Re: [ccp4bb] CCP4BB Digest - 12 Feb 2009 to 13 Feb 2009 (#2009-45)
Dear Ho, On Fri, Feb 13, 2009 at 04:45:29PM -0800, Ho-Leung Ng wrote: Can you elaborate on the effects of improper inclusion of low resolution (bogus?) reflections? Other than rejecting spots from obvious artifacts, it bothers me to discard data. But I can also see how a few inaccurate, very high intensity spots can throw off scaling. I completely agree: it also bothers me to discard data. However, the crucial word here is 'data' - which is different from Miller indices HKL. So I am mainly concerned with two types of reflections (HKL) that aren't really 'data': 1) overloads These are obviously not included into your final reflection file (unless you explicitely tell the integration software to do that - in which case you know exactly what you are doing anyway). So there is no problem ... or is there? Overloaded reflections are only very few at low resolution - and the most important reflections are obviously the ones at 1.94A resolution so that one can have a 'better-than-2A' structure in the end ... ;-) ... So still no problem, right? And who cares if the completelness of the data isn't 100% but rather 99.4%? Exactly ... so where is the problem? But: these few missing reflections are systematically the strongest ones at low(ish) resolution, and any systematically missing data is not a good thing to have. Solution: always collect a low-intensity pass to measure those strong reflections if there is a substantial amount of overloads. 2) beamstop The integration software will predict all reflections based on your parameters (apart from the 000 reflection): it doesn't care if such a reflection would be behind the beamstop shadow or not. However, a reflection behind the beamstop will obviously not actually be there - and the integrated intensity (probably a very low value) will be wrong. One example of such effects in the context of experimental phasing is bogus anomalous differences. Imagine that your beamstop is not exactly centred around the direct beam. You will have it extending a little bit more to one side (giving you maybe 20A low resolution) than to the other side (maybe 30A resolution). In one orientation of the crystal you might be able to collect a 25A (h,k,l) reflection very well (because it is on the side where the beamstop only starts at 30A) - but the (-h,-k,-l) relfection is collected in an orientation where it is on the 20A-side of the beamstop, i.e. it is predicted within the beamstop shadow. Effect: you have a valid I+ measurement but a more-or-less zero I- measurement, giving you a huge anomalous difference that shouldn't really be there. Now if you measured your data in different orientations (kappa goniostat) with high enough multiplicity, this one bogus measurement will probably be thrown out during scaling/merging. You can e.g. check the so-called ROGUES file produced by SCALA. But if you have the usual multiplicity of only 3-4 the scaling/merging process might not detect this as an outlier correctly and it ends up in your data. Sure, programs like autoSHARP will check for these outliers and try to reject them - but this is only a hack/fix for the fundamental problem: telling the integration program what the good area of the detector is. Solution: mask your beamstop. All integration programs have tools for doing that (some are better than others). I haven't seen any program being able to do it automatically in a reliable way (if reliable would mean: correctly in at least 50% of cases) - but I'm no expert in all of them by a long shot. it usually takes me only about a minute or two for masking the beamstop by hand. A small investment for a big return (good data) ;-) There are other possibly problematic reflections at ice-rings etc: these can also have effects seen in the maps. But the above effects have one thing in common: they happen mainly at low-resolution. And our models can be seen to consist of basically two real-space components (atoms in form of a PDB file and bulk-solvent in form of a mask) - one of which is a low-resolution object (solvent mask) and the other a high-resolution object (atoms). They need to be combined through some clever scaling: if there are issues with the low-resolution reflections this scaling can go wrong - sometimes really badly. Hope that helps a bit. Cheers Clemens ho UC Berkeley Date:Fri, 13 Feb 2009 17:14:38 + From:Clemens Vonrhein vonrh...@globalphasing.com Subject: Re: unstable refinement * resolution limits: are you suddenly including all those poorly measured or non existent reflections at the low resolution end (10A and lower) that are only present because the beamstop masking wasn't don properly during data
Re: [ccp4bb] CCP4BB Digest - 12 Feb 2009 to 13 Feb 2009 (#2009-45)
Just to expand a little on the beam stop problem: the outlier rejection algorithm in Scala ( I imagine in other programs) relies on a consensus, that is it essentially assumes that the majority of observations are correct (actually they are weighted by 1/ EstimatedVariance). This means that if you have 3 observations of a reflection behind the beam stop, one not, then the program is likely to throw out the one good one keep the 3 bad ones. It's hard to think of a good algorithm which would do the Right Thing (maybe we should assume that for refelctions say 20Å resolution the strong ones are right, but I'm not sure this wouldn't cause worse problems) So tell the integration program where the beam stop is! Phil On 16 Feb 2009, at 09:07, Clemens Vonrhein wrote: Dear Ho, On Fri, Feb 13, 2009 at 04:45:29PM -0800, Ho-Leung Ng wrote: Can you elaborate on the effects of improper inclusion of low resolution (bogus?) reflections? Other than rejecting spots from obvious artifacts, it bothers me to discard data. But I can also see how a few inaccurate, very high intensity spots can throw off scaling. 2) beamstop The integration software will predict all reflections based on your parameters (apart from the 000 reflection): it doesn't care if such a reflection would be behind the beamstop shadow or not. However, a reflection behind the beamstop will obviously not actually be there - and the integrated intensity (probably a very low value) will be wrong. One example of such effects in the context of experimental phasing is bogus anomalous differences. Imagine that your beamstop is not exactly centred around the direct beam. You will have it extending a little bit more to one side (giving you maybe 20A low resolution) than to the other side (maybe 30A resolution). In one orientation of the crystal you might be able to collect a 25A (h,k,l) reflection very well (because it is on the side where the beamstop only starts at 30A) - but the (-h,-k,-l) relfection is collected in an orientation where it is on the 20A-side of the beamstop, i.e. it is predicted within the beamstop shadow. Effect: you have a valid I+ measurement but a more-or-less zero I- measurement, giving you a huge anomalous difference that shouldn't really be there. Now if you measured your data in different orientations (kappa goniostat) with high enough multiplicity, this one bogus measurement will probably be thrown out during scaling/merging. You can e.g. check the so-called ROGUES file produced by SCALA. But if you have the usual multiplicity of only 3-4 the scaling/merging process might not detect this as an outlier correctly and it ends up in your data. Sure, programs like autoSHARP will check for these outliers and try to reject them - but this is only a hack/fix for the fundamental problem: telling the integration program what the good area of the detector is. Solution: mask your beamstop. All integration programs have tools for doing that (some are better than others). I haven't seen any program being able to do it automatically in a reliable way (if reliable would mean: correctly in at least 50% of cases) - but I'm no expert in all of them by a long shot. it usually takes me only about a minute or two for masking the beamstop by hand. A small investment for a big return (good data) ;-)
