Re: [ccp4bb] Death of Rmerge
> Let's say you collect data (or rather indices) to 1.4 Ang but the real > resolution is 2.8 Ang and you use all the data in refinement with no > resolution cut-off, so there are 8 times as many data. Then your 15 > mins becomes 2 hours - is that still acceptable? It's unlikely that > you'll see any difference in the results so was all that extra > computing worth the effort? > > Now work out the total number of pixels in one of your datasets (i.e. > no of pixels per image times no of images). Divide that by the no of > reflections in the a.u. and multiply by 15 mins (it's probably in the > region of 400 days!): still acceptable? Again it's unlikely you'll > see any significant difference in the results (assuming you only use > the Bragg spots), so again was it worth it? > > What matters in terms of information content is not the absolute > intensity but the ratio intensity / (expected intensity). As the data > get weaker at higher d* I falls off, but so does and the ratio I / > becomes progressively more unreliable at determining the > information content. So a zero I when the other intensities in the > same d* shell are strong is indeed a powerful constraint (this I > suspect is what Wang meant), however if the other intensities in the > shell are also all zero it tells you next to nothing. > > -- Ian > I envisioned a process of iteration through the various stages of processing, so still using integration, scaling, etc. to reduce data before refinement, but maybe feeding back model-based information to inform the processing of the images. Something like, Refmac says to Mosflm: "kill frames 1100-1200: they're too radiation-damaged." But I like your idea of using all the pixels--that would be the ultimate, wouldn't it! Actually, the best would be to have the refinement already going when collecting data, and informing which frames to take, and for how long! In a couple years that too will take no time at all, but then again, we'll probably have atomic-precision real-time in vivo microscopes by then anyway, and crystallography will have become an (interesting!) historical curiosity... JPK *** Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: j-kell...@northwestern.edu ***
Re: [ccp4bb] Death of Rmerge
Hi, as we reported in our paper in Table 1 (actually Supplementary Table 1), at the end of Scaling 2, completeness in the outer shell after aniso truncation was 54%. Whilst 96% completeness and I/sigma 0.8 is of course before aniso truncation. I/sigma after truncation would be higher, but it is not clear to me how to calculate that number exactly, since aniso truncation is done post data scaling. One could of course re-process images in Mosflm with applied aniso limits and then scale data, but that would not be exactly the same. From many trials with strongly anisotropic data we found that for map calculation and refinement it is best to cut data anisotropically where F/sigma is approaching 2.5-2.7 in each direction, as long as completeness in the outer shell remains above 50% or so. Usually the highest useful resolution is also where the correlation coefficient between random half-data-set estimates of intensities in SCALA falls below about 0.5 (as advocated by Phil Evans, I think). CC seems to be less affected by anisotropy (in this case it reached 0.5 at 3.0 angstrom, which was another criterion to cut data at 3.0). HTH. Leo I am little curious about the anisotropically truncated data for 3RKO: Percent Possible(All) 96.0 Mean I Over Sigma(Observed) 0.8 In the supplementary table of the nature paper it was made clear that this 3.16-3.0A, I/sigmaI=0.8 and Rmerge=1.216 shell was the outer shell of the anisotropically truncated data. The authors did also report the isotropically truncated resolution to be 3.2A with I/sigmaI=1.3 and Rmerge=73%. The authors also stated in the main text that "the best native data set was anisotropically scaled and truncated to 3.4 Å, 3.0 Å and 3.0 Å resolution, where the F/σ ratio drops to ~2.6–2.8 along the a*, b* and c* axes, respectively (scaling 2, Supplementary Table 1)" My question is, is the I/sigmaI=0.8 a consequence of many reflections with nearly 0 I/sigmaI being included in the calculation? Then what does the 96% completeness mean? Does it mean that 96% completeness in the spherical shell of 3.16-3.0A was achieved, by including a great number of I=0 reflections? Zhijie
Re: [ccp4bb] Death of Rmerge
Let's say you collect data (or rather indices) to 1.4 Ang but the real resolution is 2.8 Ang and you use all the data in refinement with no resolution cut-off, so there are 8 times as many data. Then your 15 mins becomes 2 hours - is that still acceptable? It's unlikely that you'll see any difference in the results so was all that extra computing worth the effort? Now work out the total number of pixels in one of your datasets (i.e. no of pixels per image times no of images). Divide that by the no of reflections in the a.u. and multiply by 15 mins (it's probably in the region of 400 days!): still acceptable? Again it's unlikely you'll see any significant difference in the results (assuming you only use the Bragg spots), so again was it worth it? What matters in terms of information content is not the absolute intensity but the ratio intensity / (expected intensity). As the data get weaker at higher d* I falls off, but so does and the ratio I / becomes progressively more unreliable at determining the information content. So a zero I when the other intensities in the same d* shell are strong is indeed a powerful constraint (this I suspect is what Wang meant), however if the other intensities in the shell are also all zero it tells you next to nothing. -- Ian On 1 June 2012 20:03, Jacob Keller wrote: > I don't think any data should be discarded, and I think that although > we are not there yet, refinement should work directly with the images, > iterating back and forth through all the various levels of data > processing. As I think was pointed out by Wang, even an intensity of 0 > provides information placing limits on the possible true values of > that reflection. It seems that the main reason data were discarded > historically was because of the limitations of (under)grad students > going through multiple layers of films, evaluating intensities for > each spot, or other similar processing limits, most of which are not > really applicable today. A whole iterated refinement protocol now > takes, what, 15 minutes? > > Jacob > > > > On Fri, Jun 1, 2012 at 1:29 PM, Ed Pozharski wrote: >> http://www.nature.com/nsmb/journal/v4/n4/abs/nsb0497-269.html >> http://scripts.iucr.org/cgi-bin/paper?S0021889800018227 >> >> Just collect 360 sweep instead of 180 on a non-decaying crystal and see >> Rmerge go up due to increase in multiplicity (and enough with redundancy >> term - the extra data is not really *redundant*). Is your resolution >> worse or better? >> >> This has been argued over before. Rmerge has some value in comparing >> two datasets collected in perfectly identical conditions to see which >> crystal is better and it may predict to some extent what R-values you >> might expect. Otherwise, it's unreliable. >> >> Given that it's been 15 years since this was pointed out in no less than >> Nature group magazine, and we still hear that