Dear Harm, In my experience there is no use in modeling 4.5% twinning. It will not change the refinement and is most likely an artefact. IMHO you have the following options: 1) refine in P3221 with overlapping alternative conformations. You could also give the clashing chains an occupancy of 0.5 and not model the alternative conformation if it is really disordered. This is what I did in some cases. 2) refine in P32. Here you have two monomers in the ASU and you could assign the alternative conformations to separate monomers, what you already sort of did. While I believe these conformations have a random distribution, you will have a "twinning" fraction of 50%. If I understand Refmac correctly, the resulting maps will have been modified by Fcalc, so the symmetry of the maps will be broken and they will be easier to interpret.
Good luck! Herman -----Original Message----- From: Harm Otten [mailto:[email protected]] Sent: Friday, September 28, 2012 9:25 AM To: [email protected] Cc: [email protected]; Schreuder, Herman R&D/DE Subject: Re: [ccp4bb] Space group choice with respect to physical plausability Thanks for the feedback. Twinning is found to be 4.5% with phenix.xtriage. I shall model it, even though I am not so eager to use that option in Refmac. I cut the N-terminal in one and used modeled it for the other monomer in P 32, but were not able to refine it better. The 2-fold symmetry is eventually present, which already let to some group internal discussions. You will see a properly modeled PDB entry at one point! Have a successful day Harm Otten -- On Mon, Sep 24, 2012 at 5:26 PM, <[email protected]> wrote: > Dear Harm, > > As Edwin pointed out, there might not be any non-crystallographic symmetry to > break, because it is crystallographic. While it is clear that no two > N-terminii can be in the same electron density at the same moment, there is > no reason why this symmetry-breaking should happen in a regular, orderly > fashion throughout the crystal. E.g. that all even molecules in a row would > have conformation A, and all odd molecules would have conformation B. > > What most likely happened, is that randomly, the N-terminus would either > occupy conformation A (in the electron density where it is fitted now) and > randomly in some other orientation outside this electron density. This > electron density A would be strong (100% occupancy; 50% from either > neigboring molecule), while the electron density B would be weak (only 50% > occupancy). I had a similar case and the only reason which convinced myself > to fit alternative conformations, was the physical plausibility argument. > > So for the overlapping N-terminus and the surface lysines, you have to > generate alternative conformations. As I said, the electron density for the > non-overlapping conformation might be quite weak (or even disordered), so you > may have to scroll your contour level quit a bit down. You also have to make > sure that your refinement program recognizes what is going on and does not > push the molecules apart. For buster you have to use the gelly EXCLUDE > keyword, refmac will probably do it automatically. > > Good luck! > Herman > > > > -----Original Message----- > From: CCP4 bulletin board [mailto:[email protected]] On Behalf Of > Harm Otten > Sent: Monday, September 24, 2012 1:10 PM > To: [email protected] > Subject: [ccp4bb] Space group choice with respect to physical > plausability > > For a protein dataset we find an "overlapping" N-terminal stretch of 6 > amino acids in P 32 2 1 (1 monomer/asymmetric unit ASU). We did > process the same case in P1 (6 mon./ASU), C2 (3 mon./ASU) and P3 (2 > mon./ASU) to resolve the physically and chemically hard to explain overlap of > the N-terminal and some surface lysins. We did try to mutate the B-factor > outliers to alanins and also to mutate all of the B-chain to alanins/glycins > (using just the backbone and C-alphas) and keep the A-chain untouched. This > did not break the symmetry, i.e. the strong NCS relationship between the > monomers. > > Some key values are highlighted in this table: > > # Molecules Space Group Cell R-Merge > MR > correlation # Clashes B-factor outliers R-factor R-free > (No.) a b c alpha beta gamma > > 70 Å2 > 6 P1 (1) 95.7 55.4 73.1 90 90 90 > 4.4 0.227 A/B > many: B/C 2 n/a n/a n/a > 3 C2 (5) 95.7 55.4 73.1 90 90 90 > 2.7 0.594 > none 8 /3mol 22 30 > C222 (21) 55.4 95.7 73.1 90 90 90 > 43.9 n/a > n/a n/a n/a n/a > 2 P32 (145) 55.4 55.4 73 90 90 120 > 3.4 n/a > N-term 13 /1mol 29 34 > 1 P32 2 1 (152) 55.4 55.4 73 90 90 120 > n/a > n/a n/a n/a n/a n/a > 1 P3 2 1 (150) 55.4 55.4 73 90 90 120 > 3.7 0.359 N-term > 1 P32 1 2 (151) 55.4 55.4 73 90 90 120 > n/a > n/a n/a n/a n/a n/a > > > Why is it so hard to break the symmetry for two (seemingly) different > monomers? > > Thanks everybody for the multitude and quality of your answers to my question > some time back. One can really find the "authority" on hands-on questions and > standards in this bulletin board and the wwPDB. > > Have a successful day > > Harm > > --- > Harm Otten, PhD > Department of Chemistry > Universitetsparken 5, Office C316 > 2100 Copenhagen, Denmark > # +45 35 32 02 86 > fax +45 35 32 03 22 > email [email protected] > web www.harmotten.com > Please consider the environment before printing this email.
