I think CC* (derived from CC1/2) is an important step forward in how to
decide where to cut off the data you give to your refinement program,
but I don't think it is a good idea to re-define what we call the
"resolution of a structure". These do NOT have to be the same thing!
Remember, what we crystallographers call "resolution" is actually
about 3x the "resolution" a normal person would use. That is, for most
types of imaging whether it be 2D (pictures of Mars) or 3D (such as
electron density) the "resolution" is the minimum feature size you can
reliably detect in the image. This definition of "resolution" makes
intuitive sense, especially to non-crystallographers. It is also
considerably less pessimistic than our current definition since the
minimum observable feature size in an electron density map is about 1/3
of the d-spacing of the highest-angle spots. This is basically because
the d-spacing is the period of a sine wave in space, but the minimum
feature size is related to the full-width at half max of this same wave.
So, all you have to do is change your definition of "resolution" and a
3.0 A structure becomes a 1.0 A structure!
However, I think proposing this new way to define "resolution" in
crystallography will be met with some resistance. Why? Because changing
the meaning of "resolution" so drastically after ~100 years would be
devastating to its usefulness in structure evaluation. I, for one, do
not want to have to check the deposition date and see if the structure
was solved before or after the end of the world (Dec 2012) before I can
figure out whether or not I need to divide or multiply by 3 to get the
"real" resolution of the structure. I don't think I'm alone in this.
Now, calling what used to be a 1.6 A structure a 1.42 A structure (one
way to interpret Karplus & Diederichs 2012) is not quite as drastic a
change as the one I flippantly propose above, but it is still a change,
and there is a real danger of "definition creep" here. Most people
these days seem to define the resolution limit of their data at the
point where the merged I/sigma(I) drops below 2. However, using CC* =
0.5 would place the new "resolution" at the point where merged
I/sigma(I) drops below 0.5. That's definitely going beyond what anyone
would have called the "resolution of the structure" last year. So,
which one is it? Is it a 1.6 A structure (refined using data out to
1.42 A), or is it actually a 1.42 A structure?
Unfortunately, if you talk to a number of experienced crystallographers,
they will each have a slightly different set of rules for defining the
"resolution limit" that they learned from their thesis advisor, who, in
turn, learned it from theirs, etc. Nearly all of these "rule sets"
include some reference to Rmerge, but the "acceptable" Rmerge seems to
vary from 30% to as much as 150%, depending on whom you talk to.
However, despite this prevalence of Rmerge in our perception of
resolution there does not seem to be a single publication anywhere in
the literature that recommends the use of Rmerge to define the
resolution limit. Several papers have been cited to that effect, but
then if you go and read them they actually made no such claim.
Mathematically, it is fairly easy to show that Rmerge is wildly unstable
as the average intensity approaches zero, so how did we get stuck on it
as a criterion for evaluating the outer resolution bin? I'm not really
sure, but I think it must have happened around 1995. Before that, there
are NO entries for Rmerge in the high-resolution bin in the PDB. Not
one. Looking at papers from the pre-1995 era, you don't see it reported
in "table 1" either. What is more, ever since 1995, the average reported
Rmerge in the high-resolution shell has been slowly rising by about 1.6
percentage points each year. Started around 20%, and now it is up to
50%. Seriously. Here is the graph:
http://bl831.als.lbl.gov/~jamesh/pickup/outershell_Rmerge.png
I think this could be yet another example of "definition creep". For any
given year, I imagine a high-resolution Rmerge that is only "a few
percent worse" than the average over "the PDB" at that time is probably
considered "okay", and the average just keeps increasing over time.
Nevertheless, Rmerge is a useful statistic for evaluating the quality of
a diffractometer, provided it is used in the way it was originally
defined by Uli Arndt: over the entire dataset for spots with I/sd > 3.
At large multiplicity, the Rmerge calculated this way asymptotically
approaches the average "% error" for measuring a single spot. If it is
more than 5% or so, then there might be something wrong with the camera
(or the space group choice, etc). This is only true for Rmerge of ALL
the data, not when it is relegated to a given resolution bin.
Perhaps it is time we did have a discussion about what we mean by "the
resolution of a structure" so that some kind of historically relevant
and "future proof" definition for it can be devised? Otherwise, we will
probably one day see 1.0 A used to describe what today we would call a
3.0 A structure? The whole point here is to be able to compare results
done by different people at different periods in history to each other,
so I think its important to try and keep our definition of "resolution"
stable, even if we do "use" spots that are beyond it.
So, what I would advise is to refine your model with data out to the
resolution limit defined by CC*, but declare the "resolution of the
structure" to be where the merged I/sigma(I) falls to 2. You might even
want to calculate your Rmerge, Rcryst, Rfree and all the other R values
to this resolution as well, since including a lot of zeroes does nothing
but artificially drive up estimates of relative error. Perhaps we
should even take a lesson from our "small molecule" friends and start
reporting "R1", where the R factor is computed only for hkls where
I/sigma(I) is above 3?
