[ccp4bb] low resolution refinement
Hi all, just be sure: does Low Resolution Refinement Pipeline in ccp4 always use Free_R_flag=0? Thanks, s -- Sebastiano Pasqualato, PhD Crystallography Unit Department of Experimental Oncology European Institute of Oncology IFOM-IEO Campus via Adamello, 16 20139 - Milano Italy tel +39 02 9437 5167 fax +39 02 9437 5990
[ccp4bb] low-resolution refinement
Dear all, I'm making first steps in the desolate world of low-resolution refinement. With dodgy 3.8A data, the magic of Phaser was able to solve the structure of a complex by MR with its components as MR models. Jelly-body refinement does wonders for R free. There are three issues that I would like to get some advice on: 1) Using external restraints calculated with ProSMART improved the structure further, but I'm worried that using restraints derived from the structures used for MR gets me into a sinkhole of model bias. Should it be either molecular replacement or homology restraints? 2) Do I recalculate restraints at each round of refinement? In particular, I substantially rebuilt a surface loop that I don't want to restrain by the model. 3) Activating map sharpening results in mtz files that look just normal and open in coot after the typical map calculation break, but no maps are displayed. This is independent of the sharpening factor I choose (between 5 and 60). Thanks for your help. Andreas -- Andreas Förster, Research Associate Paul Freemont Xiaodong Zhang Labs Department of Biochemistry, Imperial College London http://www.msf.bio.ic.ac.uk
Re: [ccp4bb] low-resolution refinement
Hi Andreas, In your case, it sounds like a reasonable strategy would be to use external restraints for a few rounds of refinement (as you have done), but then release them and instead use jelly-body restraints. This two-stage process will help to initially hold your model in a sensible conformation using external restraints, but then gently release the structure in order to reduce further bias in later rounds. The immediate subsequent use of jelly-body restraints after external restraints will ensure that the model won't deviate too far from that sensible conformation, unless the data suggests otherwise. Of course, if certain regions lose their sensible conformations in subsequent rounds of refinement, you can continue to use external restraints just in these regions. I substantially rebuilt a surface loop that I don't want to restrain by the model. In this case, I would recommend re-generating the external restraints, this time telling ProSMART not to generate restraints for these particular residues/regions. This can be done using the -restrain and -restrain_rm keywords, as described in the documentation (let me know off-board if you want help with this). If you enable map sharpening then the single output MTZ file should be the sharpened map… I'm not sure why you are finding that the map is not displayed… do you see any difference between enabling/disabling map sharpening? Cheers, Rob On 25 Sep 2012, at 11:19, Andreas Förster wrote: Dear all, I'm making first steps in the desolate world of low-resolution refinement. With dodgy 3.8A data, the magic of Phaser was able to solve the structure of a complex by MR with its components as MR models. Jelly-body refinement does wonders for R free. There are three issues that I would like to get some advice on: 1) Using external restraints calculated with ProSMART improved the structure further, but I'm worried that using restraints derived from the structures used for MR gets me into a sinkhole of model bias. Should it be either molecular replacement or homology restraints? 2) Do I recalculate restraints at each round of refinement? In particular, I substantially rebuilt a surface loop that I don't want to restrain by the model. 3) Activating map sharpening results in mtz files that look just normal and open in coot after the typical map calculation break, but no maps are displayed. This is independent of the sharpening factor I choose (between 5 and 60). Thanks for your help. Andreas -- Andreas Förster, Research Associate Paul Freemont Xiaodong Zhang Labs Department of Biochemistry, Imperial College London http://www.msf.bio.ic.ac.uk
Re: [ccp4bb] low-resolution refinement
3) Activating map sharpening results in mtz files that look just normal and open in coot after the typical map calculation break, but no maps are displayed. This is independent of the sharpening factor I choose (between 5 and 60). I haven't used coot for map sharpening, but using the ranges I'd usually have to use (in cad) to seen an effect are more in the 80-120 range. You might want to start with an unrealistically large value just to ensure that things are working, then back off to a better value. Pete
Re: [ccp4bb] low resolution refinement
Roberto steiner has told you how to use these new REFMAC5.6 features, Rob Nicholls has suggested how to generate secondary structure restraints, and Martyn Winn given a page to install a new interface to make it easier to use them.. But with such limited data it isnt surprising that the FreeR climbs steadily. Scaling Fcalcs and Fobs together at this resolution is tricky, and you should look at your plots of Rfactor and Fo/Fc at the end of refinement. There may be anomalies at the top or bottom resolution.. There really isnt a general rule for a fix. You need to know your data. You can still get useful information from the maps. Eleanor On Sat, 9 Jul 2011 16:59:29 +0800, Qixu Cai caiq...@gmail.com wrote: Dear all, Recently, I refine two low resolution structures in refmac 5.5. Their resolutions are 3A and 3.5A respectively. For 3A structure, after MR by phaser and rigidbody refinementrestraint refinement by refmac5.5, I got R factor 25% and R free 35%. And then each time, after my model building in coot and restraint refinement by refmac 5.5, the R factor stays 25%, but R free increases to 38%, even 39%. For 3.5A structure, the R factor stays 27%, but R free increases from 37% to 42% after my slightly model building in coot. Could you help me to find the reason? Maybe the reason is the overfit of the structure? I found that new version of refmac 5.6 has many new features for low resolution refinement, such as jelly boy, secondary structure restraints. But I don't know how to use these new features in old version ccp4i (6.1.13)? I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. Could you help me? Thank you very much!