Rmerge should decide >> resolution cutoff, chances are increasingly slim that I will personally >> see the dethroning of that other major oppressor, R-value. >> >> On Fri, 2012-06-01 at 10:59 -0700, aaleshin wrote: >>> Please excuse my ignorance, but I cannot understand why Rmerge is >>> unreliable for estimation of the resolution? >>> I mean, from a theoretical point of view, <1/sigma> is indeed a better >>> criterion, but it is not obvious from a practical point of view. >>> >>> <1/sigma> depends on a method for sigma estimation, and so same data >>> processed by different programs may have different <1/sigma>. Moreover, >>> HKL2000 allows users to adjust sigmas manually. Rmerge estimates sigmas >>> from differences between measurements of same structural factor, and hence >>> is independent of our preferences. But, it also has a very important >>> ability to validate consistency of the merged data. If my crystal changed >>> during the data collection, or something went wrong with the >>> diffractometer, Rmerge will show it immediately, but <1/sigma> will not. >>> >>> So, please explain why should we stop using Rmerge as a criterion of data >>> resolution? >>> >>> Alex >>> Sanford-Burnham Medical Research Institute >>> 10901 North Torrey Pines Road >>> La Jolla, California 92037 >>> >>> >>> >>> On Jun 1, 2012, at 5:07 AM, Ian Tickle wrote: >>> >>> > On 1 June 2012 03:22, Edward A. Berry wrote: >>> >> Leo will probably answer better than I can, but I would say I/SigI counts >>> >> only >>> >> the present reflection, so eliminating noise by anisotropic truncation >>> >> should >>> >> improve it, raising the average I/SigI in the last shell. >>> > >>> > We always include unmeasured reflections with I/sigma(I) = 0 in the >>> > calculation of the mean I/sigma(I) (i.e. we divide the sum of >>> > I/sigma(I) for measureds by the predicted total no of reflections incl >>> > unmeasureds), since for unmeasureds I is (almost) completely unknown >>> > and therefore sigma(I) is effectively infinite (or at least finite but >>> > large since you do have some idea of
Re: [ccp4bb] Death of Rmerge
I don't think any data should be discarded, and I think that although we are not there yet, refinement should work directly with the images, iterating back and forth through all the various levels of data processing. As I think was pointed out by Wang, even an intensity of 0 provides information placing limits on the possible true values of that reflection. It seems that the main reason data were discarded historically was because of the limitations of (under)grad students going through multiple layers of films, evaluating intensities for each spot, or other similar processing limits, most of which are not really applicable today. A whole iterated refinement protocol now takes, what, 15 minutes? Jacob On Fri, Jun 1, 2012 at 1:29 PM, Ed Pozharski wrote: > http://www.nature.com/nsmb/journal/v4/n4/abs/nsb0497-269.html > http://scripts.iucr.org/cgi-bin/paper?S0021889800018227 > > Just collect 360 sweep instead of 180 on a non-decaying crystal and see > Rmerge go up due to increase in multiplicity (and enough with redundancy > term - the extra data is not really *redundant*). Is your resolution > worse or better? > > This has been argued over before. Rmerge has some value in comparing > two datasets collected in perfectly identical conditions to see which > crystal is better and it may predict to some extent what R-values you > might expect. Otherwise, it's unreliable. > > Given that it's been 15 years since this was pointed out in no less than > Nature group magazine, and we still hear that Rmerge should decide > resolution cutoff, chances are increasingly slim that I will personally > see the dethroning of that other major oppressor, R-value. > > On Fri, 2012-06-01 at 10:59 -0700, aaleshin wrote: >> Please excuse my ignorance, but I cannot understand why Rmerge is unreliable >> for estimation of the resolution? >> I mean, from a theoretical point of view, <1/sigma> is indeed a better >> criterion, but it is not obvious from a practical point of view. >> >> <1/sigma> depends on a method for sigma estimation, and so same data >> processed by different programs may have different <1/sigma>. Moreover, >> HKL2000 allows users to adjust sigmas manually. Rmerge estimates sigmas from >> differences between measurements of same structural factor, and hence is >> independent of our preferences. But, it also has a very important ability >> to validate consistency of the merged data. If my crystal changed during the >> data collection, or something went wrong with the diffractometer, Rmerge >> will show it immediately, but <1/sigma> will not. >> >> So, please explain why should we stop using Rmerge as a criterion of data >> resolution? >> >> Alex >> Sanford-Burnham Medical Research Institute >> 10901 North Torrey Pines Road >> La Jolla, California 92037 >> >> >> >> On Jun 1, 2012, at 5:07 AM, Ian Tickle wrote: >> >> > On 1 June 2012 03:22, Edward A. Berry wrote: >> >> Leo will probably answer better than I can, but I would say I/SigI counts >> >> only >> >> the present reflection, so eliminating noise by anisotropic truncation >> >> should >> >> improve it, raising the average I/SigI in the last shell. >> > >> > We always include unmeasured reflections with I/sigma(I) = 0 in the >> > calculation of the mean I/sigma(I) (i.e. we divide the sum of >> > I/sigma(I) for measureds by the predicted total no of reflections incl >> > unmeasureds), since for unmeasureds I is (almost) completely unknown >> > and therefore sigma(I) is effectively infinite (or at least finite but >> > large since you do have some idea of what range I must fall in). A >> > shell with = 2 and 50% completeness clearly doesn't carry >> > the same information content as one with the same and >> > 100% complete; therefore IMO it's very misleading to quote >> > including only the measured reflections. This also means >> > we can use a single cut-off criterion (we use mean I/sigma(I) > 1), >> > and we don't need another arbitrary cut-off criterion for >> > completeness. As many others seem to be doing now, we don't use >> > Rmerge, Rpim etc as criteria to estimate resolution, they're just too >> > unreliable - Rmerge is indeed dead and buried! >> > >> > Actually a mean value of I/sigma(I) of 2 is highly statistically >> > significant, i.e. very unlikely to have arisen by chance variations, >> > and the significance threshold for the mean must be much closer to 1 >> > than to 2. Taking an average always increases the statistical >> > significance, therefore it's not valid to compare an _average_ value >> > of I/sigma(I) = 2 with a _single_ value of I/sigma(I) = 3 (taking 3 >> > sigma as the threshold of statistical significance of an individual >> > measurement): that's a case of "comparing apples with pears". In >> > other words in the outer shell you would need a lot of highly >> > significant individual values >> 3 to attain an overall average of 2 >> > since the majority of individual values will be < 1. >> > >> >> F/sigF is ex
Re: [ccp4bb] Death of Rmerge
