Hi I'd have a look at the CC(1/2) in the high resolution bin - it's 0.851 (in the column 2.10 - 2.16), so there is plenty of self-agreement in that range.
Also Mean(I/sd(I)) is 2.8 in the high resolution bin, which means you have (on average) significantly strong reflections to at least 2.1Å, so you can happily integrate that far. R(meas) in the high resolution bin is acceptable at 0.44 (but Rmeas is not a good criterion for a cut-off, as has been discussed ad nauseam on this BB, so if you are going to look at it, only use it to confirm what CC(1/2) and I/sd(I) have already told you, i.e. you can safely process to higher resolution than 2.1Å ). Then I'd look at the image and see that the corners go to about 1.89Å (or a bit better) , so I'd integrate into the corners rather than use the "inscribed circle" (which is the default in iMosflm) - I notice you have already set your resolution limit on that image to about 1.87Å. Don't worry about the presence of ice rings - you can exclude them in either integration or in scaling (with an otherwise good dataset like this I would exclude them in scaling unless including them in integration had caused problems). Any modern refinement program should be able to weight your (weak) high resolution reflections properly so I wouldn't apply a manual resolution cut-off subsequently with these data. I would actually have been inclined to collect (and process) the data to higher resolution (smaller crystal to detector distance) since the data may well be usable beyond your current detector limits, and it doesn't look like you have overlaps. HTH & Merry Christmas! On 22 Dec 2014, at 12:20, sreetama das wrote: > Dear all, > I have a synchrotron data which has low outer shell and overall > completeness (table below). The dataset has ice rings (picture of frame > attached). > If it reduce the higher resolution limit during scaling (aimless), the > completeness increases, but the <I/sig(I)> in outermost shell also increases. > Besides, this would also involve throwing away of high resolution data. > In these circumstances, what criteria should be looked at to determine to > what resolution the data should be used during processing? > thanks and regards, > sreetama > Overall InnerShell OuterShell > Low resolution limit 47.75 47.75 2.16 > High resolution limit 2.10 8.91 2.10 > > Rmerge (within I+/I-) 0.058 0.029 0.338 > Rmerge (all I+ and I-) 0.062 0.031 0.370 > Rmeas (within I+/I-) 0.068 0.034 0.441 > Rmeas (all I+ & I-) 0.068 0.035 0.447 > Rpim (within I+/I-) 0.035 0.018 0.280 > Rpim (all I+ & I-) 0.027 0.014 0.242 > Rmerge in top intensity bin 0.043 - - > Total number of observations 67721 1193 1985 > Total number unique 10915 189 709 > Mean((I)/sd(I)) 18.4 35.1 2.8 > Mn(I) half-set correlation CC(1/2) 0.993 0.998 0.851 > Completeness 89.8 99.6 73.5 > Multiplicity 6.2 6.3 2.8 > > Anomalous completeness 84.5 99.6 49.7 > Anomalous multiplicity 3.1 3.8 1.7 > DelAnom correlation between half-sets -0.012 0.304 0.008 > Mid-Slope of Anom Normal Probability 0.945 - - > > Estimates of resolution limits: overall > from half-dataset correlation CC(1/2) > 0.50: limit = 2.10A == maximum > resolution > from Mn(I/sd) > 2.00: limit = 2.10A == maximum > resolution > > > > ICE RINGS: > > Possible Ice Rings > > Ice Ring Summary: > reso mean_I mean_Sigma Estimated_I Zscore Completeness Ave_Completeness > 3.89 19320.73 417.47 17504.43 4.35 0.76 1.00 > 3.67 16354.48 395.80 12788.36 9.01 0.71 1.00 > 3.44 12035.71 301.90 12214.14 -0.59 0.68 1.00 > 2.66 2358.33 143.37 2288.63 0.49 0.56 1.00 > 2.24 42046.24 1430.20 656.36 28.94 0.34 0.68 > > > <frame.png> Harry -- Dr Harry Powell, MRC Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge Biomedical Campus, Cambridge CB2 0QH Chairman of International Union of Crystallography Commission on Crystallographic Computing Chairman of European Crystallographic Association SIG9 (Crystallographic Computing)
