Re: [ccp4bb] domain boundary, new fold, structure-based sequence alignment
On Sun, Oct 18, 2009 at 10:00 AM, Shaun Lott s.l...@auckland.ac.nz wrote: In my experience, DALI can be better than SSM at detecting distant structural similarities. this is interesting. Do you take into account that, in difference of DALI, SSM has a control on the remoteness of structural hits that it delivers? By default, it looks only at fairly similar hits. It grabs more hits as you pull the similarity threshold down. And if you set that similarity threshold zero, nearly every query gathers half of PDB as potential targets - because small features are shared between many structures. I would be very curious to see an example where SSM *is not able* to catch something that DALI catches - leaving aside SSM limitations like presence of secondary structure and having at least 2-3 SSEs in the common motif. Eugene You might also want to try CE. To decide if a fold is 'new' rather than 'old but decorated' often boils down to a somewhat subjective call, but asking Alexei Murzin is as good a way as any to decide :-) For domain definition, try http://pdomains.sdsc.edu/ which offers a handy interface for comparing a number of different methods for automatic domain calling. hope this helps Shaun
[ccp4bb] how to improve Rfree?
Dear all I have been refining my structure using refmac5 restrained refinement and the model currently has an R-factor of 0.19. What is concerning me is that there is quite a big discrepancy between the Rfree and the R-factor, with the Rfree being around 0.27. Furthermore, after about 10 rounds of refinement, the R-free actually starts to increase slightly although the R-factor continues to decrease. Does this mean that there is significant bias in my model? Does anybody know how I could go about addressing this problem? Thanks for any help :-) Sylvia Fanucchi Ph.D Protein Structure-Function Research Unit East Campus, Gate House Room 416 School of Molecular and Cell Biology University of the Witwatersrand Johannesburg 2050 South Africa Tel: +27 (11) 717-6348 Fax: +27 (11) 717-6351 E-mail: sylvia.fanuc...@wits.ac.za mailto:sylvia.fanuc...@wits.ac.za htmlpfont face = verdana size = 0.8 color = navyThis communication is intended for the addressee only. It is confidential. If you have received this communication in error, please notify us immediately and destroy the original message. You may not copy or disseminate this communication without the permission of the University. Only authorized signatories are competent to enter into agreements on behalf of the University and recipients are thus advised that the content of this message may not be legally binding on the University and may contain the personal views and opinions of the author, which are not necessarily the views and opinions of The University of the Witwatersrand, Johannesburg. All agreements between the University and outsiders are subject to South African Law unless the University agrees in writing to the contrary./font/p/html image001.jpg
Re: [ccp4bb] how to improve Rfree?
Hi Sylvia You didn't make it clear on what basis you think the discrepancy between R and Rfree is too big, since the Rfree you would expect to get if the refinement were bias-free depends a lot on a number of factors such as the resolution, the data completeness, what parameters you are refining, what restraints you are using, whether the restraint weights are optimal, the solvent content etc. So to judge whether Rfree is indeed higher than would be expected I would need to know a lot more about the refinement. Otherwise it's a bit like asking how long does a piece of string have to be? - I would need to know the purpose of the piece of string! Also Rfree is only meaningful at convergence at refinement (you may as well ignore all other printed values - they are worthless!). So it's not meaningful to compare Rfree during a refinement with the final value at convergence. It's only meaningful to compare Rfree values obtained at convergence of refinements done under different conditions, for example, different parameterisation, different weights, different no of waters (provided only that you don't use different X-ray data or restraints). Cheers -- Ian -Original Message- From: owner-ccp...@jiscmail.ac.uk [mailto:owner-ccp...@jiscmail.ac.uk] On Behalf Of Sylvia Fanucchi Sent: 19 October 2009 13:49 To: CCP4BB@JISCMAIL.AC.UK Subject: how to improve Rfree? Dear all I have been refining my structure using refmac5 restrained refinement and the model currently has an R-factor of 0.19. What is concerning me is that there is quite a big discrepancy between the Rfree and the R-factor, with the Rfree being around 0.27. Furthermore, after about 10 rounds of refinement, the R-free actually starts to increase slightly although the R-factor continues to decrease. Does this mean that there is significant bias in my model? Does anybody know how I could go about addressing this problem? Thanks for any help :-) Sylvia Fanucchi Ph.D Protein Structure-Function Research Unit East Campus, Gate House Room 416 School of Molecular and Cell Biology University of the Witwatersrand Johannesburg 2050 South Africa Tel: +27 (11) 717-6348 Fax: +27 (11) 717-6351 E-mail: sylvia.fanuc...@wits.ac.za mailto:sylvia.fanuc...@wits.ac.za This communication is intended for the addressee only. It is confidential. If you have received this communication in error, please notify us immediately and destroy the original message. You may not copy or disseminate this communication without the permission of the University. Only authorized signatories are competent to enter into agreements on behalf of the University and recipients are thus advised that the content of this message may not be legally binding on the University and may contain the personal views and opinions of the author, which are not necessarily the views and opinions of The University of the Witwatersrand, Johannesburg. All agreements between the University and outsiders are subject to South African Law unless the University agrees in writing to the contrary. Disclaimer This communication is confidential and may contain privileged information intended solely for the named addressee(s). It may not be used or disclosed except for the purpose for which it has been sent. If you are not the intended recipient you must not review, use, disclose, copy, distribute or take any action in reliance upon it. If you have received this communication in error, please notify Astex Therapeutics Ltd by emailing i.tic...@astex-therapeutics.com and destroy all copies of the message and any attached documents. Astex Therapeutics Ltd monitors, controls and protects all its messaging traffic in compliance with its corporate email policy. The Company accepts no liability or responsibility for any onward transmission or use of emails and attachments having left the Astex Therapeutics domain. Unless expressly stated, opinions in this message are those of the individual sender and not of Astex Therapeutics Ltd. The recipient should check this email and any attachments for the presence of computer viruses. Astex Therapeutics Ltd accepts no liability for damage caused by any virus transmitted by this email. E-mail is susceptible to data corruption, interception, unauthorized amendment, and tampering, Astex Therapeutics Ltd only send and receive e-mails on the basis that the Company is not liable for any such alteration or any consequences thereof. Astex Therapeutics Ltd., Registered in England at 436 Cambridge Science Park, Cambridge CB4 0QA under number 3751674
Re: [ccp4bb] how to improve Rfree?
