*** For details on how to be removed from this list visit the *** *** CCP4 home page http://www.ccp4.ac.uk ***
As several people have suggested, this structure may well be twinned, possibly made difficult to detect (as often happens) by the presence of pseudo-translational NCS. This import thing in such cases is always to process the data in the true space group and to use this true space group for the structure solution (e.g. by MR) and refinement. In addition to the possibility that the true space group is P6 as suggested by Eleanor, it could also be P321 or P312, though the latter is uncommon. In both these cases the necessary TWIN matrix for SHELXL is: TWIN -1 0 0 0 -1 0 0 0 1 Other interesting possibilities (like R32 with rhombohedral obverse/reverse twinning) can be eliminated in your case because you do not see any systematic absences. George Tiancen Hu wrote: > *** For details on how to be removed from this list visit the *** > *** CCP4 home page http://www.ccp4.ac.uk *** > > > Dear all, > > Thanks for all the quick and instructive replies. Here is a summary of all > the answers I received. > > Gerard DVD Kleywegt suggested that the data might be a combination of > twinning and pseudo-symmetry, and posted a reference: > http://xray.bmc.uu.se/cgi-bin/gerard/reprint_mailer.pl?pref=76 > > Paul Adams suggested trying mmtbx.xtriage program of PHENIX and it is good > for handling twinning data, and that translational symmetry and a real lower > symmetry of P6 could exist in our data. > > Trevor Sewell suggested 1) determining the space group with Phil Evan’s > POINTLESS program. 2) asking PHASER to check the translation function for all > rotation peaks (ccp4i interface doesn’t support this) > > Wei Zhiyi suggested that the monomer could have a different conformation in > part of the first one structure and two strategies could be tried: 1) Cut off > the overlapped part of the second monomer; then refine it again. 2) Feed the > model phase with r=.38 to ARP/wARP for autobuilding. > > Sameeta Bilgrami suggested that there might be some other protein attatched > to our protein from their real experience, and arp/warp (fed with the MTZ and > refined PDB) and manual building could be tried to model the extra density. > > Ezra Peisach suggested trying MR in all the possible spacegroups (e.g.: > P6(5)22, P6(3)22 etc.). Sorry I forgot to mention that we have tried all the > possible screw axises and only P622 (or P6) gave the highest translation > score and explainable density. So I guess there is no screw axis in our > crystal. > > Joshua Warren suggested 1) the 46% partial twinning of the P6 dataset is a > result of its actual P622 symmetry, and the negative perfect twinning test > could reject the possibility of twinning. 2) All the possible screw axised > could be tried. 3) keep refining the first monomer and then try fitting the > second by hand. > > Herbert J. Bernstein used > http://www.bernstein-plus-sons.com/software/iterate/ inputting my cell > parameters and IT numbers 12, 36, 13, 33, 39, 10 and 28 as possibilities. > > Sandra B. Gabelli suggested checking the screw axis and manually fitting the > extra density which could stand for a different dimer. > > Liu Jiaxin suggested trying R3/H3 spacegroup. > > Eleanor Dodson suggested that the data could probably be a twinned P6 one and > the extra density could be the twin “ghost”. She recommended us to first work > in the original P6 data and get a solution of the dimer, and then use the > twin refinement of SHELXL. The detailed procedures are: > 1) run SHELXPRO to convert the coordinates to the SHELX style. > 2) Then add to the ins file two lines: > > TWIN 0 1 0 1 0 0 0 0 -1 ( This gives twin law k,h,-l - the only one > allowed in P6i) > BASF 0.3 ( This is the twinning fraction which will be refined..) > ..... > > DELU $C_* $N_* $O_* $S_* > > etc .. > > James Sandy presented us a very interesting case in which two different > lattices co-existed in one crystal, with one lattice forming within another > one. These two lattices was separated by an operator of (0.5, 0.1086, 0.0). > This really opened our eyes and gave us a good hint, thanks Sandy:). It would > be great if you could explain a little bit more detailed for us about how you > discovered this? > > Petrus H Zwart suggested that if the true symmetry is P622, then