http://www.nature.com/nsmb/journal/v4/n4/abs/nsb0497-269.html http://scripts.iucr.org/cgi-bin/paper?S0021889800018227 Just collect 360 sweep instead of 180 on a non-decaying crystal and see Rmerge go up due to increase in multiplicity (and enough with redundancy term - the extra data is not really *redundant*). Is your resolution worse or better? This has been argued over before. Rmerge has some value in comparing two datasets collected in perfectly identical conditions to see which crystal is better and it may predict to some extent what R-values you might expect. Otherwise, it's unreliable. Given that it's been 15 years since this was pointed out in no less than Nature group magazine, and we still hear that Rmerge should decide resolution cutoff, chances are increasingly slim that I will personally see the dethroning of that other major oppressor, R-value. On Fri, 2012-06-01 at 10:59 -0700, aaleshin wrote: > Please excuse my ignorance, but I cannot understand why Rmerge is unreliable > for estimation of the resolution? > I mean, from a theoretical point of view, <1/sigma> is indeed a better > criterion, but it is not obvious from a practical point of view. > > <1/sigma> depends on a method for sigma estimation, and so same data > processed by different programs may have different <1/sigma>. Moreover, > HKL2000 allows users to adjust sigmas manually. Rmerge estimates sigmas from > differences between measurements of same structural factor, and hence is > independent of our preferences. But, it also has a very important ability to > validate consistency of the merged data. If my crystal changed during the > data collection, or something went wrong with the diffractometer, Rmerge will > show it immediately, but <1/sigma> will not. > > So, please explain why should we stop using Rmerge as a criterion of data > resolution? > > Alex > Sanford-Burnham Medical Research Institute > 10901 North Torrey Pines Road > La Jolla, California 92037 > > > > On Jun 1, 2012, at 5:07 AM, Ian Tickle wrote: > > > On 1 June 2012 03:22, Edward A. Berry wrote: > >> Leo will probably answer better than I can, but I would say I/SigI counts > >> only > >> the present reflection, so eliminating noise by anisotropic truncation > >> should > >> improve it, raising the average I/SigI in the last shell. > > > > We always include unmeasured reflections with I/sigma(I) = 0 in the > > calculation of the mean I/sigma(I) (i.e. we divide the sum of > > I/sigma(I) for measureds by the predicted total no of reflections incl > > unmeasureds), since for unmeasureds I is (almost) completely unknown > > and therefore sigma(I) is effectively infinite (or at least finite but > > large since you do have some idea of what range I must fall in). A > > shell with = 2 and 50% completeness clearly doesn't carry > > the same information content as one with the same and > > 100% complete; therefore IMO it's very misleading to quote > > including only the measured reflections. This also means > > we can use a single cut-off criterion (we use mean I/sigma(I) > 1), > > and we don't need another arbitrary cut-off criterion for > > completeness. As many others seem to be doing now, we don't use > > Rmerge, Rpim etc as criteria to estimate resolution, they're just too > > unreliable - Rmerge is indeed dead and buried! > > > > Actually a mean value of I/sigma(I) of 2 is highly statistically > > significant, i.e. very unlikely to have arisen by chance variations, > > and the significance threshold for the mean must be much closer to 1 > > than to 2. Taking an average always increases the statistical > > significance, therefore it's not valid to compare an _average_ value > > of I/sigma(I) = 2 with a _single_ value of I/sigma(I) = 3 (taking 3 > > sigma as the threshold of statistical significance of an individual > > measurement): that's a case of "comparing apples with pears". In > > other words in the outer shell you would need a lot of highly > > significant individual values >> 3 to attain an overall average of 2 > > since the majority of individual values will be < 1. > > > >> F/sigF is expected to be better than I/sigI because dx^2 = 2Xdx, > >> dx^2/x^2 = 2dx/x, dI/I = 2* dF/F (or approaches that in the limit . . .) > > > > That depends on what you mean by 'better': every metric must be > > compared with a criterion appropriate to that metric. So if we are > > comparing I/sigma(I) with a criterion value = 3, then we must compare > > F/sigma(F) with criterion value = 6 ('in the limit' of zero I), in > > which case the comparison is no 'better' (in terms of information > > content) with I than with F: they are entirely equivalent. It's > > meaningless to compare F/sigma(F) with the criterion value appropriate > > to I/sigma(I): again that's "comparing apples and pears"! > > > > Cheers > > > > -- Ian -- Edwin Pozharski, PhD, Assistant Professor University of Maryland, Baltimore ---
Re: [ccp4bb] Death of Rmerge
As the K & D paper points out, as the signal/noise declines at higher resolution, Rmerge goes up to infinity, so there is no sensible way to set a limiting value to determine "resolution". That is not to say that Rmerge has no use: as you say it's a reasonably good metric to plot against image number to detect a problem. It just not a suitable metric for deciding resolution I/sigI is pretty good for this, even though the sigma estimates are not very reliable. CC1/2 is probably better since it is independent of sigmas and has defined values from 1.0 down to 0.0 as signal/noise decreases. But we should be careful of any dogma which says what data we should discard, and what the cutoff limits should be: I/sigI > 3,2, or 1? CC1/2 > 0.2, 0.3, 0.5 ...? Usually it does not make a huge difference, but why discard useful data? Provided the data are properly weighted in refinement by weights incorporating observed sigmas (true in Refmac, not true in phenix.refine at present I believe), adding extra weak data should do no harm, at least out to some point. Program algorithms are improving in their treatment of weak data, but are by no means perfect. One problem as discussed earlier in this thread is that we have got used to the idea that nominal resolution is a single number indicating the quality of a structure, but this has never been true, irrespective of the cutoff method. Apart from the considerable problem of anisotropy, we all need to note the wisdom of Ethan Merritt > "We should also encourage people not to confuse the quality of > the data with the quality of the model." Phil On 1 Jun 2012, at 18:59, aaleshin wrote: > Please excuse my ignorance, but I cannot understand why Rmerge is unreliable > for estimation of the resolution? > I mean, from a theoretical point of view, <1/sigma> is indeed a better > criterion, but it is not obvious from a practical point of view. > > <1/sigma> depends on a method for sigma estimation, and so same data > processed by different programs may have different <1/sigma>. Moreover, > HKL2000 allows users to adjust sigmas manually. Rmerge estimates sigmas from > differences between measurements of same structural factor, and hence is > independent of our preferences. But, it also has a very important ability to > validate consistency of the merged data. If my crystal changed during the > data collection, or something went wrong with the diffractometer, Rmerge will > show it immediately, but <1/sigma> will not. > > So, please explain why should we stop using Rmerge as a criterion of data > resolution? > > Alex > Sanford-Burnham Medical