[phenixbb] So what is unique?
nechols at lbl.gov
Tue Jul 15 09:06:20 PDT 2014
On Tue, Jul 15, 2014 at 8:26 AM, Kenneth A. Satyshur <satyshur at wisc.edu>
> "Unique Reflections"
> My definition is all the possible reflections out to the highest
> resolution reported not related by symmetry.
> Where can I find this? The .mtz contains a list of all HKL calculated to
> the highest resolution. Usually, we
> are not able to measure all these diffraction spots due to limits of the
> detector, mechanical limits, crystal
> orientation, etc.
Not "all possible", "all observed". I've never seen any of the statistics
in depositions (or Table 1, etc.) interpreted to include reflections that
weren't actually measured in the experiment. But different programs are
not at all consistent about what "unique" means either: phenix.refine
counts F+ and F- separately, Refmac (and the Table 1 program in Phenix)
counts F+ and F- as a single reflection.
> The depositions server asks for total reflections. I assume it wants only
> those unique reflections we were able to
> collect, regardless of the sigma cut off. These are called 'observed'. The
> total we use in refinement will be a subset
> of the 'unique observed' that are cut on sigma.
Other way around - unique is a subset of total. My assumption would be
that this refers to the number of reflections observed in the experiment,
not accounting for symmetry or multiplicity, but it really depends on
context. (In our Table 1, "Total" = "separately integrated", prior to
You might consider checking the formal mmCIF definition as well:
although to be honest some of these are pretty vague too. Certainly the
deposition server should give clearer instructions. However, in the (near)
future the task of PDB deposition should become much simpler and the
various programs (including Phenix) will automatically take care of these
issues for you. (phenix.refine already has the ability to output mmCIF
files directly, which may save you some time.)
However, some crystallographers believe that we should not cut
> on sigma since some of the intensities may in fact be zero. Is this a
> question for the Refmac and Phenix people?
Many would argue that the traditional cutoff (e.g. I/sigma>2) is too
conservative and throws out useful information; and most modern
crystallography software (e.g. Phenix or Refmac) should behave sensibly
with very weak high-resolution data. I'm not going to rehash all of the
arguments here, but we think that CC1/2 is a much more useful statistic for
this purpose. We definitely don't recommend applying any type of filtering
beyond a single global resolution cutoff - so no filtering individual
reflections, elliptical truncation, permanent anisotropy correction, or
other selective removal/doctoring of data.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the phenixbb