Dear Sylvia, The discrepancy is only 8% and that is acceptable (you may be able to get the Rfree lower by changing the refinement program for the ultimate rounds of refinement, or by using TLS refinement if you don't do that already --- you don't mention what refinement program you are using). WRT your question concerning the increase in Rfree: once you have the R and Rfree as far down as they can be, there's only one way they can go: back up. Also, the aim of refinement is not to minimize R and Rfree but to get a model that matches best all available information. So checking the values of R and Rfree is not sufficient. Is the geometry acceptable? HTH, Fred. Sylvia Fanucchi wrote: Dear all I have been refining my structure using refmac5 restrained refinement and the model currently has an R-factor of 0.19. What is concerning me is that there is quite a big discrepancy between the Rfree and the R-factor, with the Rfree being around 0.27. Furthermore, after about 10 rounds of refinement, the R-free actually starts to increase slightly although the R-factor continues to decrease. Does this mean that there is significant bias in my model? Does anybody know how I could go about addressing this problem? Thanks for any help J **Sylvia Fanucchi Ph.D** **Protein Structure-Function Research Unit** East Campus, Gate House Room 416 School of Molecular and Cell Biology University of the Witwatersrand Johannesburg 2050 South Africa Tel: +27 (11) 717-6348 Fax: +27 (11) 717-6351 E-mail: sylvia.fanuc...@wits.ac.za mailto:sylvia.fanuc...@wits.ac.za This communication is intended for the addressee only. It is confidential. If you have received this communication in error, please notify us immediately and destroy the original message. You may not copy or disseminate this communication without the permission of the University. Only authorized signatories are competent to enter into agreements on behalf of the University and recipients are thus advised that the content of this message may not be legally binding on the University and may contain the personal views and opinions of the author, which are not necessarily the views and opinions of The University of the Witwatersrand, Johannesburg. All agreements between the University and outsiders are subject to South African Law unless the University agrees in writing to the contrary. attachment: Frederic_Vellieux.vcf
[ccp4bb] constrained monomer
Dear all, what are the best criteria to say that one monomer is more constrained by the packing than another? surfaces analyses by PISA? How to be sure that the observed differences are significant? Thanks in advance. Cedric Bauvois
Re: [ccp4bb] how to improve Rfree?
Hi Sylvia: -Original Message- From: Sylvia Fanucchi [mailto:sylvia.fanuc...@wits.ac.za] Sent: 19 October 2009 15:06 To: Ian Tickle Subject: RE: how to improve Rfree? Hi Ian Yes, sorry, admittedly this is the first time I'm doing this and I'm not sure of all the checks that need to be made. I have been checking the validity of my model in coot. I've been under the impression that the initial factor that will tell you whether the refinement has been any good and whether the model is on the right track is the R-factor. I figured that once I'd got that to a satisfactory number I would run the model through validating programmes such as procheck or molprobity. Am I missing something fundamental? This would not surprise me since I am very new to this and have essentially taught myself most of everything I know. Procheck/Molprobity is a good start for geometric validation! It's also important to look for obvious unexplained peaks in the difference Fourier since good agreement of the model with the X-ray data is also important (Coot will help with this); and there are programs which list the real-space density correlation coefficients and/or density RMSD that may help identify problem areas. The refinement program often also has tables and/or graphs of other useful statistics such as structure factor correlation coefficients and log-likelihood, as well as R factors of course. My resolution is 1.6A although I have cut it to 1.8A to bring the R-factor down. I've been performing restrained refinement in refmac5 using the default settings. The solvent content is 40% An Rfree-R difference of 0.08 at 1.8 Ang does indeed seem somewhat high, though the low solvent content may go some way towards explaining it (since it means you will have fewer data in comparison with other structures of similar size and resolution). In addition if your overall completeness is say 0.95 (or 0.8 in the outer shell) that won't help. However I assume you mean Rmerge not Rwork: the purpose of refinement is not to reduce the R factor but to get the model which best explains the observations (including the restraints), and also makes sense in comparison with the body of other structures already determined: rejecting data on the basis of Rwork and/or Rfree may not be the best way to achieve this! Personally I reject data based only on shell completeness ( 0.8) and mean I/sd(I) ( 1.5); I don't use Rmerge for this - see recent BB discussions. Someone told me once that if the Rfree increased with successive rounds of refinement while the R-factor decreased, something was very wrong and that was why I was concerned. If Rfree (or -LLfree) *ends up* being higher with one refined model / parameterisation / weighting scheme compared with another (with the same data), then one would reject the first model. However what you can't do is compare the models on the basis of the intermediate Rfree/LLfree values. Cheers -- Ian Disclaimer This communication is confidential and may contain privileged information intended solely for the named addressee(s). It may not be used or disclosed except for the purpose for which it has been sent. If you are not the intended recipient you must not review, use, disclose, copy, distribute or take any action in reliance upon it. If you have received this communication in error, please notify Astex Therapeutics Ltd by emailing i.tic...@astex-therapeutics.com and destroy all copies of the message and any attached documents. Astex Therapeutics Ltd monitors, controls and protects all its messaging traffic in compliance with its corporate email policy. The Company accepts no liability or responsibility for any onward transmission or use of emails and attachments having left the Astex Therapeutics domain. Unless expressly stated, opinions in this message are those of the individual sender and not of Astex Therapeutics Ltd. The recipient should check this email and any attachments for the presence of computer viruses. Astex Therapeutics Ltd accepts no liability for damage caused by any virus transmitted by this email. E-mail is susceptible to data corruption, interception, unauthorized amendment, and tampering, Astex Therapeutics Ltd only send and receive e-mails on the basis that the Company is not liable for any such alteration or any consequences thereof. Astex Therapeutics Ltd., Registered in England at 436 Cambridge Science Park, Cambridge CB4 0QA under number 3751674
Re: [ccp4bb] how to improve Rfree?