Research Institute > 10901 North Torrey Pines Road > La Jolla, California 92037 > > > > On Jun 1, 2012, at 5:07 AM, Ian Tickle wrote: > >> On 1 June 2012 03:22, Edward A. Berry wrote: >>> Leo will probably answer better than I can, but I would say I/SigI counts >>> only >>> the present reflection, so eliminating noise by anisotropic truncation >>> should >>> improve it, raising the average I/SigI in the last shell. >> >> We always include unmeasured reflections with I/sigma(I) = 0 in the >> calculation of the mean I/sigma(I) (i.e. we divide the sum of >> I/sigma(I) for measureds by the predicted total no of reflections incl >> unmeasureds), since for unmeasureds I is (almost) completely unknown >> and therefore sigma(I) is effectively infinite (or at least finite but >> large since you do have some idea of what range I must fall in). A >> shell with = 2 and 50% completeness clearly doesn't carry >> the same information content as one with the same and >> 100% complete; therefore IMO it's very misleading to quote >> including only the measured reflections. This also means >> we can use a single cut-off criterion (we use mean I/sigma(I) > 1), >> and we don't need another arbitrary cut-off criterion for >> completeness. As many others seem to be doing now, we don't use >> Rmerge, Rpim etc as criteria to estimate resolution, they're just too >> unreliable - Rmerge is indeed dead and buried! >> >> Actually a mean value of I/sigma(I) of 2 is highly statistically >> significant, i.e. very unlikely to have arisen by chance variations, >> and the significance threshold for the mean must be much closer to 1 >> than to 2. Taking an average always increases the statistical >> significance, therefore it's not valid to compare an _average_ value >> of I/sigma(I) = 2 with a _single_ value of I/sigma(I) = 3 (taking 3 >> sigma as the threshold of statistical significance of an individual >> measurement): that's a case of "comparing apples with pears". In >> other words in the outer shell you would need a lot of highly >> significant individual values >> 3 to attain an overall average of 2 >> since the majority of individual values will be < 1. >> >>> F/sigF is expected to be better than I/sigI because dx^2 = 2Xdx, >>> dx^2/x^2 = 2dx/x, dI/I = 2* dF/F (or approaches that in the limit . . .) >> >> That depen
Re: [ccp4bb] Death of Rmerge
Please excuse my ignorance, but I cannot understand why Rmerge is unreliable for estimation of the resolution? I mean, from a theoretical point of view, <1/sigma> is indeed a better criterion, but it is not obvious from a practical point of view. <1/sigma> depends on a method for sigma estimation, and so same data processed by different programs may have different <1/sigma>. Moreover, HKL2000 allows users to adjust sigmas manually. Rmerge estimates sigmas from differences between measurements of same structural factor, and hence is independent of our preferences. But, it also has a very important ability to validate consistency of the merged data. If my crystal changed during the data collection, or something went wrong with the diffractometer, Rmerge will show it immediately, but <1/sigma> will not. So, please explain why should we stop using Rmerge as a criterion of data resolution? Alex Sanford-Burnham Medical Research Institute 10901 North Torrey Pines Road La Jolla, California 92037 On Jun 1, 2012, at 5:07 AM, Ian Tickle wrote: > On 1 June 2012 03:22, Edward A. Berry wrote: >> Leo will probably answer better than I can, but I would say I/SigI counts >> only >> the present reflection, so eliminating noise by anisotropic truncation >> should >> improve it, raising the average I/SigI in the last shell. > > We always include unmeasured reflections with I/sigma(I) = 0 in the > calculation of the mean I/sigma(I) (i.e. we divide the sum of > I/sigma(I) for measureds by the predicted total no of reflections incl > unmeasureds), since for unmeasureds I is (almost) completely unknown > and therefore sigma(I) is effectively infinite (or at least finite but > large since you do have some idea of what range I must fall in). A > shell with = 2 and 50% completeness clearly doesn't carry > the same information content as one with the same and > 100% complete; therefore IMO it's very misleading to quote > including only the measured reflections. This also means > we can use a single cut-off criterion (we use mean I/sigma(I) > 1), > and we don't need another arbitrary cut-off criterion for > completeness. As many others seem to be doing now, we don't use > Rmerge, Rpim etc as criteria to estimate resolution, they're just too > unreliable - Rmerge is indeed dead and buried! > > Actually a mean value of I/sigma(I) of 2 is highly statistically > significant, i.e. very unlikely to have arisen by chance variations, > and the significance threshold for the mean must be much closer to 1 > than to 2. Taking an average always increases the statistical > significance, therefore it's not valid to compare an _average_ value > of I/sigma(I) = 2 with a _single_ value of I/sigma(I) = 3 (taking 3 > sigma as the threshold of statistical significance of an individual > measurement): that's a case of "comparing apples with pears". In > other words in the outer shell you would need a lot of highly > significant individual values >> 3 to attain an overall average of 2 > since the majority of individual values will be < 1. > >> F/sigF is expected to be better than I/sigI because dx^2 = 2Xdx, >> dx^2/x^2 = 2dx/x, dI/I = 2* dF/F (or approaches that in the limit . . .) > > That depends on what you mean by 'better': every metric must be > compared with a criterion appropriate to that metric. So if we are > comparing I/sigma(I) with a criterion value = 3, then we must compare > F/sigma(F) with criterion value = 6 ('in the limit' of zero I), in > which case the comparison is no 'better' (in terms of information > content) with I than with F: they are entirely equivalent. It's > meaningless to compare F/sigma(F) with the criterion value appropriate > to I/sigma(I): again that's "comparing apples and pears"! > > Cheers > > -- Ian
Re: [ccp4bb] Death of Rmerge
On 1 June 2012 03:22, Edward A. Berry wrote: > Leo will probably answer better than I can, but I would say I/SigI counts > only > the present reflection, so eliminating noise by anisotropic truncation > should > improve it, raising the average I/SigI in the last shell. We always include unmeasured reflections with I/sigma(I) = 0 in the calculation of the mean I/sigma(I) (i.e. we divide the sum of I/sigma(I) for measureds by the predicted total no of reflections incl unmeasureds), since for unmeasureds I is (almost) completely unknown and therefore sigma(I) is effectively infinite (or at least finite but large since you do have some idea of what range I must fall in). A shell with = 2 and 50% completeness clearly doesn't carry the same information content as one with the same and 100% complete; therefore IMO it's very misleading to quote including only the measured reflections. This also means we can use a single cut-off criterion (we use mean I/sigma(I) > 1), and we don't need another arbitrary cut-off criterion for completeness. As many others seem to be doing now, we don't use Rmerge, Rpim etc as criteria to estimate resolution, they're just too unreliable - Rmerge