Re: [ccp4bb] low resolution refinement
Dear All, An updated ccp4i interface to Refmac 5.6 is available at: http://www.ccp4.ac.uk/prerelease/refmac5_task_for_ccp4_6.2.tar.gz You can do System Administration - Install/uninstall Tasks from the GUI, or just copy the individual files from the tarball to the $CCP4/ccp4i area. This interface includes jelly-body, local NCS and map sharpening. It will be in the imminent CCP4 6.2.0 HTH Martyn On Sat, 2011-07-09 at 12:44 +0200, Robbie Joosten wrote: Dear Qixu, refamac 5.6 works well at these resolutions. You can add commands to your refinement in CCP4i by using the 'Run and view command script' (or something like that) option and just typing in the extra commands. Jelly-body has worked very well for me (although I use tigheter restraints than the default). Also local NCS works well (provided you have NCS). I never used reference structures, but I heared good things about it. Don't forget to use riding hydrogens, for some reason it is not the deafault. Perhaps you should also switch of the automatic X-ray weighting in favour of optimizing the matrix weight yourself (start with 0.05 and compare refinements for higher and lower values). HTH, Robbie Date: Sat, 9 Jul 2011 16:59:29 +0800 From: caiq...@gmail.com Subject: [ccp4bb] low resolution refinement To: CCP4BB@JISCMAIL.AC.UK Dear all, Recently, I refine two low resolution structures in refmac 5.5. Their resolutions are 3A and 3.5A respectively. For 3A structure, after MR by phaser and rigidbody refinementrestraint refinement by refmac5.5, I got R factor 25% and R free 35%. And then each time, after my model building in coot and restraint refinement by refmac 5.5, the R factor stays 25%, but R free increases to 38%, even 39%. For 3.5A structure, the R factor stays 27%, but R free increases from 37% to 42% after my slightly model building in coot. Could you help me to find the reason? Maybe the reason is the overfit of the structure? I found that new version of refmac 5.6 has many new features for low resolution refinement, such as jelly boy, secondary structure restraints. But I don't know how to use these new features in old version ccp4i (6.1.13)? I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. Could you help me? Thank you very much! -- *** * * * Dr. Martyn Winn * * * * STFC Daresbury Laboratory, Daresbury, Warrington, WA4 4AD, U.K. * * Tel: +44 1925 603455E-mail: martyn.w...@stfc.ac.uk* * Fax: +44 1925 603634Skype name: martyn.winn * * URL: http://www.ccp4.ac.uk/martyn/ * ***
Re: [ccp4bb] low resolution refinement
Hi Quixu, I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. using the latest phenix.refine from nightly builds ( http://www.phenix-online.org/download/nightly_builds.cgi), try combined strategy: - refinement of individual coordinates and ADPs, and - use SA (try different temperatures, say 5000K and 1K), and - use reference model, and - use weight optimization which is not much improved in the latest nightly builds and can use multiple CPUs, and - use Ramachandran plot restraints, and - if NCS is available - use it. Make sure you use it in torsion angle space (this is a new option that is much more robust and does not require ad hoc decisions about NCS group selections) - use secondary structure restraints ONLY IF a) they are determined automatically from non-distorted starting model that has good geometry, or b) manually with great deal of thought, and - try using H atoms although they may not help much f you are far far away from a final model. When exploring the maps use automatically sharpened maps which typically enhances the weak features on low resolution map. Combining sharpening with map kicking should further improve the maps. Some experimenting with the above options might be helpful. If this doesn't help then please contact me *off-list or at phenixbb* and I might be able to help. Pavel.
Re: [ccp4bb] low resolution refinement
Hi, Thank you very much. In the example5 of this page http://www.ysbl.york.ac.uk/~garib/refmac/docs/usage/examples.html#exam5, It seems that for 3A dataset, MAKE HYDRogenshttp://www.ysbl.york.ac.uk/%7Egarib/refmac/docs/keywords/restraints.html#make_hydrNo. Is it mean that the hydrogen just usefull for high resolution data? 2011/7/10 Robbie Joosten robbie_joos...@hotmail.com Hi Qixu, In CCP4i the option is in the refinement parameters: Use hydrogen atoms: [build all hydrogens] and [] output to coordinate file What is does is build all hydrogens at the expected coordinates and constrain them in refinement (i.e. adding hydrogens does not add extra parameters to the model). The effect on explaining your experminetal data is typically small, but the hydrogens help with the VdW restraints. In effect they reduce the number of bumps and improve your torsion angles. You can use a reference structure to generate external restraints: http://www.ysbl.york.ac.uk/~garib/refmac/data/refmac_news.html#External I hope someone else on the BB can explain how. I think it is also explained in the talk and tutorials of the Refmac website. HTH, Robbie From: caiq...@gmail.com Date: Sun, 10 Jul 2011 00:44:25 +0800 Subject: Re: [ccp4bb] low resolution refinement To: robbie_joos...@hotmail.com CC: CCP4BB@jiscmail.ac.uk Hi, Thank you for your suggestion. Could you tell me what is riding hydrogens? And it seems there is not reference model function in refmac5.6? 2011/7/9 Robbie Joosten robbie_joos...@hotmail.commailto:robbie_joos...@hotmail.com Dear Qixu, refamac 5.6 works well at these resolutions. You can add commands to your refinement in CCP4i by using the 'Run and view command script' (or something like that) option and just typing in the extra commands. Jelly-body has worked very well for me (although I use tigheter restraints than the default). Also local NCS works well (provided you have NCS). I never used reference structures, but I heared good things about it. Don't forget to use riding hydrogens, for some reason it is not the deafault. Perhaps you should also switch of the automatic X-ray weighting in favour of optimizing the matrix weight yourself (start with 0.05 and compare refinements for higher and lower values). HTH, Robbie Date: Sat, 9 Jul 2011 16:59:29 +0800 From: caiq...@gmail.commailto:caiq...@gmail.com Subject: [ccp4bb] low resolution refinement To: CCP4BB@JISCMAIL.AC.UKmailto:CCP4BB@JISCMAIL.AC.UK Dear all, Recently, I refine two low resolution structures in refmac 5.5. Their resolutions are 3A and 3.5A respectively. For 3A structure, after MR by phaser and rigidbody refinementrestraint refinement by refmac5.5, I got R factor 25% and R free 35%. And then each time, after my model building in coot and restraint refinement by refmac 5.5, the R factor stays 25%, but R free increases to 38%, even 39%. For 3.5A structure, the R factor stays 27%, but R free increases from 37% to 42% after my slightly model building in coot. Could you help me to find the reason? Maybe the reason is the overfit of the structure? I found that new version of refmac 5.6 has many new features for low resolution refinement, such as jelly boy, secondary structure restraints. But I don't know how to use these new features in old version ccp4i (6.1.13)? I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. Could you help me? Thank you very much!
