JRH report on CCP4 Meeting held January 2012

[JRH]

Prof John R Helliwell report to IUCr on DDDWG relevant parts of CCP4 January 4-6th 2012

The overall theme of this year's CCP4 was diffraction Data Processing with particular relevance to proteins. Over 400 people attended. Speakers came from UK, Europe and USA. The venue was Warwick University, UK. Day zero was a 'Synchrotron beam lines for macromolecular crystallography update' with particular reference to Diamond and the EMBL part of PETRA III in Hamburg. On day 2 of the main meeting ESRF, Grenoble industrial company usage was also described.

Specific mentions of diffraction data images deposition was made in the talks of Alun Ashton and of Phil Evans. The former talk described Diamond's complete archiving of data and progress towards support for doi registration of data sets underpinning publications. The latter talk referred to diffraction images deposition as the final option (with a question mark) in a list of increased ie widened data deposition options. The new upcoming policy of the PDB of deposition of processed diffraction data being deposited unmerged was also mentioned.

Two speakers ( Phil Evans and Kay Diederichs) commended a new statistical indicator of diffraction data quality, and that Rmerge should be dropped altogether. Thus, for example, a correlation coefficient (between randomised data set halves) was put forward such as dropping to 1/2 as being the resolution limit [although it was emphasised that CC1/2 was not necessarily the “best” limit; i.e. Phil being hesitant to replace a current dogma with a new dogma. The correlation coefficient has the advantage of being a familiar, and formal, statistical coefficient that would even be familiar to other areas of science. Phil emphasised that there was good precedent for this measure, created by splitting the data in half, namely that it is an old measure in statistics, going back to at least the early years of the 20C, and it has also long been in use in electron microscopy for the estimation of “resolution”. The statistic has been available from Phil's program SCALA for a number of years, and was mentioned as a resolution criterion (though not expanded on) in the publication from a previous CCP4 Study Weekend 2010, namely in Evans Acta Cryst. D67, 282-292 (2011). Any indicator of diffraction quality should also take into account anisotropy. This was well received in discussion as an improvement over other measures such as Rmerge (which in any case show a wide range in current community practice) and additional to the widely used <I/sig(I)> crossing below 2 as the resolution limit. I supported the initiative and would support it being accepted by the PDB alongside the other new statistic that I am already pressing PDB to accept namely the Cruickshank-Blow Diffraction Precision Index.
The notion that the ultimate arbiter of a publication being the raw data itself was not championed; I found this disappointing ie as an enthusiast for adopting the policy of making the raw diffraction data images linkable to a publication by a doi.

Several presentations described data measured with the pixel array detector device. This leads not only to higher data output per unit time but also opens up improved data via fine phi slicing. This latter device leads to data sets of increased total size compared with typical CCD or image plate data. These devices increase the data archiving challenge but which could be managed in my view by one or more approaches of data triage, images compression or summing of small phi increments into coarse increments. In the talk of James Holton, and at earlier conferences, emphasis was given to the low proportion of around 5% of measured diffraction data images at macromolecular crystallography beamlines leading to publications. This is also in effect a data compression factor perhaps due to the finite time researchers have to write up results, and which are thereby the most pressing and exciting results.
The presentation of Sacha Popov described the recent updates and use of the 'BEST' program available at beamlines for deciding the best parameters for diffraction data measurement. My comment after the talk was to compliment the authors, Sacha Popov and Gleb Bourenkov, on developing 'BEST' and also to commend to them though the potential importance of a retrospective analysis of the measured diffraction images of well known challenging cases such as PSII and the difficulty of the radiation sensitivity of the Mn4CaO5 oxygen evolving complex. As Kay Deiderichs has remarked to me in email discussion subsequently:- " No-one has shown in an Acta paper that one can learn a lot from old, re-processed datasets. There are of course examples from related fields (e.g. PDB_REDO [1])". I also received an email comment on version 1 of my report from Tom Terwilliger as to "what about informal discussions about diffraction data images archiving?". I replied as follows:- "Re my discussions at CCP4 2012. I worked within the approach of not initiating that topic. In spite of much discussion activity on the CCP4bb bulletin board during 2011 it was not on everyone's lips at CCP4 2012! The opinion poll on the CCP4bb organised by Ed Pozharski [2] is the best indicator of opinion on the topic that we have at present."

Prof John R Helliwell DSc

[1] R. P. Joosten, J. Salzemann, V. Bloch, H. Stockinger, A.-C. Berglund, C. Blanchet, E. Bongcam-Rudloff, C. Combet, A. L. Da Costa, G. Deleage, M. Diarena, R. Fabbretti, G. Fettahi, V. Flegel, A. Gisel, V. Kasam, T. Kervinen, E. Korpelainen, K. Mattila, M. Pagni, M. Reichstadt, V. Breton, I. J. Tickle and G. Vriend J. Appl. Cryst. (2009). 42, 376-384 [ doi:10.1107/S0021889809008784 ]
PDB_REDO: automated re-refinement of X-ray structure models in the PDB.
[2]https://docs.google.com/spreadsheet/viewanalytics?hl=en_US&formkey=dHh4cjdLZGZrSEpUOG9kV2hkb3ZXNHc6MQ

Acknowledgements
I am grateful to Kay Diederichs, Phil Evans, James Halton, Andy Karplus and Tom Terwilliger for email discussions on my report first draft as well as to Alun Ashton and Brian McMahon for numerous earlier discussions.