[ccp4bb] Technician position available at IMP - Vienna, AUSTRIA
Dear All, A LABORATORY TECHNICIAN position is open at the Research Institute of Molecular Pathology (IMP) in Vienna, Austria, in the Group of Dr. Peggy Stolt-Bergner (http://www.imp.ac.at/research/peggy-stolt-bergner/) We are seeking a scientific Technician experienced in molecular biology and protein analysis. You will be engaged in fundamental research within different projects involving structural biology and protein biochemistry. You will be involved in the following areas: ♦ MOLECULAR BIOLOGY: Cloning of DNA, expression and purification of recombinant proteins in bacteria and/or yeast. ♦ PROTEIN CHEMISTRY: 2D gel analysis, Western Blotting, FPLC and HPLC, biophysical measurements for protein characterization, protein crystallization. ♦ DATA MANAGEMENT: Handling different programs for data evaluation. Use of basic computer programs such as spreadsheets, graphing, and word processing. The ideal candidate has comprehensive experience in molecular biology and protein purification, and a fundamental knowledge of protein chemistry. Enthusiasm and motivation, the ability to work as part of a team, and excellent organizational skills are also necessary. Experience with protein crystallization and/or X-ray crystallography is an advantage, but is not required. Further training will be provided on-site. The IMP is a basic research institute focused on understanding biological mechanisms at the molecular and cellular level, and offers state-of-the-art research facilities and a vibrant, international scientific community. The working language is English. Closing Date: March 15, 2009 Please forward your CV, including the names of 2-3 references, to: Dr. Peggy Stolt-Bergner Research Institute of Molecular Pathology Dr. Bohr-gasse 7 A-1030 Vienna Austria Email: st...@imp.ac.at
[ccp4bb] Off-topic: ligand enrichment
Dear guys, Sorry for the off-topic question. After I have solved my strucutre, I have found my target ligand bound at the potential binding site. Also, I have found that there are two more ligand molecules bound along the path from solvent to the binding site. I think this can enrich the ligand to binding site, enhancing the local concentration of the ligand, thus reducing the Km of the ligand. I am wondering if anybody can give some suggestions on how to solve this problem clearly. If there is any similar case, it will be better. Thank you in advance. Best wishes, Yingjie Yingjie PENG, Ph.D. student Structural Biology Group Shanghai Institute of Biochemistry and Cell Biology (SIBCB) Shanghai Institute of Biological Sciences (SIBS) Chinese Academy of Sciences (CAS) 320 Yue Yang Road, Shanghai 200031 P. R. China 86-21-54921117 Email: yjp...@sibs.ac.cn
Re: [ccp4bb] PEG3350-based cryoprotectant
Hi Keith, I freezed my crystals with 35% PEG4000 as cryo. They grow in 16% PEG4000, 4% Isopropanol, 0.1 M Na-Acetate. I just put the hanging drop over 35% PEG4000, 4% Isopropanol, 0.1 M Na-Acetate and let it equilibrate over night. They diffracted even better than with other cryos - might be an additional shrinking effect. All the best Alfred _ From: CCP4 bulletin board [mailto:ccp...@jiscmail.ac.uk] On Behalf Of Keith Romano Sent: Sunday, February 15, 2009 3:15 AM To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] PEG3350-based cryoprotectant Hi all, I have protein crystals in complex with substrate grown in 20-30% (w/v) PEG 3350, 4% (w/v) ammonium sulfate, and 0.1M sodium MES buffer at pH 6.5. I purify and concentrate my protein in a high salt buffer (0.5M NaCl, 0.1M sodium MES at pH 6.5, 10% glycerol, 2mM DTT). I grow my crystals with vapor diffusion in 24-well format by hanging a drop of equal volume protein and precipitant solution over the reservoir of precipitant solution. Interestingly, when I do the math, the initial osmolarity of my drop is greater than that of the reservoir (due to the high NaCl in the protein solution). As far as I know, this runs against the principles of the vapor diffusion method, as vapor will leave the reservoir and enter the drop... Nevertheless, these conditions yield giant, rod-like crystals over 1mm long. However, they don't react well to direct flash freezing- the spots tend to smear and indexed refinement leads to high mosaicity. I have tried many cryoprotectant solutions by making up the given precipitant solution with 15-25% glycerol or ethylene glycol, including a range from 0mM to 350mM NaCl. In general, dipping the crystals in cryoprotectant improves the diffraction and lessens the spot smearing. However, the diffraction usually becomes highly twinned and hard to index. After transfer into the cryoprotectant, the crysatls appear to crack and often break apart, as observed under the microscope. It seems like my crystals are very sensitive to the osmotic/ionic change when transferred to the cryoprotectant. I have been unable to find a stable cryoprotectant, and I am wondering if anyone has had similar experience with high-weight PEGs and could suggest some cryoprotectants to try out. Any input would be greatly appreciated! Keith Department of Biochemistry Molecular Pharmacology 970L Lazare Research Building University of Massachusetts Medical School 364 Plantation Street Worcester, MA 01605
Re: [ccp4bb] unstable refinement
Clemens, I know we've had this discussion several times before, but I'd like to take you up on the point you made that reducing Rfree-R is necessarily always a 'good thing'. Suppose the refinement had started from a point where Rfree was biased, e.g. the test set in use had previously been part of the working set, so that Rfree-R was too small. In that case one would hope and indeed expect that Rfree-R would increase on further refinement now excluding the test set. Shouldn't the criterion be that Rfree-R should attain its expected value (dependent of course on the observation/parameter ratio and the weighting parameters), so a high value of |(Rfree-R) - Rfree-R| is bad, i.e. any significant deviations of (Rfree-R) from its expectation are bad? I would go further than that and say that anyway Rfree is meaningless unless the refinement has converged, i.e. reached its maximum (local or global) total likelihood (i.e. data+restraints). So one simply cannot compare the Rfree (or Rfree-R) values at the beginning and end of a run. The purpose of Rfree (or better free likelihood) is surely to compare the *results* of *different* runs where convergence has been attained and where the *refinement protocol* (i.e. selection of parameters to vary and weighting parameters) has been varied, and then to choose as the optimal protocol (and therefore optimal result) the one that gave the lowest Rfree (or highest free likelihood). Rfree-R is then used as a subsidiary test to verify that it has attained its expected value, if not then something is wrong, i.e. either the refinement didn't converge (Rfree-R lower than Rfree-R) or there are non-random errors (Rfree-R higher than Rfree-R), or a combination of factors. Cheers -- Ian -Original Message- From: owner-ccp...@jiscmail.ac.uk [mailto:owner-ccp...@jiscmail.ac.uk] On Behalf Of Clemens Vonrhein Sent: 13 February 2009 17:15 To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] unstable refinement * you don't mention if the R and Rfree move up identically - or if you have a faster increase in R than in Rfree, which would mean that your R-factors are increasing (bad I guess) but your Rfree-R gap is closing down (good). So moving from R/Rfree=0.20/0.35 to R/Rfree=0.32/37 is different than moving from R/Rfree=0.20/0.25 to R/Rfree=0.23/0.28. 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] Refinement