Re: [ccp4bb] low resolution refinement
When in doubt, try both. In my personal experience, adding hydrogens always works. Especially at low resolution. But don't take my word for it, experiment a little. Cheers, Robbie Date: Sun, 10 Jul 2011 16:01:59 +0800 From: caiq...@gmail.com Subject: Re: [ccp4bb] low resolution refinement To: CCP4BB@JISCMAIL.AC.UK Hi, Thank you very much. In the example5 of this page http://www.ysbl.york.ac.uk/~garib/refmac/docs/usage/examples.html#exam5, It seems that for 3A dataset, MAKE HYDRogens No. Is it mean that the hydrogen just usefull for high resolution data? 2011/7/10 Robbie Joosten robbie_joos...@hotmail.com Hi Qixu, In CCP4i the option is in the refinement parameters: Use hydrogen atoms: [build all hydrogens] and [] output to coordinate file What is does is build all hydrogens at the expected coordinates and constrain them in refinement (i.e. adding hydrogens does not add extra parameters to the model). The effect on explaining your experminetal data is typically small, but the hydrogens help with the VdW restraints. In effect they reduce the number of bumps and improve your torsion angles. You can use a reference structure to generate external restraints: http://www.ysbl.york.ac.uk/~garib/refmac/data/refmac_news.html#External I hope someone else on the BB can explain how. I think it is also explained in the talk and tutorials of the Refmac website. HTH, Robbie From: caiq...@gmail.com Date: Sun, 10 Jul 2011 00:44:25 +0800 Subject: Re: [ccp4bb] low resolution refinement To: robbie_joos...@hotmail.com CC: CCP4BB@jiscmail.ac.uk Hi, Thank you for your suggestion. Could you tell me what is riding hydrogens? And it seems there is not reference model function in refmac5.6? 2011/7/9 Robbie Joosten robbie_joos...@hotmail.commailto:robbie_joos...@hotmail.com Dear Qixu, refamac 5.6 works well at these resolutions. You can add commands to your refinement in CCP4i by using the 'Run and view command script' (or something like that) option and just typing in the extra commands. Jelly-body has worked very well for me (although I use tigheter restraints than the default). Also local NCS works well (provided you have NCS). I never used reference structures, but I heared good things about it. Don't forget to use riding hydrogens, for some reason it is not the deafault. Perhaps you should also switch of the automatic X-ray weighting in favour of optimizing the matrix weight yourself (start with 0.05 and compare refinements for higher and lower values). HTH, Robbie Date: Sat, 9 Jul 2011 16:59:29 +0800 From: caiq...@gmail.commailto:caiq...@gmail.com Subject: [ccp4bb] low resolution refinement To: CCP4BB@JISCMAIL.AC.UKmailto:CCP4BB@JISCMAIL.AC.UK Dear all, Recently, I refine two low resolution structures in refmac 5.5. Their resolutions are 3A and 3.5A respectively. For 3A structure, after MR by phaser and rigidbody refinementrestraint refinement by refmac5.5, I got R factor 25% and R free 35%. And then each time, after my model building in coot and restraint refinement by refmac 5.5, the R factor stays 25%, but R free increases to 38%, even 39%. For 3.5A structure, the R factor stays 27%, but R free increases from 37% to 42% after my slightly model building in coot. Could you help me to find the reason? Maybe the reason is the overfit of the structure? I found that new version of refmac 5.6 has many new features for low resolution refinement, such as jelly boy, secondary structure restraints. But I don't know how to use these new features in old version ccp4i (6.1.13)? I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. Could you help me? Thank you very much!
[ccp4bb] low resolution refinement
Dear all, Recently, I refine two low resolution structures in refmac 5.5. Their resolutions are 3A and 3.5A respectively. For 3A structure, after MR by phaser and rigidbody refinementrestraint refinement by refmac5.5, I got R factor 25% and R free 35%. And then each time, after my model building in coot and restraint refinement by refmac 5.5, the R factor stays 25%, but R free increases to 38%, even 39%. For 3.5A structure, the R factor stays 27%, but R free increases from 37% to 42% after my slightly model building in coot. Could you help me to find the reason? Maybe the reason is the overfit of the structure? I found that new version of refmac 5.6 has many new features for low resolution refinement, such as jelly boy, secondary structure restraints. But I don't know how to use these new features in old version ccp4i (6.1.13)? I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. Could you help me? Thank you very much!
