<html><head></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><div>Hi Tom,&nbsp;</div><div>Do you think something like this could be used during refinement to identify the "best" resolution limits? If you have an Rsleep set would Rfree be sufficient for this? I imagine &nbsp;collecting data with a ring of noise and then let the optimal resolution be determined during refinement. My understanding of this is that the modern refinement algorithms can handle some noise in the reflections, but maybe this could be a way to optimize how much signal is needed to contribute in a positive fashion?&nbsp;</div><div>Kendall</div><div><br><blockquote type="cite"><div bgcolor="#FFFFFF"><div><br></div><blockquote type="cite"><div><b>From:</b> "Terwilliger, Thomas C" &lt;<a href="mailto:terwilliger@lanl.gov">terwilliger@lanl.gov</a>&gt;<br><b>Date:</b> May 3, 2012 10:23:29 AM EDT<br><b>To:</b> PHENIX user mailing list &lt;<a href="mailto:phenixbb@phenix-online.org">phenixbb@phenix-online.org</a>&gt;<br><b>Subject:</b> <b>Re: [phenixbb] Geometry Restraints - Anisotropic truncation</b><br><b>Reply-To:</b> PHENIX user mailing list &lt;<a href="mailto:phenixbb@phenix-online.org">phenixbb@phenix-online.org</a>&gt;<br><br></div></blockquote><div></div><blockquote type="cite"><div><span>Hi Kendall,</span><br><span></span><br><span>This could work. &nbsp;You could define a fixed set of test reflections, and never touch these, and never include them in refinement, and always use this fixed set to calculate a free R. &nbsp;Then you could do whatever you want, throw away some work reflections, etc, refine, and evaluate how things are working with the fixed free R set.</span><br><span></span><br><span>All the best,</span><br><span>Tom T</span><br><span></span><br><span>________________________________________</span><br><span>From: <a href="mailto:phenixbb-bounces@phenix-online.org">phenixbb-bounces@phenix-online.org</a> [<a href="mailto:phenixbb-bounces@phenix-online.org">phenixbb-bounces@phenix-online.org</a>] on behalf of Kendall Nettles [<a href="mailto:knettles@scripps.edu">knettles@scripps.edu</a>]</span><br><span>Sent: Thursday, May 03, 2012 7:05 AM</span><br><span>To: PHENIX user mailing list</span><br><span>Subject: Re: [phenixbb] Geometry Restraints - Anisotropic truncation</span><br><span></span><br><span>Hi Pavel,</span><br><span>Could you use a similar approach to figuring out where to cut your data in general? Could you compare the effects of throwing out reflections in different bins, based on I/sigma, for example, and use this to determine what is truly noise? I might predict that as you throw out "noise" reflections you will see a larger drop in Rfree than from throwing out "signal" reflections, which should converge as you approach the "true" resolution. While we don't use I/sigma exclusively, we do to tend towards cutting most of our data sets at the same i/sigma, around 1.5. It would be great if there was a more scientific approach.</span><br><span>Best,</span><br><span>Kendall</span><br><span></span><br><span>On May 3, 2012, at 7:45 AM, Pavel Afonine wrote:</span><br><span></span><br><blockquote type="cite"><span>Hi Kendall,</span><br></blockquote><blockquote type="cite"><span></span><br></blockquote><blockquote type="cite"><span>removing same amount of data randomly gives Rwork/Rfree ~ 30/35%.</span><br></blockquote><blockquote type="cite"><span></span><br></blockquote><blockquote type="cite"><span>Pavel</span><br></blockquote><blockquote type="cite"><span></span><br></blockquote><blockquote type="cite"><span>On 5/3/12 4:13 AM, Kendall Nettles wrote:</span><br></blockquote><blockquote type="cite"><blockquote type="cite"><span>Hi Pavel,</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>What happens if you throw out that many reflections that have signal? Can you take out a random set of the same size?</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>Best,</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>Kendall</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>On May 3, 2012, at 2:41 AM, "Pavel Afonine"&lt;<a href="mailto:pafonine@lbl.gov">pafonine@lbl.gov</a>&gt; &nbsp;wrote:</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>Hi Kendall,</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>I just did this quick test: calculated R-factors using original and</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>anisotropy-corrected Mike Sawaya's data (*)</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>Original:</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;&nbsp;r_work : 0.3026</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;&nbsp;r_free : 0.3591</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;&nbsp;number of reflections: 26944</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>Truncated:</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;&nbsp;r_work : 0.2640</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;&nbsp;r_free : 0.3178</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;&nbsp;number of reflections: 18176</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>The difference in R-factors is not too surprising given how many</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>reflections was removed (about 33%).</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>Pavel</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>(*) Note, the data available in PDB is anisotropy corrected. The</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>original data set was kindly provided to me by the author.</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>On 5/2/12 5:25 AM, Kendall Nettles wrote:</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> I didnt think the structure was publishable with Rfree of 33% &nbsp;because I was expecting the reviewers to complain.</span><br></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>We have tested a number of data sets on the UCLA server and it usually doesn't make much difference. I wouldn't expect truncation alone to change Rfree by 5%, and it usually doesn't. The two times I have seen dramatic impacts on the maps ( and Rfree ), the highly anisotrophic sets showed strong waves of difference density as well, which was fixed by throwing out the noise. We have moved to using loose data cutoffs for most structures, but I do think anisotropic truncation can be helpful in rare cases.</span><br></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>Kendall</span><br></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>On May 1, 2012, at 3:07 PM, "Dale Tronrud"&lt;<a href="mailto:det102@uoxray.uoregon.edu">det102@uoxray.uoregon.edu</a>&gt; &nbsp;&nbsp;wrote:</span><br></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;While philosophically I see no difference between a spherical resolution</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>cutoff and an elliptical one, a drop in the free R can't be the justification</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>for the switch. &nbsp;A model cannot be made more "publishable" simply by discarding</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>data.</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;We have a whole bunch of empirical guides for judging the quality of this</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>and that in our field. &nbsp;We determine the resolution limit of a data set (and</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>imposing a "limit" is another empirical choice made) based on Rmrg, or Rmes,</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>or Rpim getting too big or I/sigI getting too small and there is no agreement</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>on how "too big/small" is too "too big/small".