Hear, Hear!
[rant]This *biological difference* thinking may be the source of some of the poor 'protein + ligand' structures in the PDB that look like the ligand was fitted without any attention to the rest of the model (or the refinement thereof). Well, at least these models are good teaching material.[/rant]
Using 1.42A data instead of 1.6A data may be enough to move from isotropic to anisotropic B-factors which may in turn lead to a better description of the movement of atoms. This could be biologically relevant. Adding correct waters leads to a better description of what is in the crystal, which in turn may lead to a more interpretable map. This could help you improve the rest of your model which in turn may be biologically relevant.
There is no guarantee that improving your model will lead to a biological difference, but you can only find out if you try. And like Ed said, it's not that much extra effort.
Cheers,
Robbie
> -----Original Message-----
> From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of
> Ed Pozharski
> Sent: Thursday, December 13, 2012 20:01
> To: [log in to unmask]
> Subject: Re: [ccp4bb] refining against weak data and Table I stats
>
> On Thu, 2012-12-13 at 17:50 +0000, Theresa Hsu wrote:
> > Being a beginner crystallographer, may I ask a basic question? On how
> > many occasions does it make a *biological* difference between having a
> > structure at 1.42 and 1.6 A?
>
> And your definition of "biological difference" is exactly what? Every field of
> experimental science strives to obtain results of best possible quality, why
> should macromolecular crystallography be different? Would you be satisfied
> with a report that "binding assays show presence of binding which perhaps is
> in submicromolar range but we don't care to determine actual binding
> constants"?
>
> You should also realize that while ccp4bb has evolved over years into a forum
> covering topics well beyond computational aspects of macromolecular
> crystallography, at its core it still is made of individuals who value method
> development. As Black Queen told Alice - it takes all the running you can do
> to keep in the same place.
>
> > I think this question also extends to adding in water molecules just
> > to make statistics look good.
>
> I never understood this attitude. Compared to O/CNS combination on SGIs
> (which are all excellent products) refining a model using COOT/REFMAC on a
> modern Core i7 machine is a cakewalk. Let's see - one spends from several
> months to a year or two going from gene to diffraction data, and spending a
> week carefully rebuilding in order to obtain the best most complete model
> possible is undue burden? I am not a perfectionist by any measure, but
> deliberately not placing water molecules that you can place because it "does
> not make biological difference" can hardly be justified.
>
> Cheers,
>
> Ed.
> >
>
> --
> Oh, suddenly throwing a giraffe into a volcano to make water is crazy?
> Julian, King of Lemurs
|