-James Holton
MAD Scientist
On 12/8/2012 4:04 AM, Miller, Mitchell D. wrote:
I too like the idea of reporting the table 1 stats vs resolution
rather than just the overall values and highest resolution shell.
I also wanted to point out an earlier thread from April about the
limitations of the PDB's defining the resolution as being that of
the highest resolution reflection (even if data is incomplete or weak).
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1204&L=ccp4bb&D=0&1=ccp4bb&9=A&I=-3&J=on&d=No+Match%3BMatch%3BMatches&z=4&P=376289
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1204&L=ccp4bb&D=0&1=ccp4bb&9=A&I=-3&J=on&d=No+Match%3BMatch%3BMatches&z=4&P=377673
What we have done in the past for cases of low completeness
in the outer shell is to define the nominal resolution ala Bart
Hazes' method of same number of reflections as a complete data set and
use this in the PDB title and describe it in the remark 3 other
refinement remarks.
There is also the possibility of adding a comment to the PDB
remark 2 which we have not used.
http://www.wwpdb.org/documentation/format33/remarks1.html#REMARK%202
This should help convince reviewers that you are not trying
to mis-represent the resolution of the structure.
Regards,
Mitch
-----Original Message-----
From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of Edward A.
Berry
Sent: Friday, December 07, 2012 8:43 AM
To: CCP4BB@JISCMAIL.AC.UK
Subject: Re: [ccp4bb] refining against weak data and Table I stats
Yes, well, actually i'm only a middle author on that paper for a good
reason, but I did encourage Rebecca and Stephan to use all the data.
But on a later, much more modest submission, where the outer shell
was not only weak but very incomplete (edges of the detector),
the reviewers found it difficult to evaluate the quality
of the data (we had also excluded a zone with bad ice-ring
problems). So we provided a second table, cutting off above
the ice ring in the good strong data, which convinced them
that at least it is a decent 2A structure. In the PDB it is
a 1.6A structure. but there was a lot of good data between
the ice ring and 1.6 A.
Bart Hazes (I think) suggested a statistic called "effective
resolution" which is the resolution to which a complete dataset
would have the number of reflectionin your dataset, and we
reported this, which came out to something like 1.75.
I do like the idea of reporting in multiple shells, not just overall
and highest shell, and the PDB accomodatesthis, even has a GUI
to enter it in the ADIT 2.0 software. It could also be used to
report two different overall ranges, such as completeness, 25 to 1.6 A,
which would be shocking in my case, and 25 to 2.0 which would
be more reassuring.
eab
Douglas Theobald wrote:
Hi Ed,
Thanks for the comments. So what do you recommend? Refine against weak data,
and report all stats in a single Table I?
Looking at your latest V-ATPase structure paper, it appears you favor something
like that, since you report a high res shell with I/sigI=1.34 and Rsym=1.65.
On Dec 6, 2012, at 7:24 PM, Edward A. Berry<ber...@upstate.edu> wrote:
Another consideration here is your PDB deposition. If the reason for using
weak data is to get a better structure, presumably you are going to deposit
the structure using all the data. Then the statistics in the PDB file must
reflect the high resolution refinement.
There are I think three places in the PDB file where the resolution is stated,
but i believe they are all required to be the same and to be equal to the
highest resolution data used (even if there were only two reflections in that
shell).
Rmerge or Rsymm must be reported, and until recently I think they were not
allowed
to exceed 1.00 (100% error?).
What are your reviewers going to think if the title of your paper is
"structure of protein A at 2.1 A resolution" but they check the PDB file
and the resolution was really 1.9 A? And Rsymm in the PDB is 0.99 but
in your table 1* says 1.3?
Douglas Theobald wrote:
Hello all,
I've followed with interest the discussions here about how we should be refining against weak
data, e.g. data with I/sigI<< 2 (perhaps using all bins that have a
"significant" CC1/2 per Karplus and Diederichs 2012). This all makes statistical
sense to me, but now I am wondering how I should report data and model stats in Table I.
Here's what I've come up with: report two Table I's. For comparability to legacy structure stats,
report a "classic" Table I, where I call the resolution whatever bin I/sigI=2. Use that
as my "high res" bin, with high res bin stats reported in parentheses after global stats.
Then have another Table (maybe Table I* in supplementary material?) where I report stats for the
whole dataset, including the weak data I used in refinement. In both tables report CC1/2 and Rmeas.
This way, I don't redefine the (mostly) conventional usage of "resolution", my
Table I can be compared to precedent, I report stats for all the data and for the model
against all data, and I take advantage of the information in the weak data during
refinement.
Thoughts?
Douglas
^`^`^`^`^`^`^`^`^`^`^`^`^`^`^`^`^`^`^`^`
Douglas L. Theobald
Assistant Professor
Department of Biochemistry
Brandeis University
Waltham, MA 02454-9110
dtheob...@brandeis.edu
http://theobald.brandeis.edu/
^\
/` /^. / /\
/ / /`/ / . /`
/ / ' '
'