Hi all I have two datasets of resolutions 1.6 and 1.65 Å both of the same molecule, the problem that i am facing is the refinement. The R factors are stuck at very high values 0.3329 and 0.3791 after restrained refinement, although the the map fits into the electron density very well. Regards, Rana
[ccp4bb] Detector pitch, roll, yaw
Dear bb, can somebody explain to me the exact definitions for detector roll, pitch and yaw? Can they be detector dependent or rather dependent on the beamline setup? Is there a relationship with the goniometer position? Many thanks in advance, Jonathan -- Jonathan Elegheert Ph.D. Student Unit for Structural Biology Biophysics http://www.lprobe.ugent.be/xray.html Lab for Protein Biochemistry and Biomolecular Engineering Department of Biochemistry Microbiology Ghent University, Belgium e-mail: jonathan.eleghe...@ugent.be
Re: [ccp4bb] unstable refinement
Dear Ian, That was in fact one of my reasons for only calculating the free R at the end of a SHELXL refinement run (the other reason, now less important, was to save some CPU time). I have to add that I am no longer completely convinced that I made the right decision all those years ago. A stable refinement in which R decreases but Rfree goes through a minimum and then starts to rise might be a useful indication of overfitting?! Best wishes, George Prof. George M. Sheldrick FRS Dept. Structural Chemistry, University of Goettingen, Tammannstr. 4, D37077 Goettingen, Germany Tel. +49-551-39-3021 or -3068 Fax. +49-551-39-22582 On Mon, 16 Feb 2009, Ian Tickle wrote: Clemens, I know we've had this discussion several times before, but I'd like to take you up on the point you made that reducing Rfree-R is necessarily always a 'good thing'. Suppose the refinement had started from a point where Rfree was biased, e.g. the test set in use had previously been part of the working set, so that Rfree-R was too small. In that case one would hope and indeed expect that Rfree-R would increase on further refinement now excluding the test set. Shouldn't the criterion be that Rfree-R should attain its expected value (dependent of course on the observation/parameter ratio and the weighting parameters), so a high value of |(Rfree-R) - Rfree-R| is bad, i.e. any significant deviations of (Rfree-R) from its expectation are bad? I would go further than that and say that anyway Rfree is meaningless unless the refinement has converged, i.e. reached its maximum (local or global) total likelihood (i.e. data+restraints). So one simply cannot compare the Rfree (or Rfree-R) values at the beginning and end of a run. The purpose of Rfree (or better free likelihood) is surely to compare the *results* of *different* runs where convergence has been attained and where the *refinement protocol* (i.e. selection of parameters to vary and weighting parameters) has been varied, and then to choose as the optimal protocol (and therefore optimal result) the one that gave the lowest Rfree (or highest free likelihood). Rfree-R is then used as a subsidiary test to verify that it has attained its expected value, if not then something is wrong, i.e. either the refinement didn't converge (Rfree-R lower than Rfree-R) or there are non-random errors (Rfree-R higher than Rfree-R), or a combination of factors. Cheers -- Ian -Original Message- From: owner-ccp...@jiscmail.ac.uk [mailto:owner-ccp...@jiscmail.ac.uk] On Behalf Of Clemens Vonrhein Sent: 13 February 2009 17:15 To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] unstable refinement * you don't mention if the R and Rfree move up identically - or if you have a faster increase in R than in Rfree, which would mean that your R-factors are increasing (bad I guess) but your Rfree-R gap is closing down (good). So moving from R/Rfree=0.20/0.35 to R/Rfree=0.32/37 is different than moving from R/Rfree=0.20/0.25 to R/Rfree=0.23/0.28. 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] moving the cofactor in model
i have a density map and a model and i want to place the cofactor. i use wincoot and find the electron density for the cofactor but... how can i move the cofactor and play with it in this area?
Re: [ccp4bb] Refinement
Hi Rana, Your email contains very little information which might help finding the reason for the high R-values - the resolution itself does not mean your data are of good quality. What is the completeness (overall and high resolution shell), what are Rint/ Rsym and I/sigma? Maybe you are including a lot of noise. - how did you solve the phase problem? MR or experimental phasing, i.e. could it be that you model is wrong? What's the Rfree value? - is your model complete, or are parts missing because there is no density? - have you tried arp/warp to get to a better model? - are you confident the space group is correct? Maybe the data are twinned? and probably many more issues one might consider. Tim -- Tim Gruene Institut fuer anorganische Chemie Tammannstr. 4 D-37077 Goettingen GPG Key ID = A46BEE1A On Mon, 16 Feb 2009, Rana Refaey wrote: Hi all I have two datasets of resolutions 1.6 and 1.65 ? both of the same molecule, the problem that i am facing is the refinement. The R factors are stuck at very high values 0.3329 and 0.3791 after restrained refinement, although the the map fits into the electron density very well. Regards, Rana
Re: [ccp4bb] Detector pitch, roll, yaw
Hi, this is probably dependent on the integration program you are using - there might not be one universal definition... did you compare the programs' documentation? Tim -- Tim Gruene Institut fuer anorganische Chemie Tammannstr. 4 D-37077 Goettingen GPG Key ID = A46BEE1A On Mon, 16 Feb 2009, jonathan elegheert wrote: Dear bb, can somebody explain to me the exact definitions for detector roll, pitch and yaw? Can they be detector dependent or rather dependent on the beamline setup? Is there a relationship with the goniometer position? Many thanks in advance, Jonathan -- Jonathan Elegheert Ph.D. Student Unit for Structural Biology Biophysics http://www.lprobe.ugent.be/xray.html Lab for Protein Biochemistry and Biomolecular Engineering Department of Biochemistry Microbiology Ghent University, Belgium e-mail: jonathan.eleghe...@ugent.be
Re: [ccp4bb] moving the cofactor in model
Elad, If it is only moving the coordinates of the co-factor and placing it in the density, once you load the coordinate file for your co-factor into the wincoot, it is pretty much drag and rotate process by using the option Rotate/Translate Zone But if your question is with respect to playing with the geometry of the co-factor, you need to generate/download the cif file and read it into the coot. You can refer to the previous ccp4 previous thread given below with respect to this. http://www.mail-archive.com/ccp4bb@jiscmail.ac.uk/msg07769.html Anthony On Mon, 16 Feb 2009 11:26:35 +, Elad Binshtein wrote i have a density map and a model and i want to place the cofactor. i use wincoot and find the electron density for the cofactor but... how can i move the cofactor and play with it in this area? - Anthony Addlagatta, Ph.D. Ramanujan Fellow and Senior Scientist Center for Chemical Biology Indian Institute of Chemical Technology [IICT] Tarnaka, Hyderabad- 57, INDIA Tel:+91-40-27191583 Url: http://www.iictindia.org/zacb/Dr.%20Anthony.aspx
[ccp4bb] Questions about (possibly) twinned data