*If* you are using WEIGHT AUTO, check the RMSDbonds. You don't say what your resolution is, but I assume it's in lower 2.0s. I've seen refmac5 in some cases produce somewhat unreasonable geometries (i.e. RMSDbonds0.025A) at such medium resolution and accordingly, large R/Rfree gap. If this is the case, one suggestion is to determine the optimal weight manually (e.g. the one that minimizes Rfree). This IMLE does not lower the Rfree, but increases R, thus closing the gap. You may also consider NCS restraints if you have multiple copies. Does anyone already have the PDB-wide R/Rfree gap versus resolution data (it does not seem that the PDB maintains one)? Ed. On Mon, 2009-10-19 at 14:49 +0200, Sylvia Fanucchi wrote: Dear all I have been refining my structure using refmac5 restrained refinement and the model currently has an R-factor of 0.19. What is concerning me is that there is quite a big discrepancy between the Rfree and the R-factor, with the Rfree being around 0.27. Furthermore, after about 10 rounds of refinement, the R-free actually starts to increase slightly although the R-factor continues to decrease. Does this mean that there is significant bias in my model? Does anybody know how I could go about addressing this problem? Thanks for any help J Sylvia Fanucchi Ph.D Protein Structure-Function Research Unit East Campus, Gate House Room 416 School of Molecular and Cell Biology University of the Witwatersrand Johannesburg 2050 South Africa Tel: +27 (11) 717-6348 Fax: +27 (11) 717-6351 E-mail: sylvia.fanuc...@wits.ac.za This communication is intended for the addressee only. It is confidential. If you have received this communication in error, please notify us immediately and destroy the original message. You may not copy or disseminate this communication without the permission of the University. Only authorized signatories are competent to enter into agreements on behalf of the University and recipients are thus advised that the content of this message may not be legally binding on the University and may contain the personal views and opinions of the author, which are not necessarily the views and opinions of The University of the Witwatersrand, Johannesburg. All agreements between the University and outsiders are subject to South African Law unless the University agrees in writing to the contrary. --
[ccp4bb] Postdoctoral Position @ Cold Spring Harbor Laboratory
Applications are invited for postdoctoral positions at Cold Spring Harbor Laboratory (CSHL) from highly enthusiastic individuals with strong interest in utilizing structural, electrophysiological, and biochemical approaches to study fundamental problems in ion and substrate transports across the membrane. Our biological goal is to understand molecular basis for synaptic transmission and neurological disorders and diseases. We use x-ray crystallography to obtain atomic structures of target membrane/glyco proteins and test structure-based functional hypotheses by various techniques including electrophysiology, analytical centrifugation, and fluorescence assay. We are equipped with a state-of-the-art facility to conduct membrane/glyco protein crystallography and functional studies including a high-throughput protein screening system, an automated crystallization screening device, a recently updated home x-ray source, electrophysiology rigs, and other setups such as isothermal titration colorimetry device. CSHL is also located within 30 min driving distance from the National Synchrotron Light Source at Brookhaven National Laboratory. Most importantly, a highly collaborative environment at CSHL provides a unique opportunity for postdocs to interact with scientists from various fields including cancer biology, neuroscience, and plant biology. Candidates must hold (or soon expect to hold) PhD in biochemistry, molecular biology, biophysics or related fields. Strong background in molecular biology, biochemistry, large-scale protein expression in eukaryotic and prokaryotic cells, and purification of membrane/glyco proteins is desired. Experience in x-ray crystallography or electrophysiology is desirable but not necessary. For more information, please check the laboratory website (http://www.cshl.edu/public/SCIENCE/furukawa.html) and contact me via email. Please send CV, a summary of research experience and interest, and three references to Hiro Furukawa by email at furuk...@cshl.edu
Re: [ccp4bb] how to improve Rfree?
Does anyone already have the PDB-wide R/Rfree gap versus resolution data (it does not seem that the PDB maintains one)? Ed. The friendly people in Uppsala has something here by the name Harry Plotter: http://xray.bmc.uu.se/gerard/supmat/eds/plotter.html This might be somewhat dated now. Engin -- Engin Özkan Post-doctoral Scholar Howard Hughes Medical Institute Dept of Molecular and Cellular Physiology 279 Campus Drive, Beckman Center B173 Stanford School of Medicine Stanford, CA 94305 ph: (650)-498-7111
[ccp4bb] Available Postdoctoral Position
A post-doctoral position is immediately available in the laboratory of Dr. E. Skordalakes at The Wistar Institute, UPENN to study protein nucleic acid assemblies that participate in the replication and maintenance of telomeres. We are looking for highly motivated candidates that have recently acquired a Ph.D. in biochemistry and/or macromolecular crystallography. Additional information regarding our research program can be found at the laboratory web site(s) http://www.chem.upenn.edu/chem/research/faculty.php?id=52 http://www.wistar.org/research_facilities/skordalakes/research.htm Interested applicants should submit a curriculum vitae and three letters of recommendation to Emmanuel Skordalakes at sko...@wistar.org
Re: [ccp4bb] how to improve Rfree?
My resolution is 1.6A although I have cut it to 1.8A to bring the R-factor down. I've been performing restrained refinement in refmac5 using the default settings. The solvent content is 40% This sounds fundamentally wrong. Even the Rmerge reduction by cutting resolution practice is questionable due to strong dependence of the Rmerge on redundancy, but to cut resolution to get better Rfactor in refinement... Nobody throws away the lower concentration (and thus noisier) data in Bradford assay, although I am sure that the appropriate R-factor would definitely go down.
Re: [ccp4bb] how to improve Rfree?
I agree. It cannot be said too often that the object of refinement is not to minimise the R-factor! ( while I'm at it, freeR has nothing to do with model bias (which is largely imaginary anyway, only a problem with low resolution molecular replacement)) Phil On 19 Oct 2009, at 17:07, Ed Pozharski wrote: My resolution is 1.6A although I have cut it to 1.8A to bring the R-factor down. I've been performing restrained refinement in refmac5 using the default settings. The solvent content is 40% This sounds fundamentally wrong. Even the Rmerge reduction by cutting resolution practice is questionable due to strong dependence of the Rmerge on redundancy, but to cut resolution to get better Rfactor in refinement... Nobody throws away the lower concentration (and thus noisier) data in Bradford assay, although I am sure that the appropriate R-factor would definitely go down.
Re: [ccp4bb] how to improve Rfree?
Sorry, I gave the wrong link, the R/Rfree gap is reported here: http://xray.bmc.uu.se/gerard/supmat/rfree2000/plotter.html Engin On 10/19/09 9:13 AM, Ed Pozharski wrote: Looks great, thanks. But it doesn't actually report R/Rfree gap, or am I missing something? Ed. On Mon, 2009-10-19 at 08:14 -0700, Engin Ozkan wrote: Does anyone already have the PDB-wide R/Rfree gap versus resolution data (it does not seem that the PDB maintains one)? Ed. The friendly people in Uppsala has something here by the name Harry Plotter: http://xray.bmc.uu.se/gerard/supmat/eds/plotter.html This might be somewhat dated now. Engin -- Engin Özkan Post-doctoral Scholar Howard Hughes Medical Institute Dept of Molecular and Cellular Physiology 279 Campus Drive, Beckman Center B173 Stanford School of Medicine Stanford, CA 94305 ph: (650)-498-7111
Re: [ccp4bb] how to improve Rfree?
