Developments in xia 2 Graeme Winter CCP 4

Developments in xia 2 Graeme Winter CCP 4 Dev Meeting 2008

What is xia 2? Automated robust data reduction and analysis n Thorough – takes additional steps when many users wouldn’t bother n In: images from e. g. synchrotron beamline n Out: measurements for downstream phasing via e. g. HAPPy, Mr BUMP, Phenix… n

Recent changes Inclusion in CCP 4 6. 1 n Many command line options n Integrated with Auto. Rickshaw (EMBL H) n Robust lattice determination n Support for Q 270, Pilatus n Zero input option n

3 Month plans Bio. XHit ends in June => so does xia 2 development n Include robust system to decide resolution limits etc (next slides) n Finish release 0. 3. 0 to go with release version of CCP 4 6. 1 n

Chef Let’s cook them books!

What is chef? A tool to help you use the best of the reflections you have n Uses unmerged intensities n Uses robust statistics to decide: n ¨ d*min for different functions (resolution) ¨ Dmax for different functions (dose) n Additional program “doser” to add dose information to unmerged MTZ files

In MTZ files from scala with “output unmerged” set n DOSE / TIME information for doser: n ¨ BATCH 1 DOSE 2. 5 TIME 2. 5 ¨ BATCH 2 DOSE 7. 5 TIME 8. 2 ¨…

Running doser hklin TS 03_12287_chef_INFL. mtz hklout infl. mtz < doser. in doser hklin TS 03_12287_chef_LREM. mtz hklout lrem. mtz < doser. in doser hklin TS 03_12287_chef_PEAK. mtz hklout peak. mtz < doser. in chef hklin 1 infl. mtz hklin 2 lrem. mtz hklin 3 peak. mtz << eof isigma 2. 0 resolution 1. 65 range width 30 max 1500 print comp rd rdcu anomalous on labin BASE=DOSE eof

Output Resolution vs. dose n Completeness vs. dose for each data set n

Methods n Based on “new” cumulative-pairwise R factor RCP: n Inspired by Rd in Diederichs (2006)

And RCP means. . ? How well do the measurements up to dose D agree? n Closely related to I/σ n Reasonably robust as it does not depend on sigma estimates or means n Gets bigger when systematic variation contributes to spread n

Requirements n Radiation damaged MAD data – what do I want for: ¨ Substructure determination – big anomalous / dispersive signal ¨ Phase calculation – well measured ΔF ¨ Phase extension & improvement – good F ¨ Refinement – good F n 85% Limit RCP < R(I/σ) + S(I/σ, Nm, Nu)

Example JCSG TB 0541 – heavily radiation damaged… n 3 wavelength MAD – INFL + LREM, PEAK n Massive signal n P 43212, 90 degrees * 3 => plenty of data n Chef says “use data to 1. 65 A, D=~600 s” n

Before (INFL) For TS 03/12287/INFL High resolution limit Low resolution limit Completeness Multiplicity I/sigma Rmerge Rmeas(I) Rmeas(I+/-) Rpim(I+/-) Wilson B factor Anomalous completeness Anomalous multiplicity Anomalous correlation 1. 66 52. 7 95. 8 6. 4 13. 1 0. 085 0. 117 0. 099 0. 045 0. 051 19. 372 95. 5 3. 4 0. 546 7. 41 52. 7 98. 4 5. 1 25. 6 0. 045 0. 077 0. 054 0. 032 0. 029 1. 66 1. 7 72. 5 4. 2 2. 2 0. 654 0. 808 0. 816 0. 374 0. 478 100. 0 3. 5 0. 695 72. 3 2. 1 0. 032

After (INFL – first 60 degrees) For TEST 001/12287/LREM High resolution limit Low resolution limit Completeness Multiplicity I/sigma Rmerge Rmeas(I) Rmeas(I+/-) Rpim(I+/-) Wilson B factor Anomalous completeness Anomalous multiplicity Anomalous correlation 1. 63 52. 56 92. 6 4. 1 13. 6 0. 052 0. 065 0. 066 0. 031 0. 041 18. 731 91. 8 2. 2 -0. 227 7. 3 52. 56 98. 3 3. 3 26. 2 0. 033 0. 041 0. 043 0. 021 0. 027 1. 63 1. 68 62. 9 2. 4 2. 1 0. 317 0. 504 0. 445 0. 306 0. 311 99. 4 2. 2 0. 071 59. 4 1. 3 0. 01

Why improvement? Limit radiation damage => σF more meaningful n Limit damage => ΔF better n Without systematic damage get higher resolution for given I/σ n

However… Pipe MTZ through scaleit / solve / cad / resolve / Arp/Warp and get very similar results – slight improvement though n This is most interesting, because it means that 55% of the “data” did not add to the quality of the result n

Plans Currently writing this up for J. Appl. Cryst n Chef will be included in CCP 4 6. 1 n Next: include this as part of xia 2 (makes 0. 3. 0) n Extend chef to make decisions about anomalous / dispersive differences n