Greetings to ccp4bb readers!
This is a long-delayed reply to requests that I received after I
posted an observation on the ccp4bb almost three years ago. (I think it was
early 2012). The paper had not yet been published (pending finalization of
experiments unrelated to structural biology) and so I was not free to
distribute any of the structure factors to those who were interested. The
paper now has been published, and the coordinates and structure factors
have been released. The following is the peculiarity that made this of
interest.
I refined a structure with 1.55 A data of a homotrimeric protein
bound to an inhibitor. Although the inhibitor added to the enzyme contained
a five-membered ring, the inhibitor we observed in two of the three active
sites had the ring opened. We therefore discovered that the ring-opened
form is also an inhibitor. However, in the remaining active site, we
discovered *both* the ring-opened and ring-closed forms of the inhibitor.
(The active sites are identical, but crystal contacts create asymmetry
among the three). The two forms of the inhibitor were positioned slightly
differently in this active site, and so the electron density at this
resolution clearly identified the presence of both forms, even though there
was much overlap in their positions. I used the CNS occupancy refinement
script, and then normalized the results so that the two occupancies added
up to 1, and so I came up with occupancies of 0.88 (ring-opened) and 0.12
(ring-closed). Then having all inhibitor molecules in the model, I
calculated maps with another program, which substituted Fcalc for missing
reflections, and could no longer see the lower occupancy form in the
electron density. I re-calculated maps in CNS, and again saw both forms in
the electron density. So I calculated new maps in CNS the same way as
before, but turned on the option to substitute missing reflections with
Fcalc: and the lower occupancy form was missing from the density in the
resulting maps.
So, this was the difference - when I used zero for missing
reflections, I was able to see both forms; but when I substituted Fcalc for
missing reflections in maps, I could only see the predominant form. The
data set was 99.6% complete overall, 99.2% in the highest resolution shell;
with a multiplicity of 5.8 (5.3 in the highest shell). So there didn't seem
to be many missing reflections to be substituted. The test set contained
roughly 2% of the total number of observed unique reflections.
Some had replied to me, saying that they were interested in using
the coordinates & structure factors for running tests; however, as I said,
I was not free to distribute them at the time. Now they are publicly
available: PDB ID 4K9G. I hope this helps all who are interested. Thank
you.
Gregg
***************************
*Gregg Crichlow *Visiting Research Scientist
Department of Pharmacology
Yale University School of Medicine
On Thu, Nov 27, 2014 at 5:14 PM, Keller, Jacob <[email protected]>
wrote:
> >We just had a chance to read this most interesting discussion. We would
> agree
>
> with Ian that jiggling or SA refinement may not be needed if refinement
> can in
>
> fact be run to convergence. However, this will be difficult to achieve for
> large
>
> structures, especially when only moderate to low resolution data are
> available.
>
>
>
> I find this interesting—is refinement convergence related to resolution?
> Is this because the structure-landscape is not sufficiently defined to find
> the real global minimum? I wonder what would happen if Ian Tickle’s test
> were done on many structures, and results examined as a function of
> resolution? Predictions? I guess generally there are more ways to fit fewer
> data points than many, but then perhaps refinement convergence would be
> more dependent on parameter:observation ratios than resolution per se,
> although quantities are closely related all things being equal.
>
>
>
> JPK
>