Hi Sivia, you say the resolution is 1.8A, and your R-work=0.19 and R-free=0.27. Here is the distribution of R-factors for structures in PDB deposited at similar resolution (in range between 1.75A and 1.85A): Histogram of Rwork for all model in PDB at resolution 1.75-1.85: 0.114 - 0.130 : 6 0.130 - 0.147 : 48 0.147 - 0.163 : 221 0.163 - 0.180 : 484 0.180 - 0.196 : 718 your structure 0.196 - 0.212 : 508 0.212 - 0.229 : 271 0.229 - 0.245 : 105 0.245 - 0.262 : 20 0.262 - 0.278 : 7 Histogram of Rfree for all model in PDB at resolution 1.75-1.85: 0.153 - 0.171 : 17 0.171 - 0.190 : 127 0.190 - 0.208 : 380 0.208 - 0.227 : 647 0.227 - 0.245 : 648 0.245 - 0.263 : 360 0.263 - 0.282 : 132 your structure 0.282 - 0.300 : 64 0.300 - 0.319 : 10 0.319 - 0.337 : 3 Histogram of Rfree-Rwork for all model in PDB at resolution 1.75-1.85: 0.001 - 0.013 : 41 0.013 - 0.025 : 336 0.025 - 0.036 : 852 0.036 - 0.048 : 693 0.048 - 0.060 : 309 0.060 - 0.072 : 98 0.072 - 0.084 : 40 your structure 0.084 - 0.095 : 11 0.095 - 0.107 : 4 0.107 - 0.119 : 4 You also mentioned you cut the data off (the actual resolution is 1.6A). Doing so just to improve R-factor doesn't seem like a good idea. Similar histograms for structures at around 1.6A resolution: Histogram of Rwork for all model in PDB at resolution 1.55-1.65: 0.099 - 0.119 : 5 0.119 - 0.138 : 26 0.138 - 0.158 : 132 0.158 - 0.177 : 349 0.177 - 0.197 : 447 your structure 0.197 - 0.217 : 281 0.217 - 0.236 : 129 0.236 - 0.256 : 17 0.256 - 0.275 : 6 0.275 - 0.295 : 1 Histogram of Rfree for all model in PDB at resolution 1.55-1.65: 0.137 - 0.156 : 6 0.156 - 0.175 : 45 0.175 - 0.195 : 208 0.195 - 0.214 : 391 0.214 - 0.233 : 395 0.233 - 0.253 : 200 0.253 - 0.272 : 96 your structure 0.272 - 0.291 : 37 0.291 - 0.311 : 12 0.311 - 0.330 : 3 Histogram of Rfree-Rwork for all model in PDB at resolution 1.55-1.65: 0.001 - 0.010 : 27 0.010 - 0.019 : 217 0.019 - 0.029 : 374 0.029 - 0.038 : 350 0.038 - 0.047 : 233 0.047 - 0.056 : 106 0.056 - 0.065 : 49 0.065 - 0.075 : 23 0.075 - 0.084 : 10 your structure 0.084 - 0.093 : 4 Pavel. On 10/19/09 5:49 AM, Sylvia Fanucchi wrote: Dear all I have been refining my structure using refmac5 restrained refinement and the model currently has an R-factor of 0.19. What is concerning me is that there is quite a big discrepancy between the Rfree and the R-factor, with the Rfree being around 0.27. Furthermore, after about 10 rounds of refinement, the R-free actually starts to increase slightly although the R-factor continues to decrease. Does this mean that there is significant bias in my model? Does anybody know how I could go about addressing this problem? Thanks for any help J Sylvia Fanucchi Ph.D Protein Structure-Function Research Unit East Campus, Gate House Room 416 School of Molecular and Cell Biology University of the Witwatersrand Johannesburg 2050 South Africa Tel: +27 (11) 717-6348 Fax: +27 (11) 717-6351 E-mail: sylvia.fanuc...@wits.ac.za This communication is intended for the addressee only. It is confidential. If you have received this communication in error, please notify us immediately and destroy the original message. You may not copy or disseminate this communication without the permission of the University. Only authorized signatories are competent to enter into agreements on behalf of the University and recipients are thus advised that the content of this message may not be legally binding on the University and may contain the personal views and opinions of the author, which are not necessarily the views and opinions of The University of the Witwatersrand, Johannesburg. All agreements between the University and outsiders are subject to South African Law unless the University agrees in writing to the contrary.
Re: [ccp4bb] how to improve Rfree?
