol=0)



Sincerely,
Sam

On Fri, Sep 30, 2016 at 11:45 AM, Edward d'Auvergne <edward@xxxxxxxxxxxxx>
wrote:

> On 30 September 2016 at 19:45, Mahdi, Sam <sam.mahdi.846@xxxxxxxxxxx>
> wrote:
> > Sorry, I just want to make sure I fully understand this so I can explain
> it
> > to my PI:
>
> No problems, this is by far the most complicated aspect in the field of
> NMR ;)
>
>
> > So if there is symmetry, I can upload the same pdb file with the dimer
> (set
> > A and B) but tell it to read only one set.
>
> Load rather than upload, but yes.
>
>
> > Since S^2 isn't effected too much
> > versus a dimer versus a monomer, the only thing that is important is the
> > change in co-ordinates of one set of the dimer (i.e. the differnence in
> > co-ordinates between set A in a monomer, and set A in a dimer
> co-ordinates,
> > or set A in a different version of that dimer's co-ordinates).
>
> S2 is not affected by the reference frame.  This only matters for
> comparing diffusion tensors.  Though you will only ever see one
> tensor, as that is what is in your NMR sample (if you have a
> monomer-dimer mix, then you're in trouble and will see a lot of
> artificial Rex and ns motions).
>
>
> > I say this
> > because I have already run my protein's data with the pdb structure of
> the
> > monomer, and I have 2 different pdb files the docking program gave back
> for
> > the dimer (2 different ways the dimer could form from one interface).
>
> Well, your analysis will always return the same diffusion tensor.  If
> you want these diffusion tensors to all be in the same frame, use the
> relax structure.superimpose user function with the method='fit to
> first' argument.  Then just pick which will be your reference
> structure and superimpose.  You can superimpose A and B - separately -
> onto the monomer frame.
>
> Let's pick the monomer as the reference frame.  Then:
>
>     - If you superimpose "dimer 1, struct A" to the monomer, you
> should find the same tensor.
>     - If you superimpose "dimer 1, struct B" to the monomer, you
> should find the same tensor.
>     - If you superimpose "dimer 2, struct A" to the monomer, you
> should find the same tensor.
>     - If you superimpose "dimer 2, struct B" to the monomer, you
> should find the same tensor.
>
> If the docking program did not optimise the internal monomer
> structure, you will get identical results.  Otherwise you'll see minor
> internal motion changes.  If your PI was hoping that you would be able
> to tell him that you have a monomer or one of the 2 dimers in your NMR
> tube, well then you will need to start to read many, many papers on
> diffusion tensor prediction.  But know that all prediction methods
> underestimate the diffusion tensor (e.g. David Case is working on this
> exact problem for MD simulations).  In this case, relaxation data is
> not the best NMR method for this.  It would be better to use RDCs from
> a purely steric alignment and to compare that to what PALES prediction
> comes up with (though that itself is still a very rough and imperfect
> method).
>
> Regards,
>
> Edward