is indeed dead and buried! Actually a mean value of I/sigma(I) of 2 is highly statistically significant, i.e. very unlikely to have arisen by chance variations, and the significance threshold for the mean must be much closer to 1 than to 2. Taking an average always increases the statistical significance, therefore it's not valid to compare an _average_ value of I/sigma(I) = 2 with a _single_ value of I/sigma(I) = 3 (taking 3 sigma as the threshold of statistical significance of an individual measurement): that's a case of "comparing apples with pears". In other words in the outer shell you would need a lot of highly significant individual values >> 3 to attain an overall average of 2 since the majority of individual values will be < 1. > F/sigF is expected to be better than I/sigI because dx^2 = 2Xdx, > dx^2/x^2 = 2dx/x, dI/I = 2* dF/F (or approaches that in the limit . . .) That depends on what you mean by 'better': every metric must be compared with a criterion appropriate to that metric. So if we are comparing I/sigma(I) with a criterion value = 3, then we must compare F/sigma(F) with criterion value = 6 ('in the limit' of zero I), in which case the comparison is no 'better' (in terms of information content) with I than with F: they are entirely equivalent. It's meaningless to compare F/sigma(F) with the criterion value appropriate to I/sigma(I): again that's "comparing apples and pears"! Cheers -- Ian
Re: [ccp4bb] Death of Rmerge
Leo will probably answer better than I can, but I would say I/SigI counts only the present reflection, so eliminating noise by anisotropic truncation should improve it, raising the average I/SigI in the last shell. F/sigF is expected to be better than I/sigI because dx^2 = 2Xdx, dx^2/x^2 = 2dx/x, dI/I = 2* dF/F (or approaches that in the limit . . .) On the other hand the integration software will measure spots whether they exist or not, so completeness is good even in a shell where there is no data. Anisotropic truncation removes reflections, so now calculating completeness in the outer (spherical) shell gives low completeness. In the direction where the data was truncated to 3.4 A, there are obviously no reflections in the 3.0-3.1 (e.g.) shell. Read about anisotropic truncation and scaling at the server they used: http://services.mbi.ucla.edu/anisoscale/ I think Eleanor Dodson once suggested anisotropic truncation could be preformed by a script which gets the anisotropy from the fall-off analysis in truncate, changes the cell parameters to phony ones which distort reciprocal space until falloff is the same in all directions, perform (spherical) resolution cut-off, and change the cell parameters back to the correct. I think all the refinement programs can perform anisotropic scaling, but they normally don't save the scaled data. Zhijie Li wrote: I am little curious about the anisotropically truncated data for 3RKO: Percent Possible(All) 96.0 Mean I Over Sigma(Observed) 0.8 In the supplementary table of the nature paper it was made clear that this 3.16-3.0A, I/sigmaI=0.8 and Rmerge=1.216 shell was the outer shell of the anisotropically truncated data. The authors did also report the isotropically truncated resolution to be 3.2A with I/sigmaI=1.3 and Rmerge=73%. The authors also stated in the main text that "the best native data set was anisotropically scaled and truncated to 3.4 Å, 3.0 Å and 3.0 Å resolution, where the F/σ ratio drops to ~2.6–2.8 along the a*, b* and c* axes, respectively (scaling 2, Supplementary Table 1)" My question is, is the I/sigmaI=0.8 a consequence of many reflections with nearly 0 I/sigmaI being included in the calculation? Then what does the 96% completeness mean? Does it mean that 96% completeness in the spherical shell of 3.16-3.0A was achieved, by including a great number of I=0 reflections? Zhijie -- From: "Edward A. Berry" Sent: Thursday, May 31, 2012 2:59 PM To: Subject: Re: [ccp4bb] Death of Rmerge Yes! I want a copy of this program RESCUT. REMARK 200 R SYM FOR SHELL (I) : 1.21700 I noticed structure 3RKO reported Rmerge in the last shell greater than 1, suggesting the police who were defending R-merge were fighting a losing battle. And this provides a lot of ammunition to those they are fighting. Jacob Keller wrote: Dear Crystallographers, in case you have not heard, it would appear that the Rmerge statistic has died as of the publication of PMID: 22628654. Ding Dong...? JPK -- *** Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: j-kell...@northwestern.edu ***
Re: [ccp4bb] Death of Rmerge
I am little curious about the anisotropically truncated data for 3RKO: Percent Possible(All)96.0 Mean I Over Sigma(Observed) 0.8 In the supplementary table of the nature paper it was made clear that this 3.16-3.0A, I/sigmaI=0.8 and Rmerge=1.216 shell was the outer shell of the anisotropically truncated data. The authors did also report the isotropically truncated resolution to be 3.2A with I/sigmaI=1.3 and Rmerge=73%. The authors also stated in the main text that "the best native data set was anisotropically scaled and truncated to 3.4 Å, 3.0 Å and 3.0 Å resolution, where the F/σ ratio drops to ~2.6–2.8 along the a*, b* and c* axes, respectively (scaling 2, Supplementary Table 1)" My question is, is the I/sigmaI=0.8 a consequence of many reflections with nearly 0 I/sigmaI being included in the calculation? Then what does the 96% completeness mean? Does it mean that 96% completeness in the spherical shell of 3.16-3.0A was achieved, by including a great number of I=0 reflections? Zhijie -- From: "Edward A. Berry" Sent: Thursday, May 31, 2012 2:59 PM To: Subject: Re: [ccp4bb] Death of Rmerge Yes! I want a copy of this program RESCUT. REMARK 200 R SYM FOR SHELL(I) : 1.21700 I noticed structure 3RKO reported Rmerge in the last shell greater than 1, suggesting the police who were defending R-merge were fighting a losing battle. And this provides a lot of ammunition to those they are fighting. Jacob Keller wrote: Dear Crystallographers, in case you have not heard, it would appear that the Rmerge statistic has died as of the publication of PMID: 22628654. Ding Dong...? JPK -- *** Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: j-kell...@northwestern.edu ***
Re: [ccp4bb] Fwd: [ccp4bb] Death of Rmerge
