Dear Gerard, Eleanor et al.,
In the graphs of Rmeas vs. batch number (and I or I/sigma vs. batch
number), everything looks roughly constant, which made me think that
significant radiation damage was not the issue--but are you saying that
radiation damage might also (or instead) produce a smiley in the graph of
Rmeas vs. resolution? Looking back at the diffraction images, they do seem
to degrade as each sweep progresses, though there are always some clear
diffraction spots past my current resolution cutoff (1.6 A). The beam was
100 um and the crystal was about 700x500x300 um; I translated the beam 110
um along the long axis between shots, so the spots are probably mostly but
not entirely fresh.
Another thing I noticed is that in the sweeps with the smiley-face Rmeas
vs. resolution trace (lower in the middle, roughly equal at high and low
resolution) the mean I/sigma vs resolution trace also looks different:
instead of being highest at 4-5 A and dropping steadily as resolution
increases, it's lower than I would expect at high resolution and higher at
low resolution. Perhaps the weakest high-res spots are disappearing
altogether and not being counted? Tables from an early sweep and a late
sweep from this crystal are below, in case that's interesting.
Thanks,
Veronica
Good sweep:
Overall InnerShell OuterShell
Low resolution limit 21.77 21.77 1.63
High resolution limit 1.60 8.76 1.60
Rmerge (within I+/I-) 0.025 0.028 0.113
Rmerge (all I+ and I-) 0.027 0.031 0.122
Rmeas (within I+/I-) 0.030 0.033 0.133
Rmeas (all I+ & I-) 0.030 0.034 0.132
Rpim (within I+/I-) 0.016 0.018 0.069
Rpim (all I+ & I-) 0.012 0.015 0.050
Rmerge in top intensity bin 0.024 - -
Total number of observations 108218 524 5456
Total number unique 16662 121 804
Mean((I)/sd(I)) 43.0 62.8 15.0
Mn(I) half-set correlation CC(1/2) 0.999 0.998 0.993
Completeness 99.9 92.3 100.0
Multiplicity 6.5 4.3 6.8
Questionable sweep:
Overall InnerShell OuterShell
Low resolution limit 21.77 21.77 1.63
High resolution limit 1.60 8.76 1.60
Rmerge (within I+/I-) 0.027 0.036 0.041
Rmerge (all I+ and I-) 0.029 0.039 0.045
Rmeas (within I+/I-) 0.032 0.044 0.049
Rmeas (all I+ & I-) 0.032 0.044 0.049
Rpim (within I+/I-) 0.017 0.025 0.027
Rpim (all I+ & I-) 0.013 0.020 0.020
Rmerge in top intensity bin 0.034 - -
Total number of observations 102687 511 4744
Total number unique 16644 121 799
Mean((I)/sd(I)) 53.8 53.5 34.1
Mn(I) half-set correlation CC(1/2) 0.999 0.997 0.998
Completeness 99.8 92.1 99.9
Multiplicity 6.2 4.2 5.9
On Sat, Feb 28, 2015 at 6:41 AM, Gerard Bricogne <[email protected]>
wrote:
> Dear Veronica,
>
> At first glance, it looks as if you have a textbook example of
> progressive radiation damage, i.e. the structure is changing between
> the start and the end of your experiment, and the scaling gets skewed
> towards finding a best compromise between all the measurements at
> medium (rather than low) resolution. The "smiley" shape in the Rmeas
> is something that Zbyszek Dauter identified many years ago as a
> tell-tale sign of progressive radiation damage.
>
> You say that each of your 6 datasets is "taken from a fresh spot
> on the crystal", but how big is your beam relative to the crystal? How
> far apart are these spots? Most of all: at room temperature, free
> radicals will diffuse much more readily than in a frozen crystal, so
> that your later datasets may not in fact come from truly fresh spots.
>
> I hope this is helpful.
>
>
> With best wishes,
>
> Gerard.
>
> --
> On Fri, Feb 27, 2015 at 03:40:38PM -0500, Veronica Pillar wrote:
> > Hi all,
> >
> > I have a data set from a large room-temperature lysozyme crystal
> consisting
> > of 6 90-degree sweeps, each taken from a fresh spot on the crystal. When
> I
> > scale & merge the first sweep by itself, the R[meas/merge] vs. resolution
> > trace as reported by aimless looks fairly normal (lowest values in the
> > 6-2.5 A range and steadily increasing at higher resolution). However, as
> I
> > look at the subsequent sweeps individually, the traces get progressively
> > stranger until the final sweep's R vs. resolution trace looks like a
> smiley
> > face, lowest around 2.5 A and about the same in the lowest and highest
> > resolution bins. The overall Rmeas is about the same for all sweeps
> > (3.0-3.2%). Do you have any ideas on what has happened (either during
> data
> > collection or data processing) to cause this?
> >
> > Thanks,
> > Veronica
>