This is a perennial Q, and one without a good answer. There are some pretty uncontriversible facts. If you have atomic resolution your R and Rfree will be pretty close - 90-95% of data gives you a pretty well correct structure and the FreeR will reflect that.. If you have low resolution and you get the spacegroup wrong - eg call P3122 P31 and assign the FreeR set without taking account of the extra symmetry your R and Rfree will be pretty close! but that doesnt mean you have a better structure.. If you have non-crystallographic symmetry the effects are uncertain, but in many cases this too gives you a lower than expected FreeR value, and again, that doesnt mean you have a better structure.. In general the difference in R and Rfree should be correlated with the resolution, but there doesnt seem to be any absolute level of expectation; thee are toomany other variables to consider.. Eleanor Pavel Afonine wrote: Hi Sivia, you say the resolution is 1.8A, and your R-work=0.19 and R-free=0.27. Here is the distribution of R-factors for structures in PDB deposited at similar resolution (in range between 1.75A and 1.85A): Histogram of Rwork for all model in PDB at resolution 1.75-1.85: 0.114 - 0.130 : 6 0.130 - 0.147 : 48 0.147 - 0.163 : 221 0.163 - 0.180 : 484 0.180 - 0.196 : 718 your structure 0.196 - 0.212 : 508 0.212 - 0.229 : 271 0.229 - 0.245 : 105 0.245 - 0.262 : 20 0.262 - 0.278 : 7 Histogram of Rfree for all model in PDB at resolution 1.75-1.85: 0.153 - 0.171 : 17 0.171 - 0.190 : 127 0.190 - 0.208 : 380 0.208 - 0.227 : 647 0.227 - 0.245 : 648 0.245 - 0.263 : 360 0.263 - 0.282 : 132 your structure 0.282 - 0.300 : 64 0.300 - 0.319 : 10 0.319 - 0.337 : 3 Histogram of Rfree-Rwork for all model in PDB at resolution 1.75-1.85: 0.001 - 0.013 : 41 0.013 - 0.025 : 336 0.025 - 0.036 : 852 0.036 - 0.048 : 693 0.048 - 0.060 : 309 0.060 - 0.072 : 98 0.072 - 0.084 : 40 your structure 0.084 - 0.095 : 11 0.095 - 0.107 : 4 0.107 - 0.119 : 4 You also mentioned you cut the data off (the actual resolution is 1.6A). Doing so just to improve R-factor doesn't seem like a good idea. Similar histograms for structures at around 1.6A resolution: Histogram of Rwork for all model in PDB at resolution 1.55-1.65: 0.099 - 0.119 : 5 0.119 - 0.138 : 26 0.138 - 0.158 : 132 0.158 - 0.177 : 349 0.177 - 0.197 : 447 your structure 0.197 - 0.217 : 281 0.217 - 0.236 : 129 0.236 - 0.256 : 17 0.256 - 0.275 : 6 0.275 - 0.295 : 1 Histogram of Rfree for all model in PDB at resolution 1.55-1.65: 0.137 - 0.156 : 6 0.156 - 0.175 : 45 0.175 - 0.195 : 208 0.195 - 0.214 : 391 0.214 - 0.233 : 395 0.233 - 0.253 : 200 0.253 - 0.272 : 96 your structure 0.272 - 0.291 : 37 0.291 - 0.311 : 12 0.311 - 0.330 : 3 Histogram of Rfree-Rwork for all model in PDB at resolution 1.55-1.65: 0.001 - 0.010 : 27 0.010 - 0.019 : 217 0.019 - 0.029 : 374 0.029 - 0.038 : 350 0.038 - 0.047 : 233 0.047 - 0.056 : 106 0.056 - 0.065 : 49 0.065 - 0.075 : 23 0.075 - 0.084 : 10 your structure 0.084 - 0.093 : 4 Pavel. On 10/19/09 5:49 AM, Sylvia Fanucchi wrote: Web Bug from cid:part1.0305.09050909@lbl.gov Dear all I have been refining my structure using refmac5 restrained refinement and the model currently has an R-factor of 0.19. What is concerning me is that there is quite a big discrepancy between the Rfree and the R-factor, with the Rfree being around 0.27. Furthermore, after about 10 rounds of refinement, the R-free actually starts to increase slightly although the R-factor continues to decrease. Does this mean that there is significant bias in my model? Does anybody know how I could go about addressing this problem? Thanks for any help J **Sylvia Fanucchi Ph.D** **Protein Structure-Function Research Unit** East Campus, Gate House Room 416 School of Molecular and Cell Biology University of the Witwatersrand Johannesburg 2050 South Africa Tel: +27 (11) 717-6348 Fax: +27 (11) 717-6351 E-mail: sylvia.fanuc...@wits.ac.za mailto:sylvia.fanuc...@wits.ac.za This communication is intended for the addressee only. It is confidential. If you have received this communication in error, please notify us immediately and destroy the original message. You may not copy or disseminate this communication without the permission of the University. Only authorized signatories are competent to enter into agreements on
Re: [ccp4bb] how to improve Rfree?
( while I'm at it, freeR has nothing to do with model bias (which is largely imaginary anyway, only a problem with low resolution molecular replacement)) Phil Phil, I understand what you're saying but 'model bias' in this context has a wider meaning than just failing to rebuild the chain correctly from an MR model at low resolution, and using this wider meaning Rfree has everything to do with model bias. For example, if I refine a high res structure with 1 Ang data using an isotropic-only B-factor model, then Rfree ( Rwork) will almost certainly be higher than you would expect using the optimal parameterisation of 3 xyz's + 6 ADPs per atom with the appropriate restraints (whatever the optimum happens to be). Bias is defined as 'difference from expectation', so the isotropic model is almost certainly biased, as reflected in the higher-than-expected Rwork Rfree. I'm well aware that bias by itself is not necessarily a bad thing particularly when a biased estimator happens to have the least error (difference from the truth), but I doubt one can claim that in this example. Cheers -- Ian Disclaimer This communication is confidential and may contain privileged information intended solely for the named addressee(s). It may not be used or disclosed except for the purpose for which it has been sent. If you are not the intended recipient you must not review, use, disclose, copy, distribute or take any action in reliance upon it. If you have received this communication in error, please notify Astex Therapeutics Ltd by emailing i.tic...@astex-therapeutics.com and destroy all copies of the message and any attached documents. Astex Therapeutics Ltd monitors, controls and protects all its messaging traffic in compliance with its corporate email policy. The Company accepts no liability or responsibility for any onward transmission or use of emails and attachments having left the Astex Therapeutics domain. Unless expressly stated, opinions in this message are those of the individual sender and not of Astex Therapeutics Ltd. The recipient should check this email and any attachments for the presence of computer viruses. Astex Therapeutics Ltd accepts no liability for damage caused by any virus transmitted by this email. E-mail is susceptible to data corruption, interception, unauthorized amendment, and tampering, Astex Therapeutics Ltd only send and receive e-mails on the basis that the Company is not liable for any such alteration or any consequences thereof. Astex Therapeutics Ltd., Registered in England at 436 Cambridge Science Park, Cambridge CB4 0QA under number 3751674
Re: [ccp4bb] how to improve Rfree?