In the meantime we could follow Phoebe Rice's example and put the resolution at I/sigma=2 in REMARK 2 "resolution of structure" but put the actual bleeding-edge resolution we used in the reduction and refinement statistics (At least if the PDB will allow us to have different values in these three places) And cite the REM 2 value in the article. eab PS - I believe optical resolution actually gives significantly more optimistic numbers for resolution than extending from I/sigma=2 to 0.5. The "resolution" we are used to is the d-spacing of the Fourier components, which is theoretically larger than the microscopist's definition of resolution (how close objects can be and still give separate maxima with a minimum between). Miller, Mitchell D. wrote: All three numbers (high resolution limit in remark 2, remark 3 and Remark 200) are supposed to be consistent and are defined as the highest resolution reflection used. http://mmcif.rcsb.org/dictionaries/mmcif_pdbx_v40.dic/Items/_reflns.d_resolution_high.html http://mmcif.rcsb.org/dictionaries/mmcif_pdbx_v40.dic/Items/_refine.ls_d_res_high.html Looking at the PDB specification, it shows that there is an option to add a free text comment to the remark 2 resolution -- "Additional explanatory text may be included starting with the third line of the REMARK 2 record. For example, depositors may wish to qualify the resolution value provided due to unusual experimental conditions." http://www.wwpdb.org/documentation/format33/remarks1.html On Wed, Apr 25, 2012 at 12:23 AM, Phoebe Rice wrote: I just noticed that the PDB has changed the stated resolution for one of my old structures! It was refined against a very anisotropic data set that extended to 2.2 in the best direction only. When depositing I called the resolution 2.5 as a rough average of resolution in all 3 directions, but now PDB is advertising it as 2.2, which is misleading. I'm afraid I may not have paid enough attention to the fine print on this issue - is the PDB now automatically advertising the "resolution" of a structure as that of the outermost flyspeck used in refinement, regardless of more cautious assertions by the authors? If so, I object! Dale Tronrud wrote: On 05/31/12 12:07, Jacob Keller wrote: Alas, how many lines like the following from a recent Science paper (PMID: 22605777), probably reviewer-incited, could have been avoided! "Here, we present three high-resolution crystal structures of the Thermus thermophilus (Tth) 70S ribosome in complex withRMF, HPF, or YfiA that were refined by using data extending to 3.0 Å (I/sI = 1), 3.1 Å (I/sI = 1), and 2.75 Å (I/sI = 1) resolution, respectively. The resolutions at which I/sI = 2 are 3.2 Å, 3.4 Å, and 2.9 Å, respectively." I don't see how you can avoid something like this. With the new, higher, resolution limits for data (which are good things) people will tend to assume that a "2.6 A resolution model" will have roughly the same quality as a "2.6 A resolution model" from five years ago when the old criteria were used. K&K show that the weak high resolution data contain useful information but certainly not as much information as the data with stronger intensity. The resolution limit of the data set has been such an important indicator of the quality of the resulting model (rightly or wrongly) that it often is included in the title of the paper itself. Despite the fact that we now want to include more, weak, data than before we need to continue to have a quality indicator that readers can use to assess the models they are reading about. While cumbersome, one solution is to state what the resolution limit would have been had the old criteria been used, as was done in the paper you quote. This simply gives the reader a measure they can compare to their previous experiences. Now would be a good time to break with tradition and institute a new measure of quality of diffraction data sets. I believe several have been proposed over the years, but have simply not caught on. SFCHECK produces an "optical resolution". Could this be used in the title of papers? I don't believe it is sensitive to the cutoff resolution and it produces values that are consistent with what the readers are used to. With this solution people could include whatever noisy data they want and not be guilty of overstating the quality of their model. Dale Tronrud JPK On Thu, May 31, 2012 at 1:59 PM, Edward A. Berry wrote: Yes! I want a copy of this program RESCUT. REMARK 200 R SYM FOR SHELL(I) : 1.21700 I noticed structure 3RKO reported Rmerge in the last shell greater than 1, suggesting the police who were defending R-merge were fighting a losing battle. And this provides a lot of ammunition to those they are fighting. Jacob Keller wrote: Dear Crystallographers, in case you have not heard, it would appear that the Rmerge statistic has died as of the publication of PMID: 22628654. Ding Dong...? JPK --
Re: [ccp4bb] Fwd: [ccp4bb] Death of Rmerge
> There are things you can expect to learn from a > 2Å structure that you are unlikely to learn from a 5Å structure, even > if equal care has been given to both experiments, so it makes sense > for the title to give the potential reader an idea which of the two > cases is presented. But for this purpose it isn't going to matter > whether "2Å" is really 1.8Å or 2.2Å. What should the title say when a crystal diffracts to, lets say, 3 A in one direction and 4-5 A in others? Alex Aleshin, Sanford-Burnham Medical Research Institute 10901 North Torrey Pines Road La Jolla, California 92037 On May 31, 2012, at 2:50 PM, Ethan Merritt wrote: > On Thursday, May 31, 2012 02:21:45 pm Dale Tronrud wrote: >> The resolution limit of the data set has been such an important >> indicator of the quality of the resulting model (rightly or wrongly) >> that it often is included in the title of the paper itself. Despite >> the fact that we now want to include more, weak, data than before >> we need to continue to have a quality indicator that readers can >> use to assess the models they are reading about. While cumbersome, >> one solution is to state what the resolution limit would have been >> had the old criteria been used, as was done in the paper you quote. >> This simply gives the reader a measure they can compare to their >> previous experiences. > > [\me dons flame suit] > > To the extent that reporting the resolution is simply a stand-in > for reporting the quality of the model, we would do better to cut > to the chase. For instance, if you map the Molprobity green/yellow/red > model quality scoring onto good/mediocre/poor then you can title > your paper > > Crystal Structure of Fabulous Protein Foo at Mediocre Quality > > [\me removes flame suit from back, and tongue from cheek] > > > More seriously, I don't think it's entirely true that the resolution > is reported as an indicator of quality in the sense that the model > is well-refined. There are things you can expect to learn from a > 2Å structure that you are unlikely to learn from a 5Å structure, even > if equal care has been given to both experiments, so it makes sense > for the title to give the potential reader an idea which of the two > cases is presented. But for this purpose it isn't going to matter > whether "2Å" is really 1.8Å or 2.2Å. > >> Now would be a good time to break with tradition and institute >> a new measure of quality of diffraction data sets. I believe several >> have been proposed over the years, but have simply not caught on. >> SFCHECK produces an "optical resolution". Could this be used in >> the title of papers? I don't believe it is sensitive to the cutoff >> resolution and it produces values that are consistent with what the >> readers are used to. With this solution people could include whatever >> noisy data they want and not be guilty of overstating the quality of >> their model. > > We should also encourage people not to confuse the quality of > the data with the quality of the model. > > Ethan > > -- > Ethan A Merritt > Biomolecular Structure Center, K-428 Health Sciences Bldg > University of Washington, Seattle 98195-7742
Re: [ccp4bb] Fwd: [ccp4bb] Death of Rmerge