Hello all, we have a dataset collected from multiple (2 or 3) parts of the same crystal with a microbeam (20 micron). The merged data scales OK (not great) in monoclinic (1-3% rejections). The resolution is 3.2-3.3 A, so the data is not fantastic. This is the cell (similar for other datasets): Cell: 70.012 126.449 107.98890.00089.94690.000 p21 Processing in orthorhombic makes the scaling a lot worse, so I'm assuming its monoclinic for now. Running xtriage gives the following summary: --- Twinning and intensity statistics summary (acentric data): Statistics independent of twin laws - I^2/I^2 : 1.877 - F^2/F^2 : 0.834 - |E^2-1| : 0.663 - |L|, L^2: 0.411, 0.235 Multivariate Z score L-test: 6.737 The multivariate Z score is a quality measure of the given spread in intensities. Good to reasonable data are expected to have a Z score lower than 3.5. Large values can indicate twinning, but small values do not necessarily exclude it. Statistics depending on twin laws - | Operator | type | R obs. | Britton alpha | H alpha | ML alpha | - | h,-k,-l | PM | 0.167 | 0.367 | 0.339 | 0.152| - Patterson analyses - Largest peak height : 5.962 (corresponding p value : 0.72096) The largest off-origin peak in the Patterson function is 5.96% of the height of the origin peak. No significant pseudotranslation is detected. So, I'm assuming that these crystals are monoclinic and that they are pseudo-merohedrally twinned. Is this a reasonable assumption? I get a decent solution for the P21 data from molecular replacement with a 50% identical model (LLG 900, with the rotation Z-scores low (4-5), but the corresponding translation Z-scores high (8-20)). My questions are: what would be the best way to refine? More specifically, what twin fraction should be used as the different tests give different fractions. Is the twin fraction automatically determined in phenix.refine or does this need to be specified? Finally, can twinning be responsible for the fact that the data do not scale well (using data collected on different parts of the same crystal)? Any hints appreciated! Cheers, Bert Bert van den Berg University of Massachusetts Medical School Program in Molecular Medicine Biotech II, 373 Plantation Street, Suite 115 Worcester MA 01605 Phone: 508 856 1201 (office); 508 856 1211 (lab) e-mail: bert.vandenb...@umassmed.edu http://www.umassmed.edu/pmm/faculty/vandenberg.cfm
Re: [ccp4bb] unstable refinement
Dear George I would still maintain that values of Rfree where the refinement had not attained convergence are totally uninformative, so I would say you made the right call! During a refinement run, Rfree is often observed to fall initially and then increase towards the end, though usually not significantly. One cannot deduce anything from this behaviour, and indeed it is not at all surprising: since Rfree is not the target function of the optimisation (or even correlated with it) there's no reason why it should do anything in particular. Exactly the same applies to Rwork: because it's a completely different function from the target function (it contains no weighting information for one thing), there's absolutely no reason why Rwork should be a minimum at convergence (even in the case of unrestrained refinement, and even though it surely is correlated with the target function). If that were true we would be able to use Rwork as the target function! The test for overfitting can only be done if you have at least 2 refinement runs done with different protocols (e.g. no of waters added) to compare: the one with the higher Rfree (or lower free likelihood) at convergence is overfitted. Note that this is a relative test: you can never be sure that a particular model is not overfitted. It's always possible for someone to come along in the future using a different parameter set (or different weighting) and produce a lower Rfree than you did (using the same data of course), making your model overfitted after the fact! Cheers -- Ian -Original Message- From: George M. Sheldrick [mailto:gshe...@shelx.uni-ac.gwdg.de] Sent: 16 February 2009 11:24 To: Ian Tickle Cc: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] unstable refinement Dear Ian, That was in fact one of my reasons for only calculating the free R at the end of a SHELXL refinement run (the other reason, now less important, was to save some CPU time). I have to add that I am no longer completely convinced that I made the right decision all those years ago. A stable refinement in which R decreases but Rfree goes through a minimum and then starts to rise might be a useful indication of overfitting?! Best wishes, George Prof. George M. Sheldrick FRS Dept. Structural Chemistry, University of Goettingen, Tammannstr. 4, D37077 Goettingen, Germany Tel. +49-551-39-3021 or -3068 Fax. +49-551-39-22582 On Mon, 16 Feb 2009, Ian Tickle wrote: Clemens, I know we've had this discussion several times before, but I'd like to take you up on the point you made that reducing Rfree-R is necessarily always a 'good thing'. Suppose the refinement had started from a point where Rfree was biased, e.g. the test set in use had previously been part of the working set, so that Rfree-R was too small. In that case one would hope and indeed expect that Rfree-R would increase on further refinement now excluding the test set. Shouldn't the criterion be that Rfree-R should attain its expected value (dependent of course on the observation/parameter ratio and the weighting parameters), so a high value of |(Rfree-R) - Rfree-R| is bad, i.e. any significant deviations of (Rfree-R) from its expectation are bad? I would go further than that and say that anyway Rfree is meaningless unless the refinement has converged, i.e. reached its maximum (local or global) total likelihood (i.e. data+restraints). So one simply cannot compare the Rfree (or Rfree-R) values at the beginning and end of a run. The purpose of Rfree (or better free likelihood) is surely to compare the *results* of *different* runs where convergence has been attained and where the *refinement protocol* (i.e. selection of parameters to vary and weighting parameters) has been varied, and then to choose as the optimal protocol (and therefore optimal result) the one that gave the lowest Rfree (or highest free likelihood). Rfree-R is then used as a subsidiary test to verify that it has attained its expected value, if not then something is wrong, i.e. either the refinement didn't converge (Rfree-R lower than Rfree-R) or there are non-random errors (Rfree-R higher than Rfree-R), or a combination of factors. Cheers -- Ian -Original Message- From: owner-ccp...@jiscmail.ac.uk [mailto:owner-ccp...@jiscmail.ac.uk] On Behalf Of Clemens Vonrhein Sent: 13 February 2009 17:15 To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] unstable refinement * you don't mention if the R and Rfree move up identically - or if you have a faster increase in R than in Rfree, which would mean that your R-factors are increasing (bad I guess) but your Rfree-R gap is closing down (good). So moving from R/Rfree=0.20/0.35 to R/Rfree=0.32/37 is different than moving from R/Rfree=0.20/0.25 to R/Rfree=0.23/0.28. Disclaimer This communication is confidential and may contain privileged information
Re: [ccp4bb] Detector pitch, roll, yaw