Re: [ccp4bb] low resolution refinement
Hi I highly recommend reading the two Brunger papers below: they will explain the important factors to take note of when refining low-resolution structures.� I found them very useful when I was learning about low-resolution refinement. I would suggest using the deformable elastic network with simulated annealing in CNS, and then follow that with refinement in PHENIX using tight geometery restraints and doing TLS and grouped-B factor refinement.� Your problem is that at low resolutions, the data to parameters ratio is low, and you need to use a number of restraints along with good bulk-solvent and anisotropic corections in order to keep your R-factor and R-free values within 3-4%.� Perhaps you could send your PHENIX input file and we could see what you did and maybe point what needs to be corrected. Hope that was helpful.� good luck! Schroder, Levitt, Brunger, Super-resolution biomolecular crystallography with low-resolution data Nature. 2010 Apr 22;464(7292):1218-22Brunger, AT et al.� X-ray structure determination at low resolution. Acta Cryst D. 2009 Feb;65(Pt 2):128-33 On Sat 07/09/11 4:59 AM , Qixu Cai wrote: Dear all, Recently, I refine two low resolution structures in refmac 5.5. Their resolutions are 3A and 3.5A respectively. For 3A structure, after MR by phaser and rigidbody refinementrestraint refinement by refmac5.5, I got R factor 25% and R free 35%. And then each time, after my model building in coot and restraint refinement by refmac 5.5, the R factor stays 25%, but R free increases to 38%, even 39%. For 3.5A structure, the R factor stays 27%, but R free increases from 37% to 42% after my slightly model building in coot. Could you help me to find the reason? Maybe the reason is the overfit of the structure? I found that new version of refmac 5.6 has many new features for low resolution refinement, such as jelly boy, secondary structure restraints. But I don't know how to use these new features in old version ccp4i (6.1.13)? I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. Could you help me? Thank you very much!
Re: [ccp4bb] low resolution refinement
Dear Qixu, refamac 5.6 works well at these resolutions. You can add commands to your refinement in CCP4i by using the 'Run and view command script' (or something like that) option and just typing in the extra commands. Jelly-body has worked very well for me (although I use tigheter restraints than the default). Also local NCS works well (provided you have NCS). I never used reference structures, but I heared good things about it. Don't forget to use riding hydrogens, for some reason it is not the deafault. Perhaps you should also switch of the automatic X-ray weighting in favour of optimizing the matrix weight yourself (start with 0.05 and compare refinements for higher and lower values). HTH, Robbie Date: Sat, 9 Jul 2011 16:59:29 +0800 From: caiq...@gmail.com Subject: [ccp4bb] low resolution refinement To: CCP4BB@JISCMAIL.AC.UK Dear all, Recently, I refine two low resolution structures in refmac 5.5. Their resolutions are 3A and 3.5A respectively. For 3A structure, after MR by phaser and rigidbody refinementrestraint refinement by refmac5.5, I got R factor 25% and R free 35%. And then each time, after my model building in coot and restraint refinement by refmac 5.5, the R factor stays 25%, but R free increases to 38%, even 39%. For 3.5A structure, the R factor stays 27%, but R free increases from 37% to 42% after my slightly model building in coot. Could you help me to find the reason? Maybe the reason is the overfit of the structure? I found that new version of refmac 5.6 has many new features for low resolution refinement, such as jelly boy, secondary structure restraints. But I don't know how to use these new features in old version ccp4i (6.1.13)? I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. Could you help me? Thank you very much!
Re: [ccp4bb] low resolution refinement
Hi, Thank you for your papers. I will read them carefully. 2011/7/9 Elizabeth McSweeney elizabeth.mcswee...@duke.edu Hi I highly recommend reading the two Brunger papers below: they will explain the important factors to take note of when refining low-resolution structures. I found them very useful when I was learning about low-resolution refinement. I would suggest using the deformable elastic network with simulated annealing in CNS, and then follow that with refinement in PHENIX using tight geometery restraints and doing TLS and grouped-B factor refinement. Your problem is that at low resolutions, the data to parameters ratio is low, and you need to use a number of restraints along with good bulk-solvent and anisotropic corections in order to keep your R-factor and R-free values within 3-4%. Perhaps you could send your PHENIX input file and we could see what you did and maybe point what needs to be corrected. Hope that was helpful. good luck! Schroder, Levitt, Brunger, Super-resolution biomolecular crystallography with low-resolution data Nature. 2010 Apr 22;464(7292):1218-22 Brunger, AT et al. X-ray structure determination at low resolution. Acta Cryst D. 2009 Feb;65(Pt 2):128-33 On Sat 07/09/11 4:59 AM , Qixu Cai caiq...@gmail.com wrote: Dear all, Recently, I refine two low resolution structures in refmac 5.5. Their resolutions are 3A and 3.5A respectively. For 3A structure, after MR by phaser and rigidbody refinementrestraint refinement by refmac5.5, I got R factor 25% and R free 35%. And then each time, after my model building in coot and restraint refinement by refmac 5.5, the R factor stays 25%, but R free increases to 38%, even 39%. For 3.5A structure, the R factor stays 27%, but R free increases from 37% to 42% after my slightly model building in coot. Could you help me to find the reason? Maybe the reason is the overfit of the structure? I found that new version of refmac 5.6 has many new features for low resolution refinement, such as jelly boy, secondary structure restraints. But I don't know how to use these new features in old version ccp4i (6.1.13)? I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. Could you help me? Thank you very much!