On Thursday, May 31, 2012 02:21:45 pm Dale Tronrud wrote: >The resolution limit of the data set has been such an important > indicator of the quality of the resulting model (rightly or wrongly) > that it often is included in the title of the paper itself. Despite > the fact that we now want to include more, weak, data than before > we need to continue to have a quality indicator that readers can > use to assess the models they are reading about. While cumbersome, > one solution is to state what the resolution limit would have been > had the old criteria been used, as was done in the paper you quote. > This simply gives the reader a measure they can compare to their > previous experiences. [\me dons flame suit] To the extent that reporting the resolution is simply a stand-in for reporting the quality of the model, we would do better to cut to the chase. For instance, if you map the Molprobity green/yellow/red model quality scoring onto good/mediocre/poor then you can title your paper Crystal Structure of Fabulous Protein Foo at Mediocre Quality [\me removes flame suit from back, and tongue from cheek] More seriously, I don't think it's entirely true that the resolution is reported as an indicator of quality in the sense that the model is well-refined. There are things you can expect to learn from a 2Å structure that you are unlikely to learn from a 5Å structure, even if equal care has been given to both experiments, so it makes sense for the title to give the potential reader an idea which of the two cases is presented. But for this purpose it isn't going to matter whether "2Å" is really 1.8Å or 2.2Å. >Now would be a good time to break with tradition and institute > a new measure of quality of diffraction data sets. I believe several > have been proposed over the years, but have simply not caught on. > SFCHECK produces an "optical resolution". Could this be used in > the title of papers? I don't believe it is sensitive to the cutoff > resolution and it produces values that are consistent with what the > readers are used to. With this solution people could include whatever > noisy data they want and not be guilty of overstating the quality of > their model. We should also encourage people not to confuse the quality of the data with the quality of the model. Ethan -- Ethan A Merritt Biomolecular Structure Center, K-428 Health Sciences Bldg University of Washington, Seattle 98195-7742
Re: [ccp4bb] Fwd: [ccp4bb] Death of Rmerge
Good idea, but how to get it to catch on without publishing in Science? JPK On Thu, May 31, 2012 at 4:21 PM, Dale Tronrud wrote: > > On 05/31/12 12:07, Jacob Keller wrote: >> Alas, how many lines like the following from a recent Science paper >> (PMID: 22605777), probably reviewer-incited, could have been avoided! >> >> "Here, we present three high-resolution crystal structures of the >> Thermus thermophilus (Tth) 70S ribosome in complex withRMF, HPF, or >> YfiA that were refined by using data extending to 3.0 Å (I/sI = 1), >> 3.1 Å (I/sI = 1), and 2.75 Å (I/sI = 1) resolution, respectively. The >> resolutions at which I/sI = 2 are 3.2 Å, 3.4 Å, and 2.9 Å, >> respectively." >> > > I don't see how you can avoid something like this. With the new, > higher, resolution limits for data (which are good things) people will > tend to assume that a "2.6 A resolution model" will have roughly the > same quality as a "2.6 A resolution model" from five years ago when > the old criteria were used. K&K show that the weak high resolution > data contain useful information but certainly not as much information > as the data with stronger intensity. > > The resolution limit of the data set has been such an important > indicator of the quality of the resulting model (rightly or wrongly) > that it often is included in the title of the paper itself. Despite > the fact that we now want to include more, weak, data than before > we need to continue to have a quality indicator that readers can > use to assess the models they are reading about. While cumbersome, > one solution is to state what the resolution limit would have been > had the old criteria been used, as was done in the paper you quote. > This simply gives the reader a measure they can compare to their > previous experiences. > > Now would be a good time to break with tradition and institute > a new measure of quality of diffraction data sets. I believe several > have been proposed over the years, but have simply not caught on. > SFCHECK produces an "optical resolution". Could this be used in > the title of papers? I don't believe it is sensitive to the cutoff > resolution and it produces values that are consistent with what the > readers are used to. With this solution people could include whatever > noisy data they want and not be guilty of overstating the quality of > their model. > > Dale Tronrud > >> JPK >> >> >> >> On Thu, May 31, 2012 at 1:59 PM, Edward A. Berry wrote: >>> Yes! I want a copy of this program RESCUT. >>> >>> REMARK 200 R SYM FOR SHELL (I) : 1.21700 >>> I noticed structure 3RKO reported Rmerge in the last shell greater >>> than 1, suggesting the police who were defending R-merge were fighting >>> a losing battle. And this provides a lot of ammunition to those >>> they are fighting. >>> >>> Jacob Keller wrote: Dear Crystallographers, in case you have not heard, it would appear that the Rmerge statistic has died as of the publication of PMID: 22628654. Ding Dong...? JPK -- *** Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: j-kell...@northwestern.edu *** >>> >> >> >> >> -- >> *** >> Jacob Pearson Keller >> Northwestern University >> Medical Scientist Training Program >> email: j-kell...@northwestern.edu >> *** >> >> -- *** Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: j-kell...@northwestern.edu ***
Re: [ccp4bb] Fwd: [ccp4bb] Death of Rmerge
On 05/31/12 12:07, Jacob Keller wrote: > Alas, how many lines like the following from a recent Science paper > (PMID: 22605777), probably reviewer-incited, could have been avoided! > > "Here, we present three high-resolution crystal structures of the > Thermus thermophilus (Tth) 70S ribosome in complex withRMF, HPF, or > YfiA that were refined by using data extending to 3.0 Å (I/sI = 1), > 3.1 Å (I/sI = 1), and 2.75 Å (I/sI = 1) resolution, respectively. The > resolutions at which I/sI = 2 are 3.2 Å, 3.4 Å, and 2.9 Å, > respectively." > I don't see how you can avoid something like this. With the new, higher, resolution limits for data (which are good things) people will tend to assume that a "2.6 A resolution model" will have roughly the same quality as a "2.6 A resolution model" from five years ago when the old criteria were used. K&K show that the weak high resolution data contain useful information but certainly not as much information as the data with stronger intensity. The resolution limit of the data set has been such an important indicator of the quality of the resulting model (rightly or wrongly) that it often is included in the title of the paper itself. Despite the fact that we now want to include more, weak, data than before we need to continue to have a quality indicator that readers can use to assess the models they are reading about. While cumbersome, one solution is to state what the resolution limit would have been had the old criteria been used, as was done in the paper you quote. This simply gives the reader a measure they can compare to their previous experiences. Now would be a good time to break with tradition and institute a new measure of quality of diffraction data sets. I believe several have been