Hi Ethan,
yes, you are absolutely right, you would need to define somehow where
your model is... But you could, at least hypothetically, use non-atomic
models for this! Like cylinders (*), spheres and similar shapes. This is
what the density looks like at those "super-low" resolutions.
(*)
BOROVIKOV, B. A., VAINSTEIN, B. K., GELFAND, I. M. & KALININ, D. I.
(1979). Kristallografiya, 24, 227-238.
KALININ, D. I. (1980). Kristallografiya, 25, 535-544.
V. Yu. Lunin, A. G. Urzhumtsev, E. A. Vernoslova, Yu. N. Chirgadze, N.
A. Neveskaya and N. P. Fomenkova. Acta Cryst. (1985). A41, 166-171.
Anyway, looks like we are about to diverge from the original subject so
I stop rambling -:)
All the best!
Pavel.
On 9/14/10 1:10 PM, Ethan Merritt wrote:
> On Tuesday 14 September 2010 12:44:37 pm Pavel Afonine wrote:
>> Hi Ethan,
>>
>>> I do not favor depositing riding hydrogen coordinates for
>>> the same reason that I do not like the recent PDB preference
>>> for depositing ANISOU records for structures that have been
>>> refined with TLS.
>>>
>>> In both cases the enumeration of these many thousands of
>>> parameter values gives the strong, but false, impression that
>>> they have been individually modeled. They have not.
>> following this logic one could say that the individual x,y,z coordinates
>> listed in ATOM records for a structure refined at very low resolution
>> using rigid-body refinement only (or torsion angle Simulated Annealing
>> only) also may make a false impression that these coordinates were
>> refined individually.
> I agree with this, at least for the case of true rigid-body.
> But you would still need to describe somehow the coordinates of all the
> atoms in your rigid model. If it came straight out of the PDB, then
> in principle it would suffice to give the PDB+CHAIN code and the
> rotation/translate matrix. But if any adjustments were made, which
> is I think typical if only to correct for sequence differences,then as a
> practical matter you still need to provide the true starting coordinates.
> And at that point you might as well provide the ending coordinates instead,
> since it's the same amount of information.
>
> Ethan
>
|