If you have non-crystallographic symmetry the effects are uncertain, but in many cases this too gives you a lower than expected FreeR value, and again, that doesnt mean you have a better structure.. In general the difference in R and Rfree should be correlated with the resolution, but there doesnt seem to be any absolute level of expectation; thee are toomany other variables to consider.. Eleanor I agree: in simple cases with good medium-high res data, given Rwork you can predict Rfree within the expected error in Rfree (based only on the size of the test set). The problem is that real life is never simple! - and NCS really messes things up! Cheers -- Ian Disclaimer This communication is confidential and may contain privileged information intended solely for the named addressee(s). It may not be used or disclosed except for the purpose for which it has been sent. If you are not the intended recipient you must not review, use, disclose, copy, distribute or take any action in reliance upon it. If you have received this communication in error, please notify Astex Therapeutics Ltd by emailing i.tic...@astex-therapeutics.com and destroy all copies of the message and any attached documents. Astex Therapeutics Ltd monitors, controls and protects all its messaging traffic in compliance with its corporate email policy. The Company accepts no liability or responsibility for any onward transmission or use of emails and attachments having left the Astex Therapeutics domain. Unless expressly stated, opinions in this message are those of the individual sender and not of Astex Therapeutics Ltd. The recipient should check this email and any attachments for the presence of computer viruses. Astex Therapeutics Ltd accepts no liability for damage caused by any virus transmitted by this email. E-mail is susceptible to data corruption, interception, unauthorized amendment, and tampering, Astex Therapeutics Ltd only send and receive e-mails on the basis that the Company is not liable for any such alteration or any consequences thereof. Astex Therapeutics Ltd., Registered in England at 436 Cambridge Science Park, Cambridge CB4 0QA under number 3751674
Re: [ccp4bb] how to improve Rfree?
Bias is defined as 'difference from expectation', so the isotropic model is almost certainly biased, as reflected in the higher-than-expected Rwork Rfree. Phil, sorry in my rush to go home I explained that badly: bias is defined as 'difference between the expected and true values'. In this case the expectation is the Rfree (or other statistic) that you would expect to get based on the isotropic model, and we're assuming that the optimally parameterised ADP model (whatever it is) is the one closest to the truth: this may of course be a rash assumption since even the optimal parameterisation may well not explain all the errors in the model. The bias in Rfree for the isotropic model is the difference between these. Cheers -- Ian Disclaimer This communication is confidential and may contain privileged information intended solely for the named addressee(s). It may not be used or disclosed except for the purpose for which it has been sent. If you are not the intended recipient you must not review, use, disclose, copy, distribute or take any action in reliance upon it. If you have received this communication in error, please notify Astex Therapeutics Ltd by emailing i.tic...@astex-therapeutics.com and destroy all copies of the message and any attached documents. Astex Therapeutics Ltd monitors, controls and protects all its messaging traffic in compliance with its corporate email policy. The Company accepts no liability or responsibility for any onward transmission or use of emails and attachments having left the Astex Therapeutics domain. Unless expressly stated, opinions in this message are those of the individual sender and not of Astex Therapeutics Ltd. The recipient should check this email and any attachments for the presence of computer viruses. Astex Therapeutics Ltd accepts no liability for damage caused by any virus transmitted by this email. E-mail is susceptible to data corruption, interception, unauthorized amendment, and tampering, Astex Therapeutics Ltd only send and receive e-mails on the basis that the Company is not liable for any such alteration or any consequences thereof. Astex Therapeutics Ltd., Registered in England at 436 Cambridge Science Park, Cambridge CB4 0QA under number 3751674
[ccp4bb] how to improve Rfree?
You can also pick thin shells with SFCHECK and with 12-fold NCS the referees will likely not be kind if you dont!!! -- Mark
Re: [ccp4bb] how to improve Rfree?
On 10/19/09 2:30 PM, Vellieux Frederic frederic.velli...@ibs.fr wrote: One of these people (Bart Hazes I think, correct me if I'm wrong) suggests to take reflections for the R-free as thin shells in reciprocal space. The thin shell is omitted completely from the target for refinement (I suppose omitting a shell of data completely will also have deleterious effects on the refinement, I don't know by how much). Problem is, none of the data processing programs or suites of programs has implemented this as far as I know. You can use Dataman from Uppsala Software Factory: http://xray.bmc.uu.se/usf/ Neel
Re: [ccp4bb] how to improve Rfree?
But see Acta Crystallogr D Biol Crystallogr. 2006 Mar;62(Pt 3):227-38. Epub 2006 Feb 22. on why thin shells aren't necessarily the answer. And note neither program mentioned, dataman nor sftools, will allow you to extend the set of shells to higher resolution if you want to follow the practice of preserving your existing R-free set. Ian Tickle has argued there is no need to transfer the free flags from one data set to another, so that may not be an issue. Pete -Original Message- From: CCP4 bulletin board on behalf of Dr. Mark Mayer Sent: Mon 10/19/2009 11:57 AM To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] how to improve Rfree? You can also pick thin shells with SFCHECK and with 12-fold NCS the referees will likely not be kind if you dont!!! -- Mark
Re: [ccp4bb] how to improve Rfree?
To properly omit reflections from the Rfree calculations can be done using reciprocal space masks as implemented in the program WEIGHTS from the DEMON/ANGEL program suite (Vellieux, Hunt, Roy Read [1995], J. Appl. Cryst. 28, 347-351). I don't have the time to do it. But people are welcome to request the program (they closed the ftp site at our end a few years ago), it is distributed with the free software foundation's general public license. So the program can be hacked to get the relevant code out, to be included in other programs that alre also distributed with the GPL. These are the only restrictions. Otherwise the algorithm can be coded anew into software packages that are not distributed with the GPL. Fred. attachment: Frederic_Vellieux.vcf
Re: [ccp4bb] how to improve Rfree?