Re: [ccp4bb] low resolution refinement
Hi, Thank you for your suggestion. Could you tell me what is riding hydrogens? And it seems there is not reference model function in refmac5.6? 2011/7/9 Robbie Joosten robbie_joos...@hotmail.com Dear Qixu, refamac 5.6 works well at these resolutions. You can add commands to your refinement in CCP4i by using the 'Run and view command script' (or something like that) option and just typing in the extra commands. Jelly-body has worked very well for me (although I use tigheter restraints than the default). Also local NCS works well (provided you have NCS). I never used reference structures, but I heared good things about it. Don't forget to use riding hydrogens, for some reason it is not the deafault. Perhaps you should also switch of the automatic X-ray weighting in favour of optimizing the matrix weight yourself (start with 0.05 and compare refinements for higher and lower values). HTH, Robbie Date: Sat, 9 Jul 2011 16:59:29 +0800 From: caiq...@gmail.com Subject: [ccp4bb] low resolution refinement To: CCP4BB@JISCMAIL.AC.UK Dear all, Recently, I refine two low resolution structures in refmac 5.5. Their resolutions are 3A and 3.5A respectively. For 3A structure, after MR by phaser and rigidbody refinementrestraint refinement by refmac5.5, I got R factor 25% and R free 35%. And then each time, after my model building in coot and restraint refinement by refmac 5.5, the R factor stays 25%, but R free increases to 38%, even 39%. For 3.5A structure, the R factor stays 27%, but R free increases from 37% to 42% after my slightly model building in coot. Could you help me to find the reason? Maybe the reason is the overfit of the structure? I found that new version of refmac 5.6 has many new features for low resolution refinement, such as jelly boy, secondary structure restraints. But I don't know how to use these new features in old version ccp4i (6.1.13)? I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. Could you help me? Thank you very much!
Re: [ccp4bb] low resolution refinement
Hi Qixu, In CCP4i the option is in the refinement parameters: Use hydrogen atoms: [build all hydrogens] and [] output to coordinate file What is does is build all hydrogens at the expected coordinates and constrain them in refinement (i.e. adding hydrogens does not add extra parameters to the model). The effect on explaining your experminetal data is typically small, but the hydrogens help with the VdW restraints. In effect they reduce the number of bumps and improve your torsion angles. You can use a reference structure to generate external restraints: http://www.ysbl.york.ac.uk/~garib/refmac/data/refmac_news.html#External I hope someone else on the BB can explain how. I think it is also explained in the talk and tutorials of the Refmac website. HTH, Robbie From: caiq...@gmail.com Date: Sun, 10 Jul 2011 00:44:25 +0800 Subject: Re: [ccp4bb] low resolution refinement To: robbie_joos...@hotmail.com CC: CCP4BB@jiscmail.ac.uk Hi, Thank you for your suggestion. Could you tell me what is riding hydrogens? And it seems there is not reference model function in refmac5.6? 2011/7/9 Robbie Joosten robbie_joos...@hotmail.commailto:robbie_joos...@hotmail.com Dear Qixu, refamac 5.6 works well at these resolutions. You can add commands to your refinement in CCP4i by using the 'Run and view command script' (or something like that) option and just typing in the extra commands. Jelly-body has worked very well for me (although I use tigheter restraints than the default). Also local NCS works well (provided you have NCS). I never used reference structures, but I heared good things about it. Don't forget to use riding hydrogens, for some reason it is not the deafault. Perhaps you should also switch of the automatic X-ray weighting in favour of optimizing the matrix weight yourself (start with 0.05 and compare refinements for higher and lower values). HTH, Robbie Date: Sat, 9 Jul 2011 16:59:29 +0800 From: caiq...@gmail.commailto:caiq...@gmail.com Subject: [ccp4bb] low resolution refinement To: CCP4BB@JISCMAIL.AC.UKmailto:CCP4BB@JISCMAIL.AC.UK Dear all, Recently, I refine two low resolution structures in refmac 5.5. Their resolutions are 3A and 3.5A respectively. For 3A structure, after MR by phaser and rigidbody refinementrestraint refinement by refmac5.5, I got R factor 25% and R free 35%. And then each time, after my model building in coot and restraint refinement by refmac 5.5, the R factor stays 25%, but R free increases to 38%, even 39%. For 3.5A structure, the R factor stays 27%, but R free increases from 37% to 42% after my slightly model building in coot. Could you help me to find the reason? Maybe the reason is the overfit of the structure? I found that new version of refmac 5.6 has many new features for low resolution refinement, such as jelly boy, secondary structure restraints. But I don't know how to use these new features in old version ccp4i (6.1.13)? I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. Could you help me? Thank you very much!
Re: [ccp4bb] low resolution refinement
Hi, External structure atomic bond distance restraints for use in refmac are generated using ProSMART. The webpage is here: http://www.ysbl.york.ac.uk/mxstat/Rob/index.html Email me if you want a pre-release version. Regards, Rob On 9 Jul 2011, at 18:00, Robbie Joosten wrote: Hi Qixu, In CCP4i the option is in the refinement parameters: Use hydrogen atoms: [build all hydrogens] and [] output to coordinate file What is does is build all hydrogens at the expected coordinates and constrain them in refinement (i.e. adding hydrogens does not add extra parameters to the model). The effect on explaining your experminetal data is typically small, but the hydrogens help with the VdW restraints. In effect they reduce the number of bumps and improve your torsion angles. You can use a reference structure to generate external restraints: http://www.ysbl.york.ac.uk/~garib/refmac/data/refmac_news.html#External I hope someone else on the BB can explain how. I think it is also explained in the talk and tutorials of the Refmac website. HTH, Robbie From: caiq...@gmail.com Date: Sun, 10 Jul 2011 00:44:25 +0800 Subject: Re: [ccp4bb] low resolution refinement To: robbie_joos...@hotmail.com CC: CCP4BB@jiscmail.ac.uk Hi, Thank you for your suggestion. Could you tell me what is riding hydrogens? And it seems there is not reference model function in refmac5.6? 2011/7/9 Robbie Joosten robbie_joos...@hotmail.commailto:robbie_joos...@hotmail.com Dear Qixu, refamac 5.6 works well at these resolutions. You can add commands to your refinement in CCP4i by using the 'Run and view command script' (or something like that) option and just typing in the extra commands. Jelly-body has worked very well for me (although I use tigheter restraints than the default). Also local NCS works well (provided you have NCS). I never used reference structures, but I heared good things about it. Don't forget to use riding hydrogens, for some reason it is not the deafault. Perhaps you should also switch of the automatic X-ray weighting in favour of optimizing the matrix weight yourself (start with 0.05 and compare refinements for higher and lower values). HTH, Robbie Date: Sat, 9 Jul 2011 16:59:29 +0800 From: caiq...@gmail.commailto:caiq...@gmail.com Subject: [ccp4bb] low resolution refinement To: CCP4BB@JISCMAIL.AC.UKmailto:CCP4BB@JISCMAIL.AC.UK Dear all, Recently, I refine two low resolution structures in refmac 5.5. Their resolutions are 3A and 3.5A respectively. For 3A structure, after MR by phaser and rigidbody refinementrestraint refinement by refmac5.5, I got R factor 25% and R free 35%. And then each time, after my model building in coot and restraint refinement by refmac 5.5, the R factor stays 25%, but R free increases to 38%, even 39%. For 3.5A structure, the R factor stays 27%, but R free increases from 37% to 42% after my slightly model building in coot. Could you help me to find the reason? Maybe the reason is the overfit of the structure? I found that new version of refmac 5.6 has many new features for low resolution refinement, such as jelly boy, secondary structure restraints. But I don't know how to use these new features in old version ccp4i (6.1.13)? I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. Could you help me? Thank you very much!