twin > refining the P6 data using SHELXL will also refine the twin fraction to 0.5. > He also recommend using Rvs R statistic (if the model is complete enough) to > check out if the data might be twinned. > > I am really grateful to receive so many suggestions, and now we have a lot to > try. I will post our result as soon as we find a way to solve this problem. > > Thanks again! > > Tiancen Hu > Shanghai Institute of Materia Medica > Shanghai 201203 > P.R. China > > > > >>*** For details on how to be removed from this list visit the *** >>*** CCP4 home page http://www.ccp4.ac.uk *** >> >> >>Dear all, >> >>We recently collected a set of data belonging to P622 group (as far as I am >>concerned), and we're trying to solve the structure using a 46% sequence >>identical template with molecular replacement. But some peculiar problems >>occurred during the process which is beyond our ability to solve, so I hope >>someone could give us some advice. Here is the case: >> >>1. Data information >>(1) Spacegroup: P622, Cell parameters: 147.97 A, 147.97 A, 163.47 A; 90 >>degree, 90 degree, 120 degree >>(2) Resolution: 2.4A, Completeness: 99.8%, Rmerge: ~0.1, Multiplicity: ~13 >>(3) Data processing softwares: Mosflm and Scala. We've also processed the >>data in P3 and P6 spacegroup, the data qualities were similar as the P622 one. >>(4) Pseudo translation: SFcheck and molrep haven't detected any pseudo >>translation. >>(5) Twinning: We have tested the twinning possibility of the data integrated >>and scaled in P6 spacegroup using Detwin and YEATES server. From the YEATES >>server, partial merohedral twinning test results suggested a twinning >>fraction of 0.46, similar as the Detwin result, while perfect merohedral >>twinning test results suggested untwinned data. Sorry for my ignorance, but >>could our data be a perfectly twinned P6 data? How can I find out? >> >>2. Template information >>(1) 357 residues (359 residues for our protein), 46% identical in sequence >>(2) Its biologically functioning unit is a dimer related by 2-fold axis, >>which is also true for our protein as is determined by biochemical >>experiments. The crystal structure of the template contains one monomer per >>ASU, but the biological dimer could be reconstituted by relating two monomers >>via a 2-fold crystallographic symmetry axis. >> >>3. Matthews coefficient test suggested two (24% probability) / three (76% >>probability) monomers per ASU. >> >>4. MR with one monomer first >>(1) Softwares tested: Molrep, Amore, Phaser, CNS, the results were similar. >>(2) Cross rotation gave two obvious peaks, differed only by about 20 degrees >>in alpha rotation angle. The 1st peak is much larger than the 2nd, but the >>gap between the 2nd and rest of the peaks was also distinct. >>(3) Translation with the 1st largest peak gave a good hit. After refining the >>fitted model, we could unambiguously position our model in the clear electron >>density map, including a ligand. The Rwork and Rfree were 38% and 42% >>respectively. This monomer forms a biological dimer with its >>crystallographically 2-fold related symmetry molecule. >>(4) After observing the map, we could find a large area of unexplained >>electron density suitable for another monomer but with poor constitutivity. >>When we tried to find the 2nd monomer with the 2nd largest peak fixing the >>1st model, the results produced were all unacceptable because of clashes with >>symmetry molecules. >> >>5. MR with a dimer (made by crystallographically relating two template or >>refined monomers) >>(1) Cross rotation gave one high peak. >>(2) The models after translation were all rejected by the softwares because >>of almost perfect overlap with symmetry molecules. >> >>Could any one give us some suggestions on what is going wrong with our >>process? Where does the problem lie? Is it space group, twinning data, >>pseudo-translation or something else of the data? >> >>Thanks in advance! >> >>Tiancen Hu >>Shanghai Institute of Materia Medica >>Shanghai 201203 >>P.R. China >> >> >> >> >> > > = = = = = = = = = = = = = = = = = = = = > > > 致 > 礼！ > > > Tiancen Hu > [EMAIL PROTECTED] > 2006-11-14 > > -- Prof. George M. Sheldrick FRS Dept. Structural Chemistry, University of Goettingen, Tammannstr. 4, D37077 Goettingen, Germany Tel. +49-551-39-3021 or -3068 Fax. +49-551-39-2582
