I recently solved a 4-identical chain protein structure. First I got it from initial model with NCS, good enough. Then from the initial model I process it without NCS, quality better than solved with NCS.
Smith At 2015-04-30 02:07:31, "Eleanor Dodson" <eleanor.dod...@york.ac.uk> wrote: Hmm - I think these peaks MUST be related to some internal symmetry in the 1200 aa solved structure. Is there some arrangement of helices or other features which are replicated in another part of the structure? A phi value of ~ 19 degrees can't be explained by a different related space group I don't think Eleanor On 29 April 2015 at 15:56, Ian Tickle <ianj...@gmail.com> wrote: Peer, you didn't say which program you are using for this? Polarrfn or Molrep? Do you get the same results with both programs? Also did you try sharpening the Fs with Ecalc and/or using all your data? In my experience sharpening works better with self- than with cross-rotation functions because the differences between NCS-related monomers are usually much less those than between non-isomorphous ones. Cheers -- Ian On 29 April 2015 at 15:00, Peer Mittl <mi...@bioc.uzh.ch> wrote: Zbyszek Otwinowski and Fred Vellieux suggested to run the self-rotation on Fcalcs. This suggestion "solved" the problem, since there are similar peaks on the kappa=180° planes as well. However, I wasn't able to get rid of those peaks by playing around with resolution and integration radius. I must say that I am surprized, because - as Eleanor pointed out - I also expected to find peaks on the kappa=180° planes only in case of P6522 symmetry. Anyway, this experience reminds me to run some simple tests beforehand. -Peer On 29.04.2015 15:31, Eleanor Dodson wrote: Well - PG P6/mmm (possible SG P6522) will have peaks at kappa = 180 omega = 90 phi = 0 30 60 etc.. But if there is only one molecule / asymm unit there cant be an extra 2-fold. How big are the relative domains? Your interesting domain couldnt just be cleaved off could it? Eleanor On 29 April 2015 at 12:59, Peer Mittl <mi...@bioc.uzh.ch <mailto:mi...@bioc.uzh.ch>> wrote: We are working with a multi-domain protein crystallized in SG P6_5 with one molecule per asymmetric unit. The structure was refined at 2.00 A resolution with reasonable R-factors but unfortunately the domain we are most interested in seems to be disordered. Interestingly, the self-rotation function shows peaks on the kappa=180° plane (omega=90°, phi=19° (and every 30°)), with more than 50% origin peak height. Therefore, we are wondering if perhaps the space group assignment might be sub-optimal. Any explanations how these self-rotation peaks could occur and how we could extract meaningful information to resolve the disordered domain are welcome. Best regards, Peer P.S. Some additional information: pointless suggests SG P6_5, the data doesn't seem to be twinned (L-test), the refined part of the structure has no "internal symmetry" and refinement in P1 doesn't reveal the "lost" domain.