Re: [ccp4bb] low resolution refinement
Dear CAI Qixu Refmac does an excellent job at low resolution. Many of the features available in version 5.6.x are described in a very recent Refmac paper Murshudov et al. REFMAC5 for the refinement of macromolecular crystal structures. Acta Crystallogr D Biol Crystallogr (2011) vol. 67 (Pt 4) pp. 355-67 Guidelines for Refmac use (including new functions) can be found at http://www.ysbl.york.ac.uk/~garib/refmac/data/refmac_news.html For example: Jelly body restraints are controlled by ridge distance sigma value # Default 0.1 ridge distance dmax value # Default 4.2 ridge atoms sigma ridge bvalue sigma keywords. You will find info on this under Harmonic distance restraints (Ridge regression) External restraints (distance) are controlled by a syntax like that below: exte dist first chain A residue 78 atom O second chain A residue 82 atom N value 3.000 sigma 0.05 More information at under External restraints. As Rob pointed out ProSMART can generate the above type of restraints for you. Phenix can also provide a list of external distance restraints which can be interpreted by Refmac phenix.secondary_structure_restraints model.pdb format=refmac Best wishes Roberto On 9 Jul 2011, at 17:44, CAI Qixu wrote: Hi, Thank you for your suggestion. Could you tell me what is riding hydrogens? And it seems there is not reference model function in refmac5.6? 2011/7/9 Robbie Joosten robbie_joos...@hotmail.commailto:robbie_joos...@hotmail.com Dear Qixu, refamac 5.6 works well at these resolutions. You can add commands to your refinement in CCP4i by using the 'Run and view command script' (or something like that) option and just typing in the extra commands. Jelly-body has worked very well for me (although I use tigheter restraints than the default). Also local NCS works well (provided you have NCS). I never used reference structures, but I heared good things about it. Don't forget to use riding hydrogens, for some reason it is not the deafault. Perhaps you should also switch of the automatic X-ray weighting in favour of optimizing the matrix weight yourself (start with 0.05 and compare refinements for higher and lower values). HTH, Robbie Date: Sat, 9 Jul 2011 16:59:29 +0800 From: caiq...@gmail.commailto:caiq...@gmail.com Subject: [ccp4bb] low resolution refinement To: CCP4BB@JISCMAIL.AC.UKmailto:CCP4BB@JISCMAIL.AC.UK Dear all, Recently, I refine two low resolution structures in refmac 5.5. Their resolutions are 3A and 3.5A respectively. For 3A structure, after MR by phaser and rigidbody refinementrestraint refinement by refmac5.5, I got R factor 25% and R free 35%. And then each time, after my model building in coot and restraint refinement by refmac 5.5, the R factor stays 25%, but R free increases to 38%, even 39%. For 3.5A structure, the R factor stays 27%, but R free increases from 37% to 42% after my slightly model building in coot. Could you help me to find the reason? Maybe the reason is the overfit of the structure? I found that new version of refmac 5.6 has many new features for low resolution refinement, such as jelly boy, secondary structure restraints. But I don't know how to use these new features in old version ccp4i (6.1.13)? I also used phenix.refine with the reference model ( I have high resolution model for one domain of the low resolution protein) and secondary structure restraints, but it seams the same. Any suggestion? BTW, is that simulator annealing not suitable for low resolution structure? I used the simulator annealing method of CNS and phenix.refine, but the geometry of the structure is always destroyed seriously. Could you help me? Thank you very much! Roberto Steiner, PhD Randall Division of Cell and Molecular Biophysics Group Leader King's College London Room 3.10A New Hunt's House Guy's Campus SE1 1UL, London, UK Tel 0044-20-78488216 Fax 0044-20-78486435 roberto.stei...@kcl.ac.ukmailto:roberto.stei...@kcl.ac.uk
Re: [ccp4bb] Low resolution refinement
Dear Joane, we had a case, where we had five molecules in the assymmetric unit where the biological functional unit was a homotrimer. So we had one non-crystallographic trimer and two monomers, which were located along the 3-fold symmetry axis of space group I213. One of the monomers also showed electron density of considerably lower quality, obviously going along with higher B-factors. By a careful analysis of the crystal packing we could see that this chain has only very few crystal contacts. If you want to have a closer look, see: Schuldt L, Weyand S, Kefala G, Weiss MS J. Mol. Biol. (2009), 863-879. The Section Crystal Packing and structural variation describes this in more detail. Best wishes, Linda Joane Kathelen Rustiguel schrieb: Dear all I am refining a structure at 3.4 A resolution that contains 3 molecules in the a.u. The chain A sits on a 2-fold crystallographic axis forming the dimeric functional structure expected for this class of proteins. The other two chains B and C, which also form the functional dimer, seem to be, somehow, a lot more flexible than chain A. As a result, whereas the electron density map, b-factor and geometry for chain A is pretty reasonable for a 3.4 A resolution structure, the refinement for the other two chains (B and C) does not behave well. Even playing with different weights for geometry, analysing different levels of 2Fo-Fc/Fo-Fc maps, using NCS, TLS, etc..., nothing works. The map for the helical regions is ok, but the electron density map for strands and loops of chains B and C are broken along the main chain, B-factors are really high, and the geometry keeps being distorted. Right now, the R-factor and R-free are 24.2 and 28.6, respectively. Any suggestions in how to proceed the refinement? And even a more difficult question, how do we report this type of structure? How do we deposit those coordinates? We can certainly use chain A as a model to perform interesting studies of structure-function relationship, but we know that chain B and chain C have problems. Any help will be greatly appreciated. Regards Joane -- Joane Kathelen Rustiguel Bonalumi Faculdade de Ciências Farmacêuticas de Ribeirão Preto - USP Laboratório de Cristalografia de Proteínas Departamento de Física e Química Fone: +55.16.3602.4193 *** Dr. Linda Schuldt Department of Molecular Biology University of Aarhus Science Park Gustav Wieds Vej 10c DK-8000 Århus C Denmark
