1. BEST strategy / SAD optimization Gleb Bourenkov EMBL-Hamburg Kappa Workgroup Meeting September 28-29, 2009 MAXLAB

2. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 X-Ray Dose : The energy deposited (via inelastic scattering processes) per mass of a crystal sample Units – Gray (Gy) Empirical reference figures for radiation damage : 30 MGy Recommended maximum total dose per data set for data collection Owen et al. PNAS 2006 ~1 MGy kinetic rate of (fast) site-specific damage processes the radiation damage may start affecting anomalous signal Translations to the corresponding exposure times at beamlines are available No significant dependencies on the details of how the dose is deposited – i.e. on the photon flux/exposure time, photon energy, etc.* (Significant) variation between different crystals ascribed to variation in absorbance e.g. high salt, heavy atom soaks Journal of Synchrotron Radiation Special Issues on RD - 2006, 2008 J. Holton's survey on http://biosync.rcsb.org *Storage ring, Monochromatic beam

3. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 1.5 MGy 7 MGy 30 MGy One and the same crystal sample (translated between data sets), measured with different total dose/data set %Rmrg Rpim Rano CCano 3.8 1.5 3.5 77 9.0 3.0 4.7 61 16.6 6.0 5.6 36 7.8 3.0 3.9 60 %Rmrg Rpim Rano CCano 6.5 2.9 3.5 40 17.9 6.3 5.9 24 40.8 18.6 9.6 5 12.4 4.6 4.5 23 Reso %Rmrg Rpim Rano CCano >4.0 4.8 2.5 3.6 50 2.6-2.4 21.6 8.3 7.2 20 2.1-2.0 61.3 23.7 16.4 3 All-2.0 7.8 6.7 6.1 30 Low dose High dose Optimized data collection parameters / dose distribution 7% 35% 70%

4. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 1. Modeling the data statistics as a function of data collection parameters 2. Modeling the diffraction intensity variation with X-ray dose Optimum data collection conditions for a particular crystal are assessed via modeling

5. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 Radiation Damage Model (Bulk Xtl MX, Cryo) Diffraction Intensity is a function of dose I(hkl,Dose)=scale(Dose,|hkl|)*I(hkl)+Δ(Dose) ● overall Debye-Waller factor (B) grows by 1 Å 2 per 1 MGy ● Luzatti isomorphism factor (Log σ A ) decays by 0.1 Å 2 per 1 MGy β=8π 2 s AD = 0.95 0.95 1.4 1.3 Å 2 /MGy Kmetko et al. (2006 ) Owen et al. (2006) D ½ = 4.3(±0.3) x10 7 Gy >2.5 Å ↔ β=1.05 Å 2 Popov et al. (2006) Dose RD factors, A 2 β= 1.0±0.3 Å 2 /MGy α= 0.1±0.03 Å 2 /MGy

6. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 The user choice Space group, Cell parameters, Orientation, Mosaicity I[(h,k,l), T exposure ], I background I/Sigma Resolution SAD data Dose(Time) Constrains Geometry Dose Rate Beamline Flux/BeamCrossection Reference Frames

7. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 Signal-to-Noise vs Dose bovine trypsin resolution shell 1.75-1.70 Å 180x0.5 degrees oscillation frames at ID29 Dose rate 10 5 Gy/sec

8. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 Data Collection with Variable Exposure Time and Oscillation Width BEST optimizes the data collection parameters for each crystal orientation (i.e. spindle position reached after exposure to a certain dose) individually; For convenience of data collection/processing the data collection "plan" is smoothed out to produce a small number sub-wedges with varying exposure/oscillation width Even without taking the Radiation damage into account, this is useful (e.g. severely anisotropic diffraction or long cell edge) For high-dose data collection, BEST suggest to increase the exposure time gradually during the data collection, in order to compensate the loss of the diffraction signal due to the radiation damage (according to the model-based expectations) and keep signal-to-noise at a required level.

9. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 Native data: requested I/SigI in the last resolution bin is a target spindle angle ( o ) # strong reflections BEST, strategy+predictions XDS, data statistics Total dose 21 MGy

10. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 SAD optimization Minimum of R Friedel = - |> is a target noise only, no anomalous scattering itself: decay, non-isomorphism exact pair-vice dose differences for Bijvoet mates Minimal R Friedel vs. Resolution -> relate expected anomalous signal --------------------------------------------- Resolution RFriedel(%) I/Sigma Multiplicity --------------------------------------------- 10.12 0.8 74.1 23.7 6.90 0.8 43.6 23.7 5.34 1.1 48.4 23.0 4.51 1.2 47.5 23.5 3.98 1.6 34.5 20.6 3.60 2.5 22.4 13.9 3.31 4.0 14.0 11.9 3.08 6.6 8.3 7.0 2.89 10.5 5.2 6.1 2.73 15.6 3.7 2.5 2.60 23.0 2.4 3.8 ---------------------------------------------

11. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 SAD optimization: search for an optimal crystal orientation minimal R Friedel vs. Resolution – Orientation - Symmetry 4.53.22.62.2Å

12. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 Interface: CCP4I BEST

13. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 Multi-crystal data collection

14. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 Truncate rotation range Single crystal Lower the resolution Multiple crystals

15. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 General approach (current view): All sample pre-screened (versus an incremental approach) All crystals selected for data collection will be used in the same orientation (versus "random" orientation - single-axis gonio case, or optimal orientation for each) Additional cross-characterization round to define the non- isomorphism covariance matrix between selected samples

16. Gleb Bourenkov Kappa Workgroup Meeting29/09/2009 Workflow Standard pre-screening (low dose) EDNA pre-selection CELL cluster ORIENTATION common accessible orientation space, + optimization SCALE,B-FACTOR S/N level comparable? MOSAICITY (can weaker diffractors "help" stronger ones) BACKGROUND Reduced subset, selected orientation Cross-Screening narrow wedge in a selected orientation (low dose) EDNA-improved characterization quantified non-isomorphism – σ A matrix Joint Strategy