The RFREE SHELL command in sftools is another way to select thin shells. However, for the purist there is no clean way of getting rid of correlations between the Rwork and Rfree sets that I know of. To do it properly, the thickness of the thin shell should be larger than the radius where the G-function has strong features. In practice that makes the shells so thick that you can have only a few which likely causes artifacts. Even then, the reflections at the edge of the shells are still contaminated so you'd have to exclude the ones near the edge from both the Rwork and Rfree sets. I'm pretty sure that there has been a publication long ago on thick shells (It may have been Ian, Phil or some other CCP4BB regular). An improvement would be to go from shells to donuts, basically the intersection of the spherical shell and a plane through the origin perpendicular to the NCS rotational symmetry axis. One step further is Fred's suggestion to use the explicit NCS relationships to select groups of NCS-contaminated reflections, but only works if the NCS symmetry forms a closed group. One step further again would be to use an explicit low resolution G-function model to find the strongest NCS correlations and only group those together. Enough fodder for purists to fight over but when sftools users ask me, my general response includes the fact that the higher your NCS the greater the NCS-contamination problem, but also the smaller the model bias/overfitting problem, assuming NCS restraints are used appropriately in refinement. As a reviewer I would have a problem with authors patting their backs for a small Rwork-Rfree difference without noting the impact of NCS. But if they report a low Rwork-Rfree difference and comment on the fact that the difference is likely underestimated due to NCS effects I have no problems, with or without thin shells, as long as I feel the refinement protocol and final model are acceptable for the biological interpretations that they make. Bart Vellieux Frederic wrote: Hi Ian ( ccp4bb'ers), NCS ties reflections in reciprocal space by the interference G-function effect. Nothing more. So you get an R-free value that is lower than if you don't have NCS. One should be aware of that, and referees should be aware of that. I currently have a structure that has 12-fold NCS in the asymmetric unit. The free R-factor is lower than the R-factor. I expect that future referees will not view that kindly. A number of people have suggested to use different approaches to get rid of this reciprocal space binding effect. One of these people (Bart Hazes I think, correct me if I'm wrong) suggests to take reflections for the R-free as thin shells in reciprocal space. The thin shell is omitted completely from the target for refinement (I suppose omitting a shell of data completely will also have deleterious effects on the refinement, I don't know by how much). Problem is, none of the data processing programs or suites of programs has implemented this as far as I know. A better approach would be to use the NCS operator (the transpose and the inverse of the rotation matrices in fact, including the identity matrix for all cases including the cases where the only NCS is 1-fold NCS, i.e. the presence of solvent in the asymmetric unit or unit cell) to select the subset of reflections that are going to be omitted from the refinement target: take one reflection, select all equivalents that are bound to it by the interference effect, and repeat the process until you have reached the required number of reflections to be omitted. But this requires serious programming... And someone willing to modify all data processing suites to include this approach. But that would satisfy referees because it is the only approach that is valid. Fred. Ian Tickle wrote: The problem is that real life is never simple! - and NCS really messes things up! Cheers -- Ian -- Bart Hazes (Associate Professor) Dept. of Medical Microbiology Immunology University of Alberta 1-15 Medical Sciences Building Edmonton, Alberta Canada, T6G 2H7 phone: 1-780-492-0042 fax:1-780-492-7521
[ccp4bb] Difficult molecular replacement in R32 with pseudo translation
Dear All: I got a problem with a data set in R32 space group with a resolution of 2.6. Pointless indicated that the space group was R32 and phenix triage showed that there was a pseudotranslation in the ASU. My protein is a hexamer and the matthew coeff. showed that there should just be two monomers in one ASU of R32. However, phaser in R32 and Molrep in R32 gave different solutions because Molrep used the pseudo-translation. I am not sure which one was correct because refinement of both solutions get stuck around 35% for Rfactor. I tried the molecular replacement in P1 and I got good solutions in Phaser. I think they are correct solutions because my protein was a hexamer and the crystal packing in P1 was formed by hexamers. How can I get correct solutions in R32 and finish the refinement? Many thanks in advance. Jerry McCully _ Your E-mail and More On-the-Go. Get Windows Live Hotmail Free. http://clk.atdmt.com/GBL/go/171222985/direct/01/
Re: [ccp4bb] Difficult molecular replacement in R32 with pseudo translation
On Monday 19 October 2009 12:43:43 Jerry McCully wrote: Dear All: I got a problem with a data set in R32 space group with a resolution of 2.6. Pointless indicated that the space group was R32 and phenix triage showed that there was a pseudotranslation in the ASU. My protein is a hexamer and the matthew coeff. showed that there should just be two monomers in one ASU of R32. However, phaser in R32 and Molrep in R32 gave different solutions because Molrep used the pseudo-translation. I am not sure which one was correct because refinement of both solutions get stuck around 35% for Rfactor. I tried the molecular replacement in P1 and I got good solutions in Phaser. I think they are correct solutions because my protein was a hexamer and the crystal packing in P1 was formed by hexamers. How can I get correct solutions in R32 and finish the refinement? Did you try using the hexamer as your molecular replacement probe? -- Ethan A Merritt Biomolecular Structure Center University of Washington, Seattle 98195-7742
Re: [ccp4bb] mammalian cell culture on IMAC
Brad, It looks like the chelator is a reducing agent added to media for preservation, I guess. The cure is not found yet. I'm just loading ~300 ml per one 5 ml column and it so far serves me well. The IEX approach did not work well. I've lost more than with IMAC after 3X dilution (with properly chosen resin and buffer pH). Your idea works fine: that's why I limit load to 300 ml. Thank you. ___ Vaheh From: Brad Bennett [mailto:bradbennet...@gmail.com] Sent: Thursday, October 08, 2009 4:43 PM To: Oganesyan, Vaheh Subject: Re: [ccp4bb] mammalian cell culture on IMAC Hi Vaheh- Well (gulp) you could dilute your solution say 2-10X with buffer with no salt. You could try small volumes at first and see how dilute the [salt] must be before your protein sticks to your IEX column. So, what is in the media that is stripping off the Ni ions? A chelator like EDTA? And you're sure it's not something more sinister like stripping AND reducing the Ni? If the media formulation is proprietary, you may never definitively know. But I wonder if you could add free resin to a small portion of your solution, spin it down and run the beads on a gel and detect your protein by silver stain or immunoblot? If, whatever it is, is just stripping your resin, then maybe you could pull down some of your target protein. Surely some Ni+ will get chelated but maybe it's worth a shot? Maybe as long as you have plenty of Ni-NTA, Ni-Sepharose or Talon hanging around, you could give this a try. You could try salting it out with saturating ammonium sulfate. Not elegant but it should work and you could resuspend the pellet in whatever buffer and volume you wanted. Any other epitopes/affinity handles on this protein? Like Flag or HA? HTH- Brad On Thu, Oct 8, 2009 at 4:15 PM, Oganesyan, Vaheh oganesy...@medimmune.com wrote: In case of cell lysates this may be a good idea since you can adjust the salt concentration in your sample. In case of secreted proteins it is probably not so good since the media contains ~150 mM NaCl. In my case this salt prevents protein from bindinq to Q column. Thank you anyway. From: CCP4 bulletin board on behalf of Dima Klenchin Sent: Thu 10/8/2009 3:54 PM To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] mammalian cell culture on IMAC When mammalian cell culture is being loaded to GE HisTrap resin Ni ions are being stripped off the resin, at least in my hands. Did any of you have similar experience and if so what kind of work-around was found? Volume is fairly large (3L) and concentration/dialysis have proven to cause loss of desired protein. Please share your positive experience. I get this with insect cell lysates. The solution: load onto ion exchanger first - not for purification but to get rid of whatever strips Ni2+. Crude wash with 50 mM salt (or as high as your protein allows) followed by step elution with 0.5M salt (very few proteins do not elute at 0.5M) -- load directly onto Ni-NTA. Solves the stripping problem 100%. (If you use step elution, make sure to NOT use weak exchangers or you will have pH shift of 2-3 units). If the protein binds to cation-exchangers at pH 7, even such a crude step results in significant purification in itself. If, for some reason, ion exchangers are not an option, load onto hydroxyapatite and elute with phosphate. The downside to HA is that it has a lot less capacity than ion-exchangers and that it needs frequent repacking. Good luck, Dima 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. 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] how to improve Rfree?
