Hi Dieter,
Thanks. I think it is best practice to continue to ask questions about
relion on the ccpem list, and only use the bitbucket issue tracker for
confirmed bug reports. If in doubt, ask here first, and if we think it
is a real bug, then we can ask you to report it on bitbucket. That way,
ccpem will remain the archive where people should search when they have
problems with running the program.
To answer your question. Probably the maxima of the probability
distributions (rlnMaxValueProbDistribution in data.star, and the average
rlnAveragePmax in model.star) are all very low? That's because the maps
are (still?) very similar, and the particles have similar weights
contributing to each class. The number of particles in each class (based
on rlnClassNumber in the data.star) then becomes not very informative.
The accumulated weight for each class (rlnClassDistribution in the
data_model_classes table of the model.star) is what really counts. So,
if rlnAveragePmax is indeed very low, then this is normal behaviour.
HTH,
Sjors
On 11/16/2016 11:25 AM, Dieter Blaas wrote:
> Dear all,
>
> Sjors asked me to post my question here:
>
> When running a 3D classification (C1) on relion2 I noticed that all
> volumes looked very similar (so far at iteration 5), despite some of
> the classes containing extremely few particles (according to
> run5_data.star but NOT to run5_model.star!). Please see the first
> lines of the respective files and the particle counts of the
> respective classes (attached). Is this apparent mismatch between
> run_data.star and run_model.star a normal behavior? To my
> understanding model.star shows the fraction of particles in the
> respective classes and in data.star the respective classes are
> indicated by the number in _rlnClassNumber!
>
> thanks, Dieter
>
> ------------------------------------------------------------------------
>
> Dieter Blaas,
> Max F. Perutz Laboratories
> Medical University of Vienna,
> Inst. Med. Biochem., Vienna Biocenter (VBC),
> Dr. Bohr Gasse 9/3,
> A-1030 Vienna, Austria,
> Tel: 0043 1 4277 61630,
> Fax: 0043 1 4277 9616,
> e-mail: [log in to unmask]
> ------------------------------------------------------------------------
>
--
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
|