</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;We then have other empirical guides for judging the quality of the models</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>we produce (e.g. Rwork, Rfree, rmsds of various sorts). &nbsp;Most people seem to</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>recognize that the these criteria need to be applied differently for different</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>resolutions. A lower resolution model is allowed a higher Rfree, for example.</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;Isn't is also true that a model refined to data with a cutoff of I/sigI of</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>1 would be expected to have a free R higher than a model refined to data with</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>a cutoff of 2? &nbsp;Surely we cannot say that the decrease in free R that results</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>from changing the cutoff criteria from 1 to 2 reflects an improved model. &nbsp;It</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>is the same model after all.</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;Sometimes this shifting application of empirical criteria enhances the</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>adoption of new technology. &nbsp;Certainly the TLS parametrization of atomic</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>motion has been widely accepted because it results in lower working and free</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>Rs. &nbsp;I've seen it knock 3 to 5 percent off, and while that certainly means</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>that the model fits the data better, I'm not sure that the quality of the</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>hydrogen bond distances, van der Waals distances, or maps are any better.</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>The latter details are what I really look for in a model.</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;On the other hand, there has been good evidence through the years that</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>there is useful information in the data beyond an I/sigI of 2 or an</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>Rmeas&gt; &nbsp;&nbsp;100% but getting people to use this data has been a hard slog. &nbsp;The</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>reason for this reluctance is that the R values of the resulting models</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>are higher. &nbsp;Of course they are higher! &nbsp;That does not mean the models</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>are of poorer quality, only that data with lower signal/noise has been</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>used that was discarded in the models you used to develop your "gut feeling"</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>for the meaning of R.</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span> &nbsp;&nbsp;When you change your criteria for selecting data you have to discard</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>your old notions about the acceptable values of empirical quality measures.</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>You either have to normalize your measure, as Phil Jeffrey recommends, by</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>ensuring that you calculate your R's with the same reflections, or by</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>making objective measures of map quality.</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>Dale Tronrud</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>P.S. It is entirely possible that refining a model to a very optimistic</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>resolution cutoff and calculating the map to a lower resolution might be</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>better than throwing out the data altogether.</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>On 5/1/2012 10:34 AM, Kendall Nettles wrote:</span><br></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>I have seen dramatic improvements in maps and behavior during refinement following use of the UCLA anisotropy server in two different cases. For one of them the Rfree went from 33% to 28%. I don't think it would have been publishable otherwise.</span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>Kendall</span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>On May 1, 2012, at 11:10 AM, Bryan Lepore wrote:</span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>On Mon, Apr 30, 2012 at 4:22 AM, Phil Evans&lt;<a href="mailto:pre@mrc-lmb.cam.ac.uk">pre@mrc-lmb.cam.ac.uk</a>&gt; &nbsp;&nbsp;&nbsp;wrote:</span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>Are anisotropic cutoff desirable?</span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>is there a peer-reviewed publication - perhaps from Acta</span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>Crystallographica - which describes precisely why scaling or</span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>refinement programs are inadequate to ameliorate the problem of</span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>anisotropy, and argues why the method applied in Strong, et. al. 2006</span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>satisfies this need?</span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>-Bryan</span><br></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>_______________________________________________</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span>phenixbb mailing list</span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span><a href="mailto:phenixbb@phenix-online.org">phenixbb@phenix-online.org</a></span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><blockquote type="cite"><span><a href="http://phenix-online.org/mailman/listinfo/phenixbb">http://phenix-online.org/mailman/listinfo/phenixbb</a></span><br></blockquote></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>_______________________________________________</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>phenixbb mailing list</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span><a href="mailto:phenixbb@phenix-online.org">phenixbb@phenix-online.org</a></span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span><a href="http://phenix-online.org/mailman/listinfo/phenixbb">http://phenix-online.org/mailman/listinfo/phenixbb</a></span><br></blockquote></blockquote><blockquote type="cite"><span></span><br></blockquote><blockquote type="cite"><span>_______________________________________________</span><br></blockquote><blockquote type="cite"><span>phenixbb mailing list</span><br></blockquote><blockquote type="cite"><span><a href="mailto:phenixbb@phenix-online.org">phenixbb@phenix-online.org</a></span><br></blockquote><blockquote type="cite"><span><a href="http://phenix-online.org/mailman/listinfo/phenixbb">http://phenix-online.org/mailman/listinfo/phenixbb</a></span><br></blockquote><span></span><br><span>_______________________________________________</span><br><span>phenixbb mailing list</span><br><span><a href="mailto:phenixbb@phenix-online.org">phenixbb@phenix-online.org</a></span><br><span><a href="http://phenix-online.org/mailman/listinfo/phenixbb">http://phenix-online.org/mailman/listinfo/phenixbb</a></span><br><span>_______________________________________________</span><br><span>phenixbb mailing list</span><br><span><a href="mailto:phenixbb@phenix-online.org">phenixbb@phenix-online.org</a></span><br><span><a href="http://phenix-online.org/mailman/listinfo/phenixbb">http://phenix-online.org/mailman/listinfo/phenixbb</a></span><br></div></blockquote></div></blockquote></div><br></body></html>