Why are
we not
solving
more
struct
tures?
James Holton University of California San Francisco and Advanced Light Source Lawrence Berkeley National Laboratory Berkeley, CA USA This work was supported by contributions from the ALS participating research team, a University of California Campus-Laboratory Collaboration Grant and grants from the National Institutes of Health: GM74929 and GM Beamline was funded by the National Science Foundation, the University of California, Berkeley, the University of California, San Francisco and Henry Wheeler. The Advanced Light Source is supported by the Director, Office of Science, Office of Basic Energy Sciences, Materials Sciences Division, of the US Department of Energy under contract No. DE-AC02-05CH11231 at Lawrence Berkeley National Laboratory.
About 50 data sets (MAD,SAD or native) are collected for every PDB deposition Only one in 12 MAD/SAD datasets can be solved Failures are generally due to: Overlaps – run strategy! Site-specific damage - stay under ~5 MGy Insufficient signal-to-noise - need ΔF ano > σ(ΔF ano ) Summary
SecondsDescriptionPercent Assigned and available91% Shutter open40% Collecting (3026 images)50% Something else50% Beamline operation efficiency “representative” user
SecondsDescriptionPercent Something else 100% 247s 45 Mounting11% 229s 37 Centering8% 179s 109 Strategizing19% 309s 37 Prepping12%
NumberDescriptionPercent Images (~7 TB)33% of light 2346 Data sets47% of light 449 MAD/SAD (1:2)19% of data sets 48 Published2% of data sets ALS in 2003 Structure Productivity
28 operating US beamlines ~10 11 ph/μm 2 exposure limit ÷ 2x10 9 ph/μm 2 /s x 25% beamline operation efficiency ≈ 100,000 datasets/year ÷ 1324 str in 2003 ≈ 2% efficient USA Structure Productivity Henderson (1990) Jiang & R.M.Sweet (2004) biosync.sdsc.edu
Investigated with Elves Automation Elves examine images and set-up data processing Elves run… mosflm scala solve mlphare dm arp/warp
Apr 6 – at ALS Investigated with Elves Automation 27,686images collected 31investigators 56unique cells 5 KDa – 23 MDaasymmetric unit 0.94 – 32 Åresolution (3.2 Å)
148datasets 117succeded ~3.5 (0.1-75)hours 31failed ~61 (0-231)hours 2 / 15MAD structures Why do structures fail?
Overlaps Signal to Noise Radiation Damage Why do structures fail?
Avoiding Overlaps c c
Is it real, or is it MLFSOM ? Simulate diffraction experiment to test hypotheses
MAD phasing simulations Anomalous signal to noise ratio Correlation coefficient to correct model mlphare results Threshold of interpretability
Reduce Noise: minimize background scattering Resolution (Ǻ) Photons/s/pixel Se edge with detector at 100 mm
Increase Signal: use multiple crystals and incremental strategy incremental_strategy.com merged.mtz auto.mat
e-e- e-e- + Interactions of x-rays with matter + e-e- e-e- + e-e- elastic scattering inelastic scattering Photoelectron emission fluorescence Secondary ionization
Radiation Damage Lattice Damage Site-specific Damage
Where do photons go? beamstop elastic scattering (6%) Transmitted (98%) useful/absorbed energy: 7.3% inelastic scattering (7%)Photoelectric (87%) Protein 1A x-rays Re-emitted (~0%)Absorbed (99%) Re-emitted (99%)Absorbed (~0%)
Protein crystal in oil x-rays cause sample expansion before after 20% glycerol
Data quality vs phasing quality dose (MGy) Correlation coefficient
Individual atoms decay at different rates dose (MGy) Correlation coefficient to observed data
fraction unconverted peak valley
660%