Artem Evdokimov wrote:
I can't resist asking: If we assume that the data fabrication
techniques and the techniques for discovery of such activities should
have the same sort of arms race as the development of viruses and
anti-malvare software (but of course on a much more modest scale since
structural biology is a relatively niche discipline) - can we then
I don't think this assumption holds for structure prediction, except in
the extreme asymptotic limit. All of the cases of fabricated data that
I've heard of were detected because the fabricated data didn't look like
actual experimental data - because our models for calculating data are
missing a variety of things that occur experimentally.
So a hypothetical arms race might be result in a better model of the
various components (and potential sources) of errors during data
collection and processing. But this would be a much more interesting
development in itself than any use for fabricating data.
Pete
speculate further that eventually the most sophisticated fabrication
techniques would be equivalent to de novo structure prediction :) It's
really too bad that there's no real money in this (again, relatively
speaking - not as much money as there is in software development),
because if there was then the structural biology equivalent of 'virus
hackers' would in reality approximate the same development trajectory
as the most successful (and legitimate) protein modelers. Given the
ingenuity of hackers and like-minded people in general, I sometimes
wonder if this isn't a better way to develop structure prediction
tools...
Artem
On Sun, Apr 1, 2012 at 10:09 AM, Paul Emsley <paul.ems...@bioch.ox.ac.uk> wrote:
On 31/03/12 23:08, Kevin Jin wrote:
I really wish PDB could have some people to review those important
structures, like paper reviewer.
So do the wwPDB, I would imagine.
But they can't just magic funding and positions into existence...
If the coordinate is downloaded for modeling and docking, people may not
check the density and model by themself. However this is not the worst case,
since the original data was fabricated.
1. All of data was correct and real,
Hmmm...
It will be very difficult for people to check the density and coordinated
if he/she is not a well-trained crystallographer.
I hope and believe that this is not the case. Even basically-trained
crystallographers should be able to calculate and interpret difference maps
of the kind described by Bernhard. And with the EDS and PDB_REDO server,
one does not even need to know how to make generate a difference map...
Paul.
