Dear CCP4BB-ers, A recent discussion on the CPP4BB made it clear that people want an easy way to run paired refinement on their data. This option has always been available in the PDB-REDO webserver (https://pdb-redo.eu), but it was (un)cleverly hidden in automation (see https://journals.iucr.org/m/issues/2014/04/00/be5263/). A hack was discussed to force paired refinement, but that wasn't exactly user-friendly. So due to popular demand we now added an 'advanced option' to force paired refinement. There are now three ways to run paired refinement:
1. The recommended way: Provide higher resolution data that your model has not seen yet. I.e. first refine your model with data to a conservative cut-off that won't get you into trouble with referees, then use an extended dataset with which you want to 'push the resolution'. This triggers paired refinement automatically, starting from your conservative resolution cut-off. 2. The 'the referee made me do it' way: Use the 'advanced options' and tick 'Paired refinement' when submitting a job. In this case the starting resolution is the point where the 1% rolling average of I/sigI becomes > 2 or F/sigF > 4 (starting from the highest resolution reflection). If you have high signal-to-noise data (in which case you probably need not run paired refinement) paired refinement is done with the 10% highest resolution reflections. 3. The non-recommended hacky way: Change the resolution in REMARK 3 of your input PDB file to any value where you want to start paired refinement. The resolution gap should be at least 0.1 A. Caveat: The idea of paired refinement is to see whether the extra data has a positive effect on the refinement of you model against the 'old' data. It works best if you model is already in good shape. The PDB-REDO implementation of the original algorithm from Karplus & Diederichs (http://science.sciencemag.org/content/336/6084/1030) was designed for and tested on situation 1 which (IMO) is the least biased way of running paired refinement. Cheers, Robbie and Tassos ######################################################################## To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1
