[3dem] Relion-4 3D classification

[Takanori Nakane]

Hi,

First of all, please upgrade to 4.0.0 from the beta version.

How different are the pixel sizes?
Are they from different data collection sessions?
Are they from different detectors?

When many parameters of the experiments are different, they introduce
systematic differences to the particles, which are larger than
conformational changes of interest. In such cases, classification is
very hard.

Best regards,

Takanori Nakane

On 2022/10/15 7:49, Reza Khayat wrote:
> ​Hi,
> 
> We are having issues properly running 3D classification on data collected at 3 different pixel values -we are using 4.0-beta-2-commit-2dffb8?. The classification runs with no errors; however, classification is performed on the pixel size rather than then conformation of the complex (i.e. each class is dominated by data with a different pixel size). Web searches identified the following entry in github/3dem/relion/issues (https://urldefense.com/v3/__https://github.com/3dem/relion/issues/530__;!!Mih3wA!GdOSCrM397i8nVaqvkqjA0gF9XHLTLvYS4O-up-fozGgTt6