 |
 |
Yes, that’s true, but you also see relatively high TFZ values associated with poor packing in other cases, such as pseudosymmetry misidentified as crystallographic symmetry. Very incomplete models also tend to pile up on symmetry operators, leading to poor packing, which is something we’d like to recognise and compensate for. Best wishes, Randy ----- Randy J. Read Department of Haematology, University of Cambridge Cambridge Institute for Medical Research Tel: +44 1223 336500 Wellcome Trust/MRC Building Fax: +44 1223 336827 Hills Road E-mail: [log in to unmask] Cambridge CB2 0XY, U.K. www-structmed.cimr.cam.ac.uk > On 21 Apr 2016, at 20:14, Keller, Jacob <[log in to unmask]> wrote: > > I believe static/statistical disorder would produce good TFZs with poor packing. > > See the paper below for example. > > JPK > > Sapan A Shah, Axel T Brunger, The 1.8 å crystal structure of a statically disordered 17 base-pair RNA duplex: principles of RNA crystal packing and its effect on nucleic acid structure1, Journal of Molecular Biology, Volume 285, Issue 4, 29 January 1999, Pages 1577-1588, ISSN 0022-2836, http://dx.doi.org/10.1006/jmbi.1998.2385. > (http://www.sciencedirect.com/science/article/pii/S0022283698923853) > Abstract: The crystal structure of a 17 base RNA oligomer, r(CACCGGAUG GUUCGGUG), has been solved to a resolution of 1.8 Å through a combination of molecular replacement, multiple isomorphous replacement phasing, and analysis of observed intensity distributions. The oligomer, which forms a stem-loop in solution, crystallized as a pseudo-infinite duplex in spacegroup P321. The asymmetric unit of the crystal contains four superimposed orientations of the duplex that are out of register, such that backbones superimpose, but base identity differs. This static disorder was initially discovered by brominating a single residue per strand in the sequence, and observing four peaks per strand in difference maps phased with a native molecular replacement solution. The presence of four superimposed duplex “motifs” related by non-crystallographic hypersymmetry was detected by computing 〈I2〉/〈I〉2 and Wilson ratios for the observed intensities. The observed ratios matched those produced from calculated intensities of a 4-fold statically disordered model. Multi-conformer simulated annealing refinement against a maximum-likelihood target incorporating experimental phase information was used to refine the 4-fold disordered model to an Rfree and R of 29.35% and 25.5%, respectively. The resulting structure reveals four distinct conformations of the duplex, with an average pairwise backbone rmsd of 2.35 Å. The structural differences between the four conformations, which can be attributed to differences in packing environment, highlight the possible influence of crystal packing forces on nucleic acid X-ray structures. Analysis of inter-helical packing between symmetry-related molecules reveals an RNA “zipper” that mediates direct phosphate oxygen-2′ hydroxyl interactions between close-packed phosphate-sugar backbones. This may be a general mode for RNA tertiary interaction that does not depend on metal ions or primary sequence. > Keywords: X-ray crystallography; RNA; disorder; intensity statistics; Wilson ratio > > > > > > -----Original Message----- > From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of Randy Read > Sent: Thursday, April 21, 2016 3:08 PM > To: [log in to unmask] > Subject: Re: [ccp4bb] Twin and tNCS > > Hi, > > On your point 2, if this isn’t already a frequently asked question somewhere it should be! With twinning, you have to decide first whether or not the crystal appears to be twinned. This is done on the basis of the twin tests that depend only on the distribution of intensities but not on a twin law, such as the second moments test or the L-test. If those tests say that your crystal is twinned, then you start to think about what the twin law is. > > The tests based on twin laws compare the reflections related by twin laws to see if they are more closely related to each other than you would expect them to be randomly. Unfortunately, if you have assigned a symmetry that is too low, then there are reflections that are unrelated by the symmetry you have assigned, but they are related by the symmetry that you have missed in your space group assignment, and those symmetry operators are also possible twin laws in the lower symmetry space group. So the various tests that depend on twin laws will say that you have perfect twinning, because