Dear Nicholas,
Yes we are diverging, but you make a few important and valid points. So
let's diverge some more...
I fully agree that validation tools should not tell people how to do
science. That would kill free thought and lead to circularity in the
results. A typical example is the question what the RMSD from E&H bond
lengths should be at a given resolution. If a popular validation tool
would give a specific value and everyone would stick to that, then
eventually the PDB would reflect that even if there was no proof for the
initially claimed target value.
If you run into the problem you describe, that is, if a validation program
says there is something wrong with your structure because it fails to
recognise the cool stuff you did with your model, please complain. That is
by far the best way to ensure validation software gets better. You won't
just help yourself but the entire community.
As for the maps, I'll happily trust the maps of a good flesh-and-blood
crystallographer. But the avarage PDB user does not know who are the good
crystallographers and who are not. It also happens that crystallographers
make gross errors without realising it (the penta-retraction for
instance). I agree that a good flesh and blood person can get (much)
better results than an automatated procedure. But errors happen and we
should keep that in mind.
An example (not naming names here) is a structure I recently saw with a
large loop half a residue out-of-register. All the side chain density was
filled with water. It was an error of the type "How can anybody miss
that?". Their maps would certainly be very interesting. But I will put
more faith in the ones generated by the EDS. I hope you agree.
Cheers,
Robbie
> Date: Sun, 13 Dec 2009 17:20:04 +0200
> From: [log in to unmask]
> Subject: Re: [ccp4bb] FW: pdb-l: Retraction of 12 Structures...
> To: [log in to unmask]
>
> Dear Robbie, List,
>
> This thread is steadily diverging. Apologies for my contribution to its
> diversification.
>
>
> > Who knows what they did to the maps in terms of (unwarrented) density
> > modefication to make them look cleaner?
> >
> > The advantage of the EDS is that it is impartial and
> > uniform. The maps are generated in a clear and well-described way.
>
>
> I agree with you that map deposition is probably a waste of resources.
>
> I strongly disagree, though, with the existence of validation tools that
> have strong views about how best I should do science. For example, your
> sentence above imply that the validation tool is more fit (than myself) to
> decide which maps I should be looking at. Which means that if I chose to
> calculate (and view) not the simple FFT-derived map, but its maximum
> entropy estimate, I am in danger of being accused that 'I did something to
> the maps to make them look cleaner', where in fact, I'm just doing a
> better job out of the existing data than the validation tool (which
> probably generate maps in a clear, well-described and wrong way :-)
>
> The take home message of what I'm saying is this: We should not be
> deterred from practising our craft as best as we could, even if that
> implies that our models contain information that a validation tool can not
> reproduce. It is only fair that a well-informed and well-educated human
> being can do a better job than a fixed-frozen automated procedure. Fraud
> is a moral issue, and can not (and should not) be used as an excuse for
> converting validation tools to the sacred holders of scientific standards.
>
>
> My twocents,
> Nicholas
>
>
> --
>
>
> Dr Nicholas M. Glykos, Department of Molecular Biology
> and Genetics, Democritus University of Thrace, University Campus,
> Dragana, 68100 Alexandroupolis, Greece, Tel/Fax (office) +302551030620,
> Ext.77620, Tel (lab) +302551030615, http://utopia.duth.gr/~glykos/
|