Hi Jonathan, Pitch, Roll and Yaw come from flight dynamics, see e.g. http://www.answers.com/topic/pitch-yaw-roll. With X parallel to the X-ray beam, Z pointing to the Zenith and Y being perpendicular to X Y, they correspond to the roty, rotz and rotx of the detector with respect to the primary beam. So, they are dependent on the beam line setup. In principle, the pitch roll and yaw should be independent of omega, kappa (or chi for that matter), phi, theta-swing or detector distance. In theory there could be a difference between the Pitch, Roll and Yaw at small and at large distance, e.g. if your detector track is bent or curved. I have not seen this in practice with the equipment I most commonly use. As far as I know, there is one integration program that I know of (SAINT), that uses this definition of detector-rotations. Best wishes, Bram jonathan elegheert wrote: Dear bb, can somebody explain to me the exact definitions for detector roll, pitch and yaw? Can they be detector dependent or rather dependent on the beamline setup? Is there a relationship with the goniometer position? Many thanks in advance, Jonathan -- Dr. Bram Schierbeek Application Scientist Structural Biology Bruker AXS B.V. Oostsingel 209, P.O.Box 811 2600 AV Delft, the Netherlands Tel.: +31 (15) 2152508 Fax: +31 (15)2152599 bram.schierb...@bruker-axs.nl www.bruker-axs.com
Re: [ccp4bb] Off-topic: ligand enrichment
Yingjie Peng wrote: .. After I have solved my strucutre, I have found my target ligand bound at the potential binding site. Also, I have found that there are two more ligand molecules bound along the path from solvent to the binding site. I think this can enrich the ligand to binding site, enhancing the local concentration of the ligand, thus reducing the Km of the ligand. I've heard this kind of explanation for alternate binding sites before, but I am skeptical. To the extent that the bound ligands are in equilibrium with the bulk phase, the local activity of the ligand will be the same as in the bulk phase- i.e. the bound ligands don't count in figuring the effective concentration. If anything they lower the activity, competing with the active site if the concentration of ligand is not [enzyme]. There would be a local buffering effect, so if the enzyme is gated by a nerve impulse or absorption of a quantum of light so that it is usually inactive and turns on suddenly, the local binding sites could release their load in response to the local depletion faster than ligand could diffuse in from the bulk- Then during the next off period all the local binding sites recharge. This would be like the function of a bypass capacitor in a digital electronic circuit. But during steady-state turnover I don't see how the bound ligand could help any. It may be easy to get ligand in physiologically irrelevant low-affinity sites due to the high concentration of protein in crystallization experiments. The protein is a good fraction of 1 mM, whereas physiologically important binding sites are often in the nM to uM range. So if you add a 3-fold excess of your ligand, and one equivalent binds at the specific site, there will still be 100 uM or so free ligand, which may bind at low affinity non-specific sites. Whether this loosely bound ligand will be well-ordered enough to identify in the density is another question. Just my thoughts on the matter, Ed
Re: [ccp4bb] Off-topic: ligand enrichment
Dear Yingjie, I agree with Ed Berry that I do not believe that nearby binding sites influence the Km (~Kd) which depend on bound and unbound concentrations. However, there could be a strong kinetic effect, e.g. these secondary binding sites could act as stepping stones when the path to the primary binding site would otherwise be difficult to pass. The crystal structure of the potassium channel provides a beautiful example of this. Best regards, Herman From: CCP4 bulletin board [mailto:ccp...@jiscmail.ac.uk] On Behalf Of Yingjie Peng Sent: Monday, February 16, 2009 11:09 AM To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] Off-topic: ligand enrichment Dear guys, Sorry for the off-topic question. After I have solved my strucutre, I have found my target ligand bound at the potential binding site. Also, I have found that there are two more ligand molecules bound along the path from solvent to the binding site. I think this can enrich the ligand to binding site, enhancing the local concentration of the ligand, thus reducing the Km of the ligand. I am wondering if anybody can give some suggestions on how to solve this problem clearly. If there is any similar case, it will be better. Thank you in advance. Best wishes, Yingjie Yingjie PENG, Ph.D. student Structural Biology Group Shanghai Institute of Biochemistry and Cell Biology (SIBCB) Shanghai Institute of Biological Sciences (SIBS) Chinese Academy of Sciences (CAS) 320 Yue Yang Road, Shanghai 200031 P. R. China 86-21-54921117 Email: yjp...@sibs.ac.cn
Re: [ccp4bb] Questions about (possibly) twinned data
Hi Bert, It seems unikely you are experiencing merohedral twinning in your crystal since none of your unit cell dimensions are equal length or integer multiples. For your cell, you would expect to see multiple lattices. Is it possible you have a dimer in the asymmetric unit? Strong NCS parallel to a principle lattice direction can sometimes give twin-like statistics especially at lower resolutions. Hope this helps, Chris On Mon, 16 Feb 2009, Van Den Berg, Bert wrote: Hello all, we have a dataset collected from multiple (2 or 3) parts of the same crystal with a microbeam (20 micron). The merged data scales OK (not great) in monoclinic (1-3% rejections). The resolution is 3.2-3.3 A, so the data is not fantastic. This is the cell (similar for other datasets): Cell: 70.012 126.449 107.98890.00089.94690.000 p21 Processing in orthorhombic makes the scaling a lot worse, so I'm assuming its monoclinic for now. Running xtriage gives the following summary: --- Twinning and intensity statistics summary (acentric data): Statistics independent of twin laws - I^2/I^2 : 1.877 - F^2/F^2 : 0.834 - |E^2-1| : 0.663 - |L|, L^2: 0.411, 0.235 Multivariate Z score L-test: 6.737 The multivariate Z score is a quality measure of the given spread in intensities. Good to reasonable data are expected to have a Z score lower than 3.5. Large values can indicate twinning, but small values do not necessarily exclude it. Statistics depending on twin laws - | Operator | type | R obs. | Britton alpha | H alpha | ML alpha | - | h,-k,-l | PM | 0.167 | 0.367 | 0.339 | 0.152| - Patterson analyses - Largest peak height : 5.962 (corresponding p value : 0.72096) The largest off-origin peak in the Patterson function is 5.96% of the height of the origin peak. No significant pseudotranslation is detected. So, I'm assuming that these crystals are monoclinic and that they are pseudo-merohedrally twinned. Is this a reasonable assumption? I get a decent solution for the P21 data from molecular replacement with a 50% identical model (LLG 900, with the rotation Z-scores low (4-5), but the corresponding translation Z-scores high (8-20)). My questions are: what would be the best way to refine? More specifically, what twin fraction should be used as the different tests give different fractions. Is the twin fraction automatically determined in phenix.refine or does this need to be specified? Finally, can twinning be responsible for the fact that the data do not scale well (using data collected on different parts of the same crystal)? Any hints appreciated! Cheers, Bert Bert van den Berg University of Massachusetts Medical School Program in Molecular Medicine Biotech II, 373 Plantation Street, Suite 115 Worcester MA 01605 Phone: 508 856 1201 (office); 508 856 1211 (lab) e-mail: bert.vandenb...@umassmed.edu http://www.umassmed.edu/pmm/faculty/vandenberg.cfm Christopher L. Colbert, Ph.D. InstructorPhone: (214) 645 5944 University of Texas Southwestern Medical Center FAX: (214) 645 5945 6001 Forest Park Lane Dallas, TX 75390
Re: [ccp4bb] Questions about (possibly) twinned data