Two very low-brow suggestions that might help: 1. Do a round of idealization in Refmac, to tighten up the geometry as much as possible. Then return to your current approach. 2. Try using phenix.refine (with the same mtz, test reflections, latest PDB, etc). The difference can sometimes be rather dramatic, especially after simulated annealing. Best of luck. Bill On Mon, October 19, 2009 5:49 am, Sylvia Fanucchi wrote: Dear all I have been refining my structure using refmac5 restrained refinement and the model currently has an R-factor of 0.19. What is concerning me is that there is quite a big discrepancy between the Rfree and the R-factor, with the Rfree being around 0.27. Furthermore, after about 10 rounds of refinement, the R-free actually starts to increase slightly although the R-factor continues to decrease. Does this mean that there is significant bias in my model? Does anybody know how I could go about addressing this problem? Thanks for any help :-) Sylvia Fanucchi Ph.D Protein Structure-Function Research Unit East Campus, Gate House Room 416 School of Molecular and Cell Biology University of the Witwatersrand Johannesburg 2050 South Africa Tel: +27 (11) 717-6348 Fax: +27 (11) 717-6351 E-mail: sylvia.fanuc...@wits.ac.za mailto:sylvia.fanuc...@wits.ac.za htmlpfont face = verdana size = 0.8 color = navyThis communication is intended for the addressee only. It is confidential. If you have received this communication in error, please notify us immediately and destroy the original message. You may not copy or disseminate this communication without the permission of the University. Only authorized signatories are competent to enter into agreements on behalf of the University and recipients are thus advised that the content of this message may not be legally binding on the University and may contain the personal views and opinions of the author, which are not necessarily the views and opinions of The University of the Witwatersrand, Johannesburg. All agreements between the University and outsiders are subject to South African Law unless the University agrees in writing to the contrary./font/p/html William G. Scott Contact info: http://chemistry.ucsc.edu/~wgscott/
Re: [ccp4bb] crystallography teaching advice: f(S) ?
On Oct 7, 2009, at 1:56 AM, Kevin Cowtan wrote: William G. Scott wrote: On Oct 6, 2009, at 1:32 AM, Morten Kjeldgaard wrote: teleports the students across the hermeneutic circle ;-) (As a consequence, I recommended the postmodernism generator website to the students: http://www.elsewhere.org/pomo/ ) It is easy to mock postmodernism, Well, there is a reason for that ... but it needs to be treated seriously. Why? Do historians likewise need to take Holocaust deniers seriously? It is based on a valid set of critiques of the modern paradigm, That is highly debatable. Even the idea that there is such thing as a paradigm (modern, post-modern, or otherwise) is only an assertion (The idea originates from Thomas Kuhn, who started as a PhD physics student, presumably doing normal science and generalized when he defected to the sociology department. Social sciences slavishly aping what they wrongly perceived to be the aims and the methods of the physical sciences immediately latched on to the idea of paradigms, but that hardly translates into universal acceptance of the idea, especially amongst scientists.) some of which have arisen from within science (notably the cognitive sciences, complex systems and QM). I don't know much about the cognitive sciences beyond what Chomsky did, and I know nothing about complex systems. I do know a little bit about QM, however, and it appears to me that the supposed need to revise our ideas about reality are based on a confusion about what probability is, along with an implicit rejection of indeterminism. Many of the classic problems don't arise if you don't posit a wave- particle duality, but instead ascribe primary reality to particles. (Chemists, not just physicists, deal with these supposed riddles all the time -- how can an unoccupied molecular orbital determine the stereochemistry of a reaction product if there are no electrons in it? -- but their world view isn't exactly placed in mortal danger of collapsing as a consequence). While pomo reacts against the problems in the modern worldview, and in doing so overreacts going off into fantasy land, any useful 21st century philosophy of science needs to take the critiques of the modern worldview - which itself has been significantly shaped by the scientific revolution - very seriously indeed, otherwise it will end up being irrelevant. If scientists remain entrenched in the broken modern paradigm, Again, I have yet to see any compelling evidence that it is at all broken. (The neo-liberal economic system that claims to be informed by this stuff and produces the likes of Tony Blair is another story, however.) they will be increasingly unable to communicate with the outside world, and the pomo paradigm shift may become more deeply anti- science. The failure of many scientists and scientific communicators to take an interest in philosophy of science and sociology In my case at least, it isn't a question of failing to take an interest, but rather a complete sense of frustration when trying to communicate with people who insist on speaking utter jibberish. The point of the postmodernism generator website is to show just how mechanical, arbitrary, and ultimately how vacuous this stuff really can be. On a more serious level, we have the case of Sokal's Social Text Affair. Sokal is a physicist who published a paper in the prestigious journal Social Text entitled Transgressing the Boundaries: Toward a Transformative Hermeneutics of Quantum Gravity. The paper was a total spoof, but the editors and reviewers (none of which were qualified to pass judgement on quantum gravity) decided it was a great achievement and worthy of publication. Then when he admitted the hoax, all hell broke loose. cf: http://www.physics.nyu.edu/faculty/sokal/ have been a significant handicap in the countering of arguments from the creationists, IDers, and climate change deniers, who have (ironically and unwittingly in some cases) tapped into pomo rather more successfully. The pomo suspicion of arguments-from-authority threatens scientific funding and evidence-based policy making at a more general level. However, the modern worldview is broken, only if we start to accept this kind of proof by assertion ... and the pomo paradigm shift may well be a juggernaught. We cannot stop it, we need to both understand it and respond constructively if we are going to advocate and communicate science. The hermenutic circle is one starting point in understanding where pomo comes from. The idea that a text has a meaning is highly problematic and may well be dualistic. Those last two sentences I have to confess kind of frighten and confuse me. I think as scientists, the best we can rationally hope do is to engage the world using simple, straightforward terminology and explicit logical argumentation, both of which