Hi Leo,
In the polishing itself, all particles go in the reconstruction with only
a single angle (i.e. like in max-cross-correlation approaches). In
practice, that gives (slightly) lower resolutions, so your observed
behaviour is normal. You could have a look at the script on the wiki to
analyse your motion tracks
(http://www2.mrc-lmb.cam.ac.uk/relion/index.php/AnalyseMovement.csh). That
way you can analyse how far motions are not all-frame-only, and see
whether your fits in the polishing are reasonable. That is a more informed
analysis than only assessing final resolution.
HTH,
S
> Hi, we are observing something which is a bit difficult to interpret, so I
> wonder if others have seen similar.
>
> Case A. If we don't do any whole movie frame alignment before Relion, then
> after movie processing and particle polishing inside Relion, the
> resolution after movie+polishing is better than before it and gets much
> better after re-refinement of polished particles, all as expected. (All
> this is after post-processing).
>
> Case B. However, if we do whole movie frame alignment before Relion using
> Li et al. motioncorr program, then the resolution is good even before any
> movie processing in Relion but gets worse after movie+polishing. Only
> after re-refinement of polished particles it gets to be slightly better
> than in case A. What I don't understand here is why the resolution gets
> worse after polishing - even if there were no extra movements of particles
> relative to whole frame movement, b-factor scaling of frames should still
> help, and b-factors look good, as expected and roughly as they did in case
> A. We tried to vary SD and Dist parameters in movies processing to limit
> the extent of particle movement, but that does not seem to help.
>
> I guess it is possible that there are no reliably measurable (by Relion)
> particle movements apart from whole frame movements in our case and
> b-factor scaling only has minor contributions. Still, not clear why the
> resolution gets worse after polishing in case B even when possible
> particle movements are minimised (although we did not yet try to get them
> close to 0)?
>
> Thanks for any insights.
>
>
--
Sjors Scheres
MRC Laboratory of Molecular Biology
Francis Crick Avenue, Cambridge Biomedical Campus
Cambridge CB2 0QH, U.K.
tel: +44 (0)1223 267061
http://www2.mrc-lmb.cam.ac.uk/groups/scheres
|