Bert, Your self-Patterson peak may be real, i.e. you have pseudo translation, which can then make the statistics *look* like the crystal is twinned. Try a self-Patterson (perhaps sharpened) at somewhat lower resolution, e.g 6 A. Maybe the peak is real, but is only 6% of origin due to a slight mis-orientation of the molecules. Dave David Borhani, Ph.D. D. E. Shaw Research, LLC 120 West Forty-Fifth Street, 39th Floor New York, NY 10036 david.borh...@deshawresearch.com 212-478-0698 http://www.deshawresearch.com http://www.deshawresearch.com/ From: CCP4 bulletin board [mailto:ccp...@jiscmail.ac.uk] On Behalf Of Van Den Berg, Bert Sent: Monday, February 16, 2009 9:12 AM To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] Questions about (possibly) twinned data Hello all, we have a dataset collected from multiple (2 or 3) parts of the same crystal with a microbeam (20 micron). The merged data scales OK (not great) in monoclinic (1-3% rejections). The resolution is 3.2-3.3 A, so the data is not fantastic. This is the cell (similar for other datasets): Cell: 70.012 126.449 107.98890.00089.94690.000 p21 Processing in orthorhombic makes the scaling a lot worse, so I'm assuming its monoclinic for now. Running xtriage gives the following summary: --- Twinning and intensity statistics summary (acentric data): Statistics independent of twin laws - I^2/I^2 : 1.877 - F^2/F^2 : 0.834 - |E^2-1| : 0.663 - |L|, L^2: 0.411, 0.235 Multivariate Z score L-test: 6.737 The multivariate Z score is a quality measure of the given spread in intensities. Good to reasonable data are expected to have a Z score lower than 3.5. Large values can indicate twinning, but small values do not necessarily exclude it. Statistics depending on twin laws - | Operator | type | R obs. | Britton alpha | H alpha | ML alpha | - | h,-k,-l | PM | 0.167 | 0.367 | 0.339 | 0.152 | - Patterson analyses - Largest peak height : 5.962 (corresponding p value : 0.72096) The largest off-origin peak in the Patterson function is 5.96% of the height of the origin peak. No significant pseudotranslation is detected. So, I'm assuming that these crystals are monoclinic and that they are pseudo-merohedrally twinned. Is this a reasonable assumption? I get a decent solution for the P21 data from molecular replacement with a 50% identical model (LLG 900, with the rotation Z-scores low (4-5), but the corresponding translation Z-scores high (8-20)). My questions are: what would be the best way to refine? More specifically, what twin fraction should be used as the different tests give different fractions. Is the twin fraction automatically determined in phenix.refine or does this need to be specified? Finally, can twinning be responsible for the fact that the data do not scale well (using data collected on different parts of the same crystal)? Any hints appreciated! Cheers, Bert Bert van den Berg University of Massachusetts Medical School Program in Molecular Medicine Biotech II, 373 Plantation Street, Suite 115 Worcester MA 01605 Phone: 508 856 1201 (office); 508 856 1211 (lab) e-mail: bert.vandenb...@umassmed.edu http://www.umassmed.edu/pmm/faculty/vandenberg.cfm
[ccp4bb] deadline for beamtime proposals during April/May 2009 for BM14/ESRF 20th Feb. 2009
Dear all, deadline for submission of proposals for beamtime at BM14 (ESRF) for the second run of 2009 [April/May] is this Friday, the20th February 2009. Beamline proposals can be submitted on-line from the BM14 webpages at http://www.bm14.eu PLEASE remember this call is distinct from ESRF proposal call and allows you direct access to BM14 -We provide funding for UK, EMBL and EU member state countries for travel and subsistence - special requests can be forwarded directly to wa...@esrf and we will do our best to accommodate you. BEAMLINE FEATURES **MARMOSIAC 225 CCD detector ( http://www.mar-usa.com/products/mx_series.htm http://www.mar-usa.com/products/mx_series.htm) **MD2 Microdiffractometer (http://www.embl-grenoble.fr/groups/instr/MD2-17.pdf) **MiniKappa goniostat (http://www.embl-grenoble.fr/groups/instr/MK2.pdf -videos available here - http://www.embl-grenoble.fr/groups/instr/mk2videos.html ) **Robotic Sample changer ( http://journals.iucr.org/d/issues/2006/10/00/gx5085/index.html http://journals.iucr.org/d/issues/2006/10/00/gx5085/index.html) **Easily Tuneable over the 6.5 - 18 keV range with X-ray flux at sample through a 100micron aperture of ~1.5x10*10 photons/s/200mA synchrotron beam current at 12.7 keV **REMOTE ACCESS for data collection ** HC1 HUMIDITY CONTROL DEVICE (Upon request and subject to availability) as a tool for improving diffraction quality. See: http://www.embl-grenoble.fr/groups/instr/humidifier_page1.html http://www.embl-grenoble.fr/groups/instr/humidifier_page1.html FULL CALL DETAILS follow: SYNCHROTRON BEAM TIME FOR MACROMOLECULAR CRYSTALLOGRAPHY AT THE UK/EMBL MAD BEAMLINE BM14, ESRF Scheduling Period: April/May 2009 DEADLINE for proposals: 20th February 2009 - Call for proposals from EMBL member states, EU states and associated states for Apr/May 2009 synchrotron run at ESRF. The bending magnet MAD beamline BM14 at the ESRF is being operated as a UK Collaborative Research Group(CRG) beamline in collaboration with the EMBL Grenoble Outstation. Proposals for beam time can be made via a user-friendly web based application form, both for monochromatic and multi-wavelength experiments. Beam time is available to groups from the UK and EMBL, EU and EU associated member states. Deadline(20th Feb. 2009) for beamtime proposals at BM14/ESRF for April/May 2009 Reimbursement is available for travel and subsistence for groups based in the UK and EU member or associate member states (see NeedToKnow Link in Main menu on Beamline Homepage for details on reimbursement and other useful information) Full details of the beamline characteristics, the user program and the procedure for the submission of electronic proposals can be found at the beamline homepage http://www.bm14.eu http://www.bm14.eu/ or by following the links from http://ww.embl-grenoble.fr/ http://ww.embl-grenoble.fr or http://www.esrf.eu/ http://www.esrf.eu -- Please don't hesistate to contact me if you have any queries Thanks Martin - Martin Walsh, MRC Group Leader and BM14 Responsible Medical Research Council (MRC) France, CRG BM14, c/o ESRF, 6, rue Jules Horowitz, B.P. 220 38043 Grenoble CEDEX France Tel: +33 4 38.88.19.69 Fax: +33 4 76.88.23.80 email1: mailto:wa...@esrf.fr wa...@esrf.fr email2: mailto:walshb...@gmail.com walshb...@gmail.com
[ccp4bb] PDB protein strucutrues as screen saver
Dear Scientists, It may be too much... But as a biophysics student I would like to appreciate and feel happy to have pdb structures as my computers screen savers than to have some funny and fancy stuffs. And it may help me as a motivator to solve my own structures in future I want to ask is there any existing script that grep strucutres one by one with one line definition of that structure. S.Jayashankar Research Student Institute for Biophysical Chemistry Hannover Medical School Germany.
Re: [ccp4bb] PDB protein strucutrues as screen saver
Hi, You may want to have a look at http://www.luminorum.com/html/luminorum_ltd___extras.html. hth Nadir -- Pr. Nadir T. Mrabet Cellular Molecular Biochemistry INSERM U-724 Nancy University, School of Medicine 9, Avenue de la Foret de Haye, BP 184 54505 Vandoeuvre-les-Nancy Cedex France Phone: +33 (0)3.83.68.32.73 Fax: +33 (0)3.83.68.32.79 E-mail: nadir.mra...@medecine.uhp-nancy.fr Jayashankar wrote: Dear Scientists, It may be too much... But as a biophysics student I would like to appreciate and feel happy to have pdb structures as my computers screen savers than to have some funny and fancy stuffs. And it may help me as a motivator to solve my own structures in future I want to ask is there any existing script that grep strucutres one by one with one line definition of that structure. S.Jayashankar Research Student Institute for Biophysical Chemistry Hannover Medical School Germany.
[ccp4bb] reifine metal with phenix
Hi all, The resolution of my structure is 3.1A. There are three Mg binds in this structure. I try to refine it with phenix. There are cleare extrea density before I add Mg atoms. But when I put Mg in the central of density with good coordination and try to refine it by phenix, those Mg move away.Does phenix need addiational paramters file to refine Mg and keedp it 2.1A 2.1A from it coordinated atom? Thanks. Lisa
Re: [ccp4bb] PDB protein strucutrues as screen saver
Get a Mac, render some images in Pymol and run a slideshow with Ken burns effect if you want. Jürgen On 16 Feb 2009, at 13:22, Jayashankar wrote: Dear Scientists, It may be too much... But as a biophysics student I would like to appreciate and feel happy to have pdb structures as my computers screen savers than to have some funny and fancy stuffs. And it may help me as a motivator to solve my own structures in future I want to ask is there any existing script that grep strucutres one by one with one line definition of that structure. S.Jayashankar Research Student Institute for Biophysical Chemistry Hannover Medical School Germany. - Jürgen Bosch Johns Hopkins Bloomberg School of Public Health Biochemistry and Molecular Biology, W8708 615 North Wolfe Street Baltimore, MD 21205 Phone: +1-410-614-4742 Fax: +1-410-955-3655
Re: [ccp4bb] PDB protein strucutrues as screen saver
But as a biophysics student I would like to appreciate and feel happy to have pdb structures as my computers screen savers than to have some funny and fancy stuffs. And it may help me as a motivator to solve my own structures in future http://74.125.47.132/search?q=cache:_WRPDpKtbtcJ:ubuntuforums.org/showthread.php%3Ft%3D291503+linux+pdb+screen+saverhl=enct=clnkcd=1gl=us Ubuntu forums was down so there's the google cache page. I'm sure you can port the directions to your distro of Linux, if that's what you're running.
[ccp4bb] Looking for a free program for calculating powder diffraction pattern
Hi, I'm looking for a free program for calculating powder diffraction pattern, given a PDB file. I googled for hours and only found a bunch of junks... Thanks for your help! Owen
Re: [ccp4bb] PDB protein strucutrues as screen saver
On Feb 16, 2009, at 10:22 AM, Jayashankar wrote: Dear Scientists, It may be too much... But as a biophysics student I would like to appreciate and feel happy to have pdb structures as my computers screen savers than to have some funny and fancy stuffs. And it may help me as a motivator to solve my own structures in future I want to ask is there any existing script that grep strucutres one by one with one line definition of that structure. S.Jayashankar Research Student Institute for Biophysical Chemistry Hannover Medical School Germany. I've been using this on my ppc G5. It is free. Unfortunately it seems to have a bias toward amino acids: http://www.sourcecod.com/structure/
[ccp4bb] Occupancy Refinement
Dear All, I have a question about the occupancy refinement of a ligand. I have a dataset of 2.3 angstrom and the ligand binds in multiple conformations in the active site. My question is if it is possible to tell which orientation(s) has/have the highest occupancy based on occupancy refinement. What is the best way to refine the occupancy? Thanks, Mariah -- Mariah Jones Department of Biochemistry University of Florida
Re: [ccp4bb] unstable refinement
Dear Ian, I totally agree with your observations and recommendations. If one is concerned about instability of the optimizer (minimization and/or simulated annealing) I suggest to also monitor the value of the total energy function (X-ray maximum likelihood term plus all restraints). Another source for slight variations in R values can occur after recalculation of the bulk solvent mask and model parameters if the model has significantly moved between solvent mask updates. Axel On Feb 16, 2009, at 6:21 AM, Ian Tickle wrote: Dear George I would still maintain that values of Rfree where the refinement had not attained convergence are totally uninformative, so I would say you made the right call! During a refinement run, Rfree is often observed to fall initially and then increase towards the end, though usually not significantly. One cannot deduce anything from this behaviour, and indeed it is not at all surprising: since Rfree is not the target function of the optimisation (or even correlated with it) there's no reason why it should do anything in particular. Exactly the same applies to Rwork: because it's a completely different function from the target function (it contains no weighting information for one thing), there's absolutely no reason why Rwork should be a minimum at convergence (even in the case of unrestrained refinement, and even though it surely is correlated with the target function). If that were true we would be able to use Rwork as the target function! The test for overfitting can only be done if you have at least 2 refinement runs done with different protocols (e.g. no of waters added) to compare: the one with the higher Rfree (or lower free likelihood) at convergence is overfitted. Note that this is a relative test: you can never be sure that a particular model is not overfitted. It's always possible for someone to come along in the future using a different parameter set (or different weighting) and produce a lower Rfree than you did (using the same data of course), making your model overfitted after the fact! Cheers -- Ian -Original Message- From: George M. Sheldrick [mailto:gshe...@shelx.uni-ac.gwdg.de] Sent: 16 February 2009 11:24 To: Ian Tickle Cc: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] unstable refinement Dear Ian, That was in fact one of my reasons for only calculating the free R at the end of a SHELXL refinement run (the other reason, now less important, was to save some CPU time). I have to add that I am no longer completely convinced that I made the right decision all those years ago. A stable refinement in which R decreases but Rfree goes through a minimum and then starts to rise might be a useful indication of overfitting?! Best wishes, George Prof. George M. Sheldrick FRS Dept. Structural Chemistry, University of Goettingen, Tammannstr. 4, D37077 Goettingen, Germany Tel. +49-551-39-3021 or -3068 Fax. +49-551-39-22582 On Mon, 16 Feb 2009, Ian Tickle wrote: Clemens, I know we've had this discussion several times before, but I'd like to take you up on the point you made that reducing Rfree-R is necessarily always a 'good thing'. Suppose the refinement had started from a point where Rfree was biased, e.g. the test set in use had previously been part of the working set, so that Rfree-R was too small. In that case one would hope and indeed expect that Rfree-R would increase on further refinement now excluding the test set. Shouldn't the criterion be that Rfree-R should attain its expected value (dependent of course on the observation/parameter ratio and the weighting parameters), so a high value of |(Rfree-R) - Rfree-R| is bad, i.e. any significant deviations of (Rfree-R) from its expectation are bad? I would go further than that and say that anyway Rfree is meaningless unless the refinement has converged, i.e. reached its maximum (local or global) total likelihood (i.e. data+restraints). So one simply cannot compare the Rfree (or Rfree-R) values at the beginning and end of a run. The purpose of Rfree (or better free likelihood) is surely to compare the *results* of *different* runs where convergence has been attained and where the *refinement protocol* (i.e. selection of parameters to vary and weighting parameters) has been varied, and then to choose as the optimal protocol (and therefore optimal result) the one that gave the lowest Rfree (or highest free likelihood). Rfree-R is then used as a subsidiary test to verify that it has attained its expected value, if not then something is wrong, i.e. either the refinement didn't converge (Rfree-R lower than Rfree-R) or there are non-random errors (Rfree-R higher than Rfree-R), or a combination of factors. Cheers -- Ian -Original Message- From: owner-ccp...@jiscmail.ac.uk [mailto:owner-ccp...@jiscmail.ac.uk] On Behalf Of Clemens Vonrhein Sent: 13 February 2009 17:15 To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] unstable
Re: [ccp4bb] reifine metal with phenix
Dear Lisa, you can specify custom bond and angle restraints by using the option refinement.geometry_restraints.edits. Best is to save them to a file (e.g. restraints_edits.params) and use this as input for your next phenix.refine run. Assuming the ideal distance for your Mg is 2.1A and the amino acid is asp113 of chain A, the according file entry could look like this: refinement.geometry_restraints.edits { bond { action = *add atom_selection_1 = name MG01 and chain G and resname MG and resseq 1 atom_selection_2 = name OD2 and chain A and resname ASP and resseq 113 distance_ideal = 2.1000 sigma = 0.01 } You can check the phenix hompage for detailed documentation: http://phenix-online.org/documentation/refinement.htm Yours, Joern ** Address: Joern Krausze University of Leipzig Centre for Biotechnology and Biomedicine Deutscher Platz 5 04103 Leipzig Germany eMail: krau...@bbz.uni-leipzig.de Phone: +49 (0)341 9731312 Fax:+49 (0)341 9731319 ** On Mon, 16 Feb 2009, Lisa Wang wrote: Hi all, The resolution of my structure is 3.1A. There are three Mg binds in this structure. I try to refine it with phenix. There are cleare extrea density before I add Mg atoms. But when I put Mg in the central of density with good coordination and try to refine it by phenix, those Mg move away.Does phenix need addiational paramters file to refine Mg and keedp it 2.1A 2.1A from it coordinated atom? Thanks. Lisa
[ccp4bb] include or exclude overloads
Hi Clemens, Thank you for the clarification. I had thought you were advocating using a general low resolution cutoff, with which I would disagree. I spend a lot of time troubleshooting data collected and processed by other people. Those are good reminders to go back and check beamstop settings and overloaded spots. Phil, I was thinking specifically of what to do with overloads when a short exposure pass wasn't done, when I asked what should be done with poorly measured data. Exclusion gives you zeroes instead of what should be the highest intensity spots in your data set. But inclusion could throw off scaling, which would be worse. SCALA by default rejects estimated intensities from overloads from mosflm, which I presume is for a very good reason. Would it be better to not use the estimated intensities for scale determination but keep them in the data set? ho UC Berkeley -- Date:Mon, 16 Feb 2009 09:07:38 + From:Clemens Vonrhein vonrh...@globalphasing.com Subject: Re: CCP4BB Digest - 12 Feb 2009 to 13 Feb 2009 (#2009-45) Dear Ho, On Fri, Feb 13, 2009 at 04:45:29PM -0800, Ho-Leung Ng wrote: Can you elaborate on the effects of improper inclusion of low resolution (bogus?) reflections? Other than rejecting spots from obvious artifacts, it bothers me to discard data. But I can also see how a few inaccurate, very high intensity spots can throw off scaling. I completely agree: it also bothers me to discard data. However, the crucial word here is 'data' - which is different from Miller indices HKL. So I am mainly concerned with two types of reflections (HKL) that aren't really 'data': 1) overloads These are obviously not included into your final reflection file (unless you explicitely tell the integration software to do that - in which case you know exactly what you are doing anyway). So there is no problem ... or is there? Overloaded reflections are only very few at low resolution - and the most important reflections are obviously the ones at 1.94A resolution so that one can have a 'better-than-2A' structure in the end ... ;-) ... So still no problem, right? And who cares if the completelness of the data isn't 100% but rather 99.4%? Exactly ... so where is the problem? But: these few missing reflections are systematically the strongest ones at low(ish) resolution, and any systematically missing data is not a good thing to have. Solution: always collect a low-intensity pass to measure those strong reflections if there is a substantial amount of overloads. 2) beamstop The integration software will predict all reflections based on your parameters (apart from the 000 reflection): it doesn't care if such a reflection would be behind the beamstop shadow or not. However, a reflection behind the beamstop will obviously not actually be there - and the integrated intensity (probably a very low value) will be wrong. One example of such effects in the context of experimental phasing is bogus anomalous differences. Imagine that your beamstop is not exactly centred around the direct beam. You will have it extending a little bit more to one side (giving you maybe 20A low resolution) than to the other side (maybe 30A resolution). In one orientation of the crystal you might be able to collect a 25A (h,k,l) reflection very well (because it is on the side where the beamstop only starts at 30A) - but the (-h,-k,-l) relfection is collected in an orientation where it is on the 20A-side of the beamstop, i.e. it is predicted within the beamstop shadow. Effect: you have a valid I+ measurement but a more-or-less zero I- measurement, giving you a huge anomalous difference that shouldn't really be there. Now if you measured your data in different orientations (kappa goniostat) with high enough multiplicity, this one bogus measurement will probably be thrown out during scaling/merging. You can e.g. check the so-called ROGUES file produced by SCALA. But if you have the usual multiplicity of only 3-4 the scaling/merging process might not detect this as an outlier correctly and it ends up in your data. Sure, programs like autoSHARP will check for these outliers and try to reject them - but this is only a hack/fix for the fundamental problem: telling the integration program what the good area of the detector is. Solution: mask your beamstop. All integration programs have tools for doing that (some are better than others). I haven't seen any program being able to do it automatically in a reliable way (if reliable would mean: correctly in at least 50% of cases) - but I'm no expert in all of them by a long shot. it usually takes me only about a minute or
[ccp4bb] Protein crystallization
Hi all I am dealing with a protein which size is about 200kDa. Due to the impurity and degradation problem, the protein has gone thru 3 purification steps (affinity column ion exchange column gel filtration column). The buffer condition for the last step of purification was 50mM MOPS pH7.0, 500mM NaCl, 20% glycerol, and 1mM DTT. Although there are few crystals were observed in different buffer condition, but they are too small and fragile and almost no diffraction at all. I have tried to optimize crystallization condition in 24-well format by hanging drop of equal volume of protein and buffer, but it is not producible. I believed that my protein might have conformational change and I have no idea how to solve it. Please advice Thanks vid
Re: [ccp4bb] PDB protein strucutrues as screen saver
I just emailed the guy today and asked him if there was any hope of getting an intel version in the future, and he wrote back almost immediately and said he is working on it and is about 90% done. The current one runs only on PPC. I tried to hint subtly that it would be kind of cool to have nucleic acids. ... On Feb 16, 2009, at 7:42 PM, Engin Ozkan wrote: Does this screensaver run on 10.5 intel Macs? It looks like it was developed a while ago, and not updated since then. Engin Manish Chandra Pathak wrote: For mac, a screen saver (structure) is already available. moreover it's free and displays couple of structural information exactly the way Jayashankar wants. http://www.sourcecod.com/structure/ I hope their is something like this for Linux/Windows also. *From:* Jürgen Bosch jubo...@jhsph.edu *To:* CCP4BB@JISCMAIL.AC.UK *Sent:* Monday, February 16, 2009 2:44:50 PM *Subject:* Re: [ccp4bb] PDB protein strucutrues as screen saver Get a Mac, render some images in Pymol and run a slideshow with Ken burns effect if you want. Jürgen On 16 Feb 2009, at 13:22, Jayashankar wrote: Dear Scientists, It may be too much... But as a biophysics student I would like to appreciate and feel happy to have pdb structures as my computers screen savers than to have some funny and fancy stuffs. And it may help me as a motivator to solve my own structures in future I want to ask is there any existing script that grep strucutres one by one with one line definition of that structure. S.Jayashankar Research Student Institute for Biophysical Chemistry Hannover Medical School Germany. - Jürgen Bosch Johns Hopkins Bloomberg School of Public Health Biochemistry and Molecular Biology, W8708 615 North Wolfe Street Baltimore, MD 21205 Phone: +1-410-614-4742 Fax: +1-410-955-3655
Re: [ccp4bb] Looking for a free program for calculating powder diffraction pattern
Dear Owen, The GSAS package can do that: http://www.ccp14.ac.uk/solution/gsas/ It is important to put some values for the solvent contribution if you want to get a meaningful pattern. Shout if you need a hand. Good luck, Jon wob wrote: Hi, I'm looking for a free program for calculating powder diffraction pattern, given a PDB file. I googled for hours and only found a bunch of junks... Thanks for your help! Owen