Discussion on Strategies
Introductory Notes
- omega vs. phi scans
- beam polarization
- single sweep vs. multi sweep
- xtal shape as re-orientation/re-centering
factor
Reconstruction of the mean reflection intensities using limited experimental data set:
<E2> profiles – a feature of PROTEINS, NOT APPLICABLE TO SMALL MOLECULES
ˆJ( h )  1  ˆJ (h ) Exp(  h  B  h T )
u
s
12.5
1hq3 : [ ]=0.63, []=0.06
1at0 : [ ]=0.00, []=0.60
12
1d5t : [ ]=0.27, []=0.23
11.5
11
<
>I
n
l 10.5
10
9.5
9
(a)
10
5
3.33
2.5
2 1.67 1.43 1.25 1.11
o
resolution d (A)
1 0.909
Optimization target: Signal/Noise
• NOT the time to be spent for experiment,
number of frames to collect, etc …
• ALL the data collection parameters (multisub-wedge, variable exposure time, etc.)
are optimized simultaneously Example: multiplicity vs exposure time
Radiation Damage • Compensation of intensity decay by adjusting
(increasing) the exposure time / frame is
essential :
Total dose per data set is not important
– defined by the long exposure of the LAST frames
– short exposures of the FIRST frames are critical
What works in BEST now?
optimal orientation with respect to:
• Overlaps
(~90% of failing experiments – J. Holton )
- also with isometric cells @ high mosaicity
• Intrinsic diffraction anisotropy
each diffraction pattern is maximally isotropic,
S/N in a weak direction compensated by exposure
(small effect when judged by standard "resolution shell" statistics)
• Low noise in anomalous difference data
anomalous difference error model (radiation induced
non-isomorphism) accounts for the difference in dose between the observed
Bijvoet mates
Minimal RFriedel= <|<E+>-<E->|> vs. Resolution and Orientation (error
contribution to the difference only, no anomalous scattering contribution
16
16
14
14
P2
10
8
P222
12
010
100
110
Random
Rfriedel, %
Rfriedel, %
12
6
10
8
6
4
4
2
2
0
random
100
110
111
0
0
0.05
0.1
0.15
0.2
0.25
0
0.05
0.1
1/d^2
0.15
0.2
0.25
1/d^2
16
P23
14
16
P4
14
12
Rfriedel, %
Rfriedel, %
12
10
001
100
111
8
6
10
100
110
111
random
8
6
4
4
2
2
0
0
0
0.05
0.1
0.15
1/d^2
0.2
0.25
0
0.05
0.1
1/d^2 0.15
0.2
0.25
Data collection using multiple crystals
Reference images
Experimental aim
Auto-indexing
BEST
Crystal characterization and ranking
Determination of maximal achievable resolution
Optimal crystal orientation(s)
Crystal 3
D.C. plan
Completeness 23%
Crystal 5
Completeness 58%
Crystal 1
Completeness 91%
Crystal 8
Completeness 99.7%
Omega vs. Phi scans
Omega scans
- orientation wrt scan axis is optimized
Overlaps
Radiation-induced non-isomorphism
Multi-crystals
AAS
Phi scans
- orientation wrt BEAM (direction/electric field vector) is varied
"true redundancy" (– no advantage wrt. Omega,
but - may be - less limitations)
Blind region reduction ( - when in a symmetric setting)
AAS?
Beam polarization
• Isotropic scattering –
Scan axis || Electic Filed vector is optimal, though only
important at high resolution ( < 2*wavelength)
Vertical OMEGA is of advantage for the microbeam (gravity)
PHI is mechanically non-micro
• AAS
BEST minimizes the noise in anomalous diffrence
data (fully applicable to AAS data)
the target describing the AAS signal is required
Single Sweep vs. Multi Sweep
Multi sweep on a single crystal:
Blind region completion
Multiplicity
Partial data set completion (disaster scenario)
From the point of view of implementation in BEST, MultiSweep strategy is a particular case of multiple crystal data
collection optimization with the goniometric limitations
Single Sweep vs. Multi Sweep
• "Fast" coverage of an asymmetric unit on a
single crystal – no advantage in signal-to-noise!
Disadvantage – Inhomogeneous S/N
Single sweep
Radiation damage
Single sweep
RD compensation
Multiple sweeps
xtal shape as re-orientation/recentering factor
• Exploiting ALL of the crystal volume is
critically important
• Severe mismatch of Xtal/Beam size –
major limitation to sample characterization,
strategy and data quality in general
• Use Kappa to match the Xtal/Beam size
(at least in a vertical direction), Simplify line scans
along Omega