Not surprising since I'm sure they are both based on the Crick-Magdoff
equation (http://dx.doi.org/10.1107/S0365110X56002552) or the very
similar Hendrickson-Teeter equation (http://dx.doi.org/10.1038/290107a0).
Practically important thing is the signal-to-noise ratio. Roughly
speaking, to measure a difference of 1.5% you need a signal-to-noise
that is at least 1/1.5% = 60. So, if you typically get I/sd ~ 10
(cumulative to the "resolution of interest", wich is usually ~3.5A for
SAD phasing), then you need to average over (60/10)^2 = 36 crystals to
bump the overall I/sd up to what you need. This is assuming the
crystals are all isomorphous of course. With a 3% Bijvoet ratio, you
only need I/sd ~ 30. This is not all that hard to do with a single
good-quality crystal at "normal" multiplicity (~3-4). I think this is
why most successful MAD/SAD structure determinations have Bijvoet ratios
around 3%. The world record
(http://dx.doi.org/10.1107/S0907444906038534) used a Bijvoet ratio of
0.5% and very high multiplicity. Kay wrote a paper about this
"asymptotic" I/sd (http://dx.doi.org/10.1107/S0907444910014836), which
tops out around 30 for most diffractometers. The source of this
limitation is actually detector calibration. That hasn't exactly been
published yet, except in the detector literature of course
(http://dx.doi.org/10.1063/1.1488674 section IV.B), but who has time to
read that?
-James Holton
MAD Scientist
On 6/2/2013 12:43 PM, Edward A. Berry wrote:
Similar result. Taking all defaults except #atoms, amt protein:
kDa #res #sites BjvtRatio
protein of interest:
LBL: 469 12 0.0153 (4 e-)
Ruppweb 4273 12 0.016 (edge + 1 ev)
Ruppweb 4273 12 0.013 (CuKa)
Hemoglobin protomer
LBL 32 2 0.0239 (4 e-)
Ruppweb 291 2 0.024 (edge + 1 ev)
Ruppweb 291 2 0.020 (CuKa)
James Holton wrote:
I put together a little jiffy for feasibility of MAD experiments here:
http://bl831.als.lbl.gov/xtalsize.html
It does not calculate the f" value for you, but in general f" is ~4
electrons for K edges and ~10 electrons for L edges. Crossec can give
you more accurate values than this as long as you are "far away" from
the absorption edge (10-100 eV), but if you are right on top of the
"white line", then you have to measure f" anyway using an absorption
scan. That said, it is very rare for the white line to be more than
double the edge jump and Fe is not particularly spectacular for white
lines, so I wouldn't expect much more than ~6 electrons from iron.
-James Holton
MAD Scientist
On 6/2/2013 8:47 AM, Edward A. Berry wrote:
Is Ethan Merritt's anomalous scattering page at:
http://www.bmsc.washington.edu/scatter/
down or moved, or the firewall I'm behind is blocking it?
I want to check feasibility of a native-iron MAD experiment,
and I'm not very good at math.
thanks,
eab