the supposedly “twin-related” reflections are almost perfectly correlated to each other. If you reprocess your data in point group P622, you will find that there are no twin laws identified, and nothing in xtriage or truncate will say that your crystal is twinned. > > As for the RMS value, I agree with Ray that the value of 1.5A often works well for NMR structures. You could also try something more optimistic, like 1-1.25A. > > I’m not sure why you’re getting solutions with high TFZ that fail to pack. However, I’d like to see the results of searching in what I think is the correct point group first. > > Best wishes, > > Randy > > ----- > Randy J. Read > Department of Haematology, University of Cambridge > Cambridge Institute for Medical Research Tel: +44 1223 336500 > Wellcome Trust/MRC Building Fax: +44 1223 336827 > Hills Road E-mail: [log in to unmask] > Cambridge CB2 0XY, U.K. www-structmed.cimr.cam.ac.uk > >> On 21 Apr 2016, at 17:49, Qixu Cai <[log in to unmask]> wrote: >> >> Dear Dr. Read, >> >> 1, Yes, I have tried to turn off the tNCS option in phaser and still got the similar result that the high TFZ solutions were rejected because of packing clash. >> >> 2, According to the result of Xtriage, the twin fraction seems to be close to perfect twin. What's the relation between twin and tNCS? Are they independent? >> >> 3, The model I used for MR in phaser is NMR structure. The sequence is 100% identical but because it's NMR structure, I don't know how to judge RMS in phaser. >> >> Do you think that the rejection of high TFZ solution is related to the twinning of the dataset? >> >> Thanks a lot! >> Best wishes, >> Qizu CAI >> >> Qixu Cai >> Email: [log in to unmask] >> >> >> 2016-04-21 22:41 GMT+08:00 Randy Read <[log in to unmask]>: >> Hi, >> >> There are some odd things that I would want to sort out. >> >> First, xtriage thinks that your data merge in much higher symmetry, i.e. point group P622 rather than P3. Using the ccp4i cell content analysis option under Molecular Replacement -> Analysis, 2 copies of your DNA in P622 should give 50% solvent. That would fit with the apparent translational NCS. However, I would want to keep in mind the possibility that, if your DNA adopts some helical arrangement, that would also generate large Patterson peaks. So you could test the result of telling Phaser not to apply the tNCS correction. >> >> Second, it's not clear whether there is any twinning. After correcting for tNCS, Phaser says the second intensity moment is very nearly 2, which would imply no twinning. The L-test in xtriage might, however, imply at least partial twinning. If the crystal is twinned, then you would indeed have to consider that the true point group might be lower than the apparent P622. >> >> Third, I would worry about the presence of a lot of outliers. There appears to be a large number of reflections with significant net negative intensities, which implies that something might have gone wrong in the integration step. Maybe these come from the weak direction of diffraction, because the diffraction is strongly anisotropic. >> >> You've told Phaser that the model is 100% identical, from which it is inferring a rather small RMS. However, it sounds like you have greater uncertainty about the quality of your model, so you might be better off describing model quality in terms of expected RMS error. >> >> By the way, we've only just realised that Phaser wasn't dealing properly with cell content analysis for nucleic acids. The development version now takes into account the different partial specific volumes of protein and nucleic acids, but the distributed version doesn't give the right results for the solvent content. >> >> Good luck! >> >> Randy Read >> >>> On 21 Apr 2016, at 14:39, Qixu Cai <[log in to unmask]> wrote: >>> >>> Dear all, >>> >>> I have a twinning dataset with translational NCS. And I cannot get molecular replacement solution while lots of solutions with high TFZ (>10) are rejected because of packing. >>> >>> The log files of xtriage and phaser are attached. >>> >>> Could you please give me any suggestion? >>> >>> Best wishes, >>> >>> Qixu Cai >>> Email: [log in to unmask] >>> >>> <xtriage_4.log><3_phaser_MR.log> >> >> ------ >> Randy J. Read >> Department of Haematology, University of Cambridge >> Cambridge Institute for Medical Research Tel: + 44 1223 336500 >> Wellcome Trust/MRC Building Fax: + 44 1223 336827 >> Hills Road E-mail: [log in to unmask] >> Cambridge CB2 0XY, U.K. www-structmed.cimr.cam.ac.uk >> >>