Vincent, You are on the right track. Notice now that both maps at chi/kappa=180 have strong peaks at the y-axis and a “mirror” plane perpendicular to this. This is what would be expected for P21, i.e., Y is b/b*. Then depending on the orthogonalization convention of MolRep, the a- and b-axes are in the X-Z plane. Now the two maps are consistent. So going to an integration radius of 30 Å has made a marked improvement in the SRF maps. What Herman has pointed out is that you can (and should) explore different integration radii, but always being bias towards choosing a MINIMUM radius that gives you a clear and unequivocal signal.
> I'm puzzled by this radius of 30A as I read in the documentation that is > corresponds to the approximate radius of the protein. My protein is about > 130Ax40A in the closed form to 130Ax70A in the open form according to other > structures available. Shouldn't I compute maps with a radius of 70A or more? > I'm not posting here these results here as they look more "artsy". Absolutely not!!!!! Part of the problem is that almost no one looks at Patterson maps now days. To find NCS, you want to stay within the region of Patterson vectors that contains the intermolecular vectors arising from within a putative NCS system (say a tetramer), but also contains almost NO intermolecular vectors between a tetramer and a neighboring tetramer (from within the same or from a different ASU). I have always followed the rule of thumb to start with an integration radius of about 1/2 of the minimum molecular dimension (~35 Å in your case, but I have used routinely 20-30 Å for many large proteins). As Herman rightly points out, lowering the integration radius does lower the signal, but any noise/signal from cross-peaks can make the maps almost impossible to interpret. Both maps at chi/kappa=180 also have clear peaks in the X-Z plane that are 90 degrees apart. What are their peak heights compared to the origin peak? If they are essentially 100%, you may have a P222 crystal system; if not, you may have NCS that mimics P222 (all major peaks separated by 90 degrees). The NCS peaks I’ve seen are generally between 50-80% of the origin peak height. Finally, the second map at at chi/kappa=180 shows a series of peaks that form a "great circle" between the crystallographic b/b* axis and the second peak in the X-Z plane. These could be weak NCS two-folds. How many NCS systems do you suspect in the ASU? Cheers and good luck, Michael **************************************************************** R. Michael Garavito, Ph.D. Professor of Biochemistry & Molecular Biology 603 Wilson Rd., Rm. 513 Michigan State University East Lansing, MI 48824-1319 Office: (517) 355-9724 Lab: (517) 353-9125 FAX: (517) 353-9334 Email: rmgarav...@gmail.com **************************************************************** > On Jun 2, 2017, at 6:05 AM, vincent Chaptal <vincent.chap...@ibcp.fr> wrote: > > Hi Michael, > > thank you for your explanations. > > On 01/06/2017 15:41, R. Michael Garavito wrote: >> Vincent, >> >> I see a few of problems with your SRF (the maps) which would impact the >> interpretation. >> >> First you say that both crystals are processed as P21, which you would >> expect a very strong peak (100% of your origin peak) on kappa/chi=180 at the >> b* axis (one of the major axes of your map); this arises from the >> crystallographic symmetry. Where this axis is depends on what the >> conventions are of the program you are using. Your SRF has a major peak >> along Z, but is that the b* axis? (My b* axis is always placed North-South >> or your X, á la the Rossmann conventions. I don’t know the conventions for >> Molrep) Then you see major off-axial peaks. The sad thing about your >> off-axial peaks is that they are badly split. > The datasets were processed by XDS in primitive monoclinic P2 (space group > 3), the reindexed to P21 by Aimless as the most probable space group. I just > double checked that it is indeed P21, you made me doubt. I tried to look for > the conventions for Molrep but couldn't find them, I'm sorry. > Following your 3rd comment bellow, I reproduced the maps with a radius of 30A > as you suggested (see attached maps). I guess I can see the same features on > both maps, with a peak along Y that would correspond to the 2 fold axis (if > the convention is turned 90° compared to yours), and 2 off origin peaks along > X. The maps from the crystal without additive is an awful mess. I should add > that my monomer has an internal 2fold (pseudo)symmetry. Could the two main > peaks mean that I have 2 monomers, each with his own 2 fold symmetry? > > I'm puzzled by this radius of 30A as I read in the documentation that is > corresponds to the approximate radius of the protein. My protein is about > 130Ax40A in the closed form to 130Ax70A in the open form according to other > structures available. Shouldn't I compute maps with a radius of 70A or more? > I'm not posting here these results here as they look more "artsy". > > All the best > Vincent >> >> The second problem is that the second crystal shows very strong peak (1001% >> of your origin peak) on kappa/chi=180, but at different position (along Y). >> The control peak is the expected peak which arises from the P21 >> crystallographic symmetry. If the data sets were processed as P21, they >> should be indexed with the 2-fold axis along b/b*, thus the control peak is >> the expected peak which arises from the P21 crystallographic symmetry. Once >> that peak appears in the same place on each map, you can then compare the >> maps with more confidence. >> >> The third problem is that all your maps have the “appearance" of mirror >> symmetry across the x-axis, which would be expected as it is the consequence >> of the P2(1) symmetry and you are looking at a hemisphere. But it suggests >> that the 2-fold axis is along Y (b/b*). Then map 1 is missing the expected >> peak on kappa/chi=180 at the b* axis. >> >> Final comment is that the maps labeling suggest that the radius of >> integration is 62-66 Å, which is way, way too large, even for an empty unit >> cell. If it is, reduce it down to 20-30 Å to avoid intermolecular >> cross-peaks and see if the maps become clearer. I have attached a SRF map >> from a P21 crystal form (radius of integration = 20 Å, resol. = 2.8 Å) with >> the 2-fold axis indexed along b/b* (Y) from polarrfn; note the “appearance" >> of mirror symmetry perpendicular to across the b/b*-axis (North-South or >> y-axis). >> >> Cheers, >> >> Michael >> >> >> >> >> **************************************************************** >> R. Michael Garavito, Ph.D. >> Professor of Biochemistry & Molecular Biology >> 603 Wilson Rd., Rm. 513 >> Michigan State University >> East Lansing, MI 48824-1319 >> Office: (517) 355-9724 Lab: (517) 353-9125 >> FAX: (517) 353-9334 Email: rmgarav...@gmail.com >> <mailto:garav...@gmail.com> >> **************************************************************** > > -- > Vincent Chaptal, PhD > Institut de Biologie et Chimie des Protéines > Drug Resistance and Membrane Proteins Laboratory > 7 passage du Vercors > 69007 LYON > FRANCE > +33 4 37 65 29 01 > http://www.ibcp.fr <http://www.ibcp.fr/> > > <crystal-with-additive_rf-radius-30.pdf><crystal-without-additive_rf-radius-30.pdf>