proposed over the years, but have simply not caught on. SFCHECK produces an "optical resolution". Could this be used in the title of papers? I don't believe it is sensitive to the cutoff resolution and it produces values that are consistent with what the readers are used to. With this solution people could include whatever noisy data they want and not be guilty of overstating the quality of their model. Dale Tronrud > JPK > > > > On Thu, May 31, 2012 at 1:59 PM, Edward A. Berry wrote: >> Yes! I want a copy of this program RESCUT. >> >> REMARK 200 R SYM FOR SHELL(I) : 1.21700 >> I noticed structure 3RKO reported Rmerge in the last shell greater >> than 1, suggesting the police who were defending R-merge were fighting >> a losing battle. And this provides a lot of ammunition to those >> they are fighting. >> >> Jacob Keller wrote: >>> >>> Dear Crystallographers, >>> >>> in case you have not heard, it would appear that the Rmerge statistic >>> has died as of the publication of PMID: 22628654. Ding Dong...? >>> >>> JPK >>> >>> -- >>> *** >>> Jacob Pearson Keller >>> Northwestern University >>> Medical Scientist Training Program >>> email: j-kell...@northwestern.edu >>> *** >>> >> > > > > -- > *** > Jacob Pearson Keller > Northwestern University > Medical Scientist Training Program > email: j-kell...@northwestern.edu > *** > >
[ccp4bb] Fwd: [ccp4bb] Death of Rmerge
Alas, how many lines like the following from a recent Science paper (PMID: 22605777), probably reviewer-incited, could have been avoided! "Here, we present three high-resolution crystal structures of the Thermus thermophilus (Tth) 70S ribosome in complex withRMF, HPF, or YfiA that were refined by using data extending to 3.0 Å (I/sI = 1), 3.1 Å (I/sI = 1), and 2.75 Å (I/sI = 1) resolution, respectively. The resolutions at which I/sI = 2 are 3.2 Å, 3.4 Å, and 2.9 Å, respectively." JPK On Thu, May 31, 2012 at 1:59 PM, Edward A. Berry wrote: > Yes! I want a copy of this program RESCUT. > > REMARK 200 R SYM FOR SHELL (I) : 1.21700 > I noticed structure 3RKO reported Rmerge in the last shell greater > than 1, suggesting the police who were defending R-merge were fighting > a losing battle. And this provides a lot of ammunition to those > they are fighting. > > Jacob Keller wrote: >> >> Dear Crystallographers, >> >> in case you have not heard, it would appear that the Rmerge statistic >> has died as of the publication of PMID: 22628654. Ding Dong...? >> >> JPK >> >> -- >> *** >> Jacob Pearson Keller >> Northwestern University >> Medical Scientist Training Program >> email: j-kell...@northwestern.edu >> *** >> > -- *** Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: j-kell...@northwestern.edu *** -- *** Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: j-kell...@northwestern.edu ***
Re: [ccp4bb] Death of Rmerge
Yes! I want a copy of this program RESCUT. REMARK 200 R SYM FOR SHELL(I) : 1.21700 I noticed structure 3RKO reported Rmerge in the last shell greater than 1, suggesting the police who were defending R-merge were fighting a losing battle. And this provides a lot of ammunition to those they are fighting. Jacob Keller wrote: Dear Crystallographers, in case you have not heard, it would appear that the Rmerge statistic has died as of the publication of PMID: 22628654. Ding Dong...? JPK -- *** Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: j-kell...@northwestern.edu ***
Re: [ccp4bb] Death of Rmerge
Rpim (from the East) just told me "You don't need to be helped any longer. You've always had the power to go back to Kansas." -Original Message- From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of Oganesyan, Vaheh Sent: Thursday, May 31, 2012 2:27 PM To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] Death of Rmerge It wasn't doing well lately. So, it was expected. From: CCP4 bulletin board [CCP4BB@JISCMAIL.AC.UK] on behalf of Jacob Keller [j-kell...@fsm.northwestern.edu] Sent: Thursday, May 31, 2012 2:20 PM To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] Death of Rmerge Dear Crystallographers, in case you have not heard, it would appear that the Rmerge statistic has died as of the publication of PMID: 22628654. Ding Dong...? JPK -- *** Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: j-kell...@northwestern.edu *** To the extent this electronic communication or any of its attachments contain information that is not in the public domain, such information is considered by MedImmune to be confidential and proprietary. This communication is expected to be read and/or used only by the individual(s) for whom it is intended. If you have received this electronic communication in error, please reply to the sender advising of the error in transmission and delete the original message and any accompanying documents from your system immediately, without copying, reviewing or otherwise using them for any purpose. Thank you for your cooperation. Notice: This e-mail message, together with any attachments, contains information of Merck & Co., Inc. (One Merck Drive, Whitehouse Station, New Jersey, USA 08889), and/or its affiliates Direct contact information for affiliates is available at http://www.merck.com/contact/contacts.html) that may be confidential, proprietary copyrighted and/or legally privileged. It is intended solely for the use of the individual or entity named on this message. If you are not the intended recipient, and have received this message in error, please notify us immediately by reply e-mail and then delete it from your system.
Re: [ccp4bb] Death of Rmerge
It wasn't doing well lately. So, it was expected. From: CCP4 bulletin board [CCP4BB@JISCMAIL.AC.UK] on behalf of Jacob Keller [j-kell...@fsm.northwestern.edu] Sent: Thursday, May 31, 2012 2:20 PM To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] Death of Rmerge Dear Crystallographers, in case you have not heard, it would appear that the Rmerge statistic has died as of the publication of PMID: 22628654. Ding Dong...? JPK -- *** Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: j-kell...@northwestern.edu *** To the extent this electronic communication or any of its attachments contain information that is not in the public domain, such information is considered by MedImmune to be confidential and proprietary. This communication is expected to be read and/or used only by the individual(s) for whom it is intended. If you have received this electronic communication in error, please reply to the sender advising of the error in transmission and delete the original message and any accompanying documents from your system immediately, without copying, reviewing or otherwise using them for any purpose. Thank you for your cooperation.
Re: [ccp4bb] Death of Rmerge
Meant to include the following link in the previous message: http://www.youtube.com/watch?v=9Jn8K8EA7-Q JPK On Thu, May 31, 2012 at 1:20 PM, Jacob Keller wrote: > Dear Crystallographers, > > in case you have not heard, it would appear that the Rmerge statistic > has died as of the publication of PMID: 22628654. Ding Dong...? > > JPK > > -- > *** > Jacob Pearson Keller > Northwestern University > Medical Scientist Training Program > email: j-kell...@northwestern.edu > *** -- *** Jacob Pearson Keller Northwestern University Medical Scientist Training Program email: j-kell...@northwestern.edu ***