Hmmm - not much help, but MR can work with twinned data .. What's the sequence match between your models and your protein? And do you expect them to form a dimer?
Presumably you found the d3:d3 dimer using MR? I would be a bit worried that the twinning could mislead a dimer search - are the two d3-s related by (x,y,z) (-x,-y,z) and possibly a consequence of twinning imposing an extra 2 fold axes in the data? And what do you see in the self rotation function? Is there any indication of how the two copies are related? Lots of questions, no answers! Eleanor On Tue, 8 Sep 2020 at 19:53, Andrew Lovering <[email protected]> wrote: > Dear all, > > > I've a project with two historical datasets (i.e. not sure of when/if I > can reproduce to solve problem by getting better data) that are twinned. > > > The spacegroup is P32, with operator -h, -k, l > > cell = 83.5 83.5 64.2 90 90 120 > > > protein has three domains d1-d2-d3, likely 50% solvent, 2 copies in asu > > > dataset 1, roughly 3.6 angstrom resolution, twin fraction 0.25 > > dataset 2, roughly 3.1 angstrom resolution, twin fraction 0.4 > > > I've managed to solve this (unambiguously) by finding a d3:d3 dimer in > detwinned data, then seeing via phenix autobuild that it finds a d2 also in > map. At present the ability to interpret the map further stalls there, and > it can't be "the limit" because lack of crystal contacts suggest there > should be more protein to find..... > > > I realise I can go two ways, either using the detwinned data or using > twinned data and supplying the operator.....the former giving ridiculously > optimistic R-values with a suspiciously "clean map", the latter giving > noisier but maybe more informative maps. > > > Two-fold averaging is not going to be great because the d2-d3 arrangement > observed in monomer 1 clashes when placed over monomer 2 (so at present can > only average the small % of ASU between d3 copies). Critically, looking for > a second d2 using MR fails but the lack of crystal contacts suggests > success should be possible....its really unlikely that a third entity is in > there (d1 and d2 are actually 50% similar, so by searching for d2 and > failing, I'm actually missing three spots it could potentially occupy if > the protein is not proteolysed) > > > Sorry for long email but the devil is in the details. My question would > be, what is best practice in a low res poor twinned dataset where you can > only fill roughly just over half of what should be in there? > > > Feel free to tell me its doomed! > > > Best wishes & thanks in advance > > Andy > > ------------------------------ > > To unsubscribe from the CCP4BB list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1 > ######################################################################## To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1 This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