[ccp4bb] Low resolution refinement
Dear all I am refining a structure at 3.4 A resolution that contains 3 molecules in the a.u. The chain A sits on a 2-fold crystallographic axis forming the dimeric functional structure expected for this class of proteins. The other two chains B and C, which also form the functional dimer, seem to be, somehow, a lot more flexible than chain A. As a result, whereas the electron density map, b-factor and geometry for chain A is pretty reasonable for a 3.4 A resolution structure, the refinement for the other two chains (B and C) does not behave well. Even playing with different weights for geometry, analysing different levels of 2Fo-Fc/Fo-Fc maps, using NCS, TLS, etc..., nothing works. The map for the helical regions is ok, but the electron density map for strands and loops of chains B and C are broken along the main chain, B-factors are really high, and the geometry keeps being distorted. Right now, the R-factor and R-free are 24.2 and 28.6, respectively. Any suggestions in how to proceed the refinement? And even a more difficult question, how do we report this type of structure? How do we deposit those coordinates? We can certainly use chain A as a model to perform interesting studies of structure-function relationship, but we know that chain B and chain C have problems. Any help will be greatly appreciated. Regards Joane -- Joane Kathelen Rustiguel Bonalumi Faculdade de Ciências Farmacêuticas de Ribeirão Preto - USP Laboratório de Cristalografia de Proteínas Departamento de Física e Química Fone: +55.16.3602.4193
Re: [ccp4bb] Low resolution refinement
Missing density could point to incomplete data, how is the completeness at high and low resolution? Sent from my iPhone On May 19, 2011, at 11:54 AM, Jon Schuermann schue...@anl.gov wrote: Dear Joane, Bert's recommendations are very good, but I'd like to add a little caution. If just part of the molecule has bad density then it is not unusual, but if the whole chain does not look very good (missing side-chains or backbone) you may have a couple of things going on. First, I would first suspect NCS issues (as Bert has) where you might be trying to force too high a space group since an NCS operator might be darn close to a crystallographic operator. Second it could also be caused by twinning, but since your R-factors are pretty respectable I would imagine that is probably not the case although you may want to look into it. Third, it could be static disorder where the crappy density might be two molecules on top of each other but one is slightly shifted. A way to tell is to scale down in your electron density maps to see if there are extra strands and helices next to the strands and helices you have built. Hopefully this is not the case and it is just NCS messing with you... Jon -- Jonathan P. Schuermann, Ph. D. Beamline Scientist NE-CAT, Building 436E Advanced Photon Source (APS) Argonne National Laboratory 9700 South Cass Avenue Argonne, IL 60439 email: schue...@anl.gov Tel: (630) 252-0682 Fax: (630) 252-0687 On 05/19/2011 10:06 AM, Van Den Berg, Bert wrote: One more thing though: have you refined with the NCS restraints off or on? Presumably on, seeing that you have a small gap between R and Rfree? It may be worth turning the restraints off or modify the NCS selections: if your 3 molecules are substantially different, applying NCS may make things worse rather than better (ie you’re averaging things that are different). Good luck, Bert On 5/19/11 9:02 AM, Joane Kathelen Rustiguel jkrustig...@usp.br wrote: Dear all I am refining a structure at 3.4 A resolution that contains 3 molecules in the a.u. The chain A sits on a 2-fold crystallographic axis forming the dimeric functional structure expected for this class of proteins. The other two chains B and C, which also form the functional dimer, seem to be, somehow, a lot more flexible than chain A. As a result, whereas the electron density map, b-factor and geometry for chain A is pretty reasonable for a 3.4 A resolution structure, the refinement for the other two chains (B and C) does not behave well. Even playing with different weights for geometry, analysing different levels of 2Fo-Fc/Fo-Fc maps, using NCS, TLS, etc..., nothing works. The map for the helical regions is ok, but the electron density map for strands and loops of chains B and C are broken along the main chain, B-factors are really high, and the geometry keeps being distorted. Right now, the R-factor and R-free are 24.2 and 28.6, respectively. Any suggestions in how to proceed the refinement? And even a more difficult question, how do we report this type of structure? How do we deposit those coordinates? We can certainly use chain A as a model to perform interesting studies of structure-function relationship, but we know that chain B and chain C have problems. Any help will be greatly appreciated. Regards Joane -- Joane Kathelen Rustiguel Bonalumi Faculdade de Ciências Farmacêuticas de Ribeirão Preto - USP Laboratório de Cristalografia de Proteínas Departamento de Física e Química Fone: +55.16.3602.4193
Re: [ccp4bb] Low resolution refinement
