Dear Wouter,
Perhaps the errors come from instabilities in the cluster. Or from one or
a few strange (all-black or so) particles?
You cannot make side views by throwing away top views, so either you get
small data sets by throwing away many top views, or you could more data.
Sorry I cannot be more helpful.
Best, Sjors
> Dear Sjors and others,
>
> I'm having problems when running 3D classification. The error I get
> is: "slave 23 encountered error: ERROR!!! zero sum of weights...." and
> than it stops at iteration 7. (I could also send the full error.output
> file if necesary). The problem seems to depend on the size of the
> dataset (a smaller dataset works perfect). These were my parameters:
>
> Class3D/runC25Cnewmask150 --i particlesbestclassesRUNC25C.star
> --particle_diameter 150 --angpix 1.7 --ref references/X.mrc
> --firstiter_cc --ini_high 60 --ctf --ctf_corrected_ref --iter 25
> --tau2_fudge 4 --K 4 --flatten_solvent --zero_mask --oversampling 1
> --healpix_order 2 --offset_range 5 --offset_step 2 --sym C5 --norm
> --scale --j 2 --memory_per_thread 20']
>
> I also have an additional question related to preferential
> orientations. The data clearly shows that the amount of top-views is
> about 10-20 times larger than side views. Although the 2D classes look
> reasonable, the 3D volumes are bad. My guess was to decrease the ratio
> top-side to 3:1, but the problem that I have is that the total number
> of particles decreases to 20-15 000. Are there any other tricks to
> solve this problem (besides collecting more data)? Additional: the
> programm says: 5000-10000 particles per class for 3D volumes. Although
> I have enough particles to do this, I wondered if you could also
> express this as how many different side-views/top-views there should
> be to start with 3d classification.
>
> best wishes and apologies if these questions are basic,
>
> Wouter
>
>
>
> --
> Wouter Van Putte
> Laboratory for Protein Biochemistry & Biomolecular Engineering
> Department of Biochemistry and Microbiology
> Ghent University
> K.L. Ledeganckstraat 35
> B-9000 Gent
> Belgium
>
> http://www.lprobe.ugent.be/
>
> Tel.: +32 (0)9 - 264 5175
>
--
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
|