This reminds me of: Pig heart short chain L-3-hydroxyacyl-CoA dehydrogenase revisited: Sequence analysis and crystal structure determination Barycki JJ, O'Brien LK, Birktoft JJ, Strauss AW, Banaszak LJ Protein Science (Oct 1999) Vol 8, pp 2010-2018. In which the protein in question also had one monomer forming a dimer about a crystallographic axis, and two monomers forming a dimer elsewhere in the asymmetric unit. A portion of the molecule had messy density, which caused difficulties literally for decades. The eventual solution was to switch to a different species (human). After which, a MR solution of the original porcine data was possible. I believe the disorder was judged to be intrinsic. On 05/19/11 09:02, Joane Kathelen Rustiguel wrote: Dear all I am refining a structure at 3.4 A resolution that contains 3 molecules in the a.u. The chain A sits on a 2-fold crystallographic axis forming the dimeric functional structure expected for this class of proteins. The other two chains B and C, which also form the functional dimer, seem to be, somehow, a lot more flexible than chain A. As a result, whereas the electron density map, b-factor and geometry for chain A is pretty reasonable for a 3.4 A resolution structure, the refinement for the other two chains (B and C) does not behave well. Even playing with different weights for geometry, analysing different levels of 2Fo-Fc/Fo-Fc maps, using NCS, TLS, etc..., nothing works. The map for the helical regions is ok, but the electron density map for strands and loops of chains B and C are broken along the main chain, B-factors are really high, and the geometry keeps being distorted. Right now, the R-factor and R-free are 24.2 and 28.6, respectively. Any suggestions in how to proceed the refinement? And even a more difficult question, how do we report this type of structure? How do we deposit those coordinates? We can certainly use chain A as a model to perform interesting studies of structure-function relationship, but we know that chain B and chain C have problems. Any help will be greatly appreciated. Regards Joane -- === All Things Serve the Beam === David J. Schuller modern man in a post-modern world MacCHESS, Cornell University schul...@cornell.edu
Re: [ccp4bb] Low resolution refinement
Dear Joane, We have had very good luck refining low resolution structures with dramatic improvement using several newish options in refmac - particularly when NCS is present! Those options are jelly body restraints, automatic NCS restraints, and map sharpening. Here is a description cut-and-pasted from Garib's presentation (www.ccp4.ac.uk/schools/China-2011/tutorials/refmac_tutorial.pdf) V) Low resolution refinement If you have older version of ccp4 then you can follow the instructions described in the presentation: www.ysbl.york.ac.uk/refmac/Presentations/Refmac_Erice_workshop.ppt slides 45-50 Full description of new features are in: www.ysbl.york.ac.uk/refmac/data/refmac_news.html In the new version of ccp4 there are options for jelly body, automatic NCS and map sharpening. a) Jelly body is under “Refinement Parameters”. You need to click “Use jelly- body refinement with sigma”. Change sigma to 0.01 or 0.02. This value defines “jelliness”. Smaller value means tighter restraints b) Automatic NCS restraints are under “Setup Non-Crystallographic Symmetry”. You need to cick “use automatically generated local NCS restraints”. You can also use global NCS c) Map sharpening is under “Monitoring and Output Options”. You need to click “Perform map sharpening with B value 20.0”. B value should a little bit smaller than Wilson’s B value. more in depth information can be found here: www.ccp4.ac.uk/schools/China-2011/talks/refmac_Shanghai.pdf This is definitely something you should explore. good luck, Eric Eric T. Larson, PhD Biomolecular Structure Center Department of Biochemistry Box 357742 University of Washington Seattle, WA 98195 email: larso...@u.washington.edu On Thu, 19 May 2011, David Schuller wrote: This reminds me of: Pig heart short chain L-3-hydroxyacyl-CoA dehydrogenase revisited: Sequence analysis and crystal structure determination Barycki JJ, O'Brien LK, Birktoft JJ, Strauss AW, Banaszak LJ Protein Science (Oct 1999) Vol 8, pp 2010-2018. In which the protein in question also had one monomer forming a dimer about a crystallographic axis, and two monomers forming a dimer elsewhere in the asymmetric unit. A portion of the molecule had messy density, which caused difficulties literally for decades. The eventual solution was to switch to a different species (human). After which, a MR solution of the original porcine data was possible. I believe the disorder was judged to be intrinsic. On 05/19/11 09:02, Joane Kathelen Rustiguel wrote: Dear all I am refining a structure at 3.4 A resolution that contains 3 molecules in the a.u. The chain A sits on a 2-fold crystallographic axis forming the dimeric functional structure expected for this class of proteins. The other two chains B and C, which also form the functional dimer, seem to be, somehow, a lot more flexible than chain A. As a result, whereas the electron density map, b-factor and geometry for chain A is pretty reasonable for a 3.4 A resolution structure, the refinement for the other two chains (B and C) does not behave well. Even playing with different weights for geometry, analysing different levels of 2Fo-Fc/Fo-Fc maps, using NCS, TLS, etc..., nothing works. The map for the helical regions is ok, but the electron density map for strands and loops of chains B and C are broken along the main chain, B-factors are really high, and the geometry keeps being distorted. Right now, the R-factor and R-free are 24.2 and 28.6, respectively. Any suggestions in how to proceed the refinement? And even a more difficult question, how do we report this type of structure? How do we deposit those coordinates? We can certainly use chain A as a model to perform interesting studies of structure-function relationship, but we know that chain B and chain C have problems. Any help will be greatly appreciated. Regards Joane -- === All Things Serve the Beam === David J. Schuller modern man in a post-modern world MacCHESS, Cornell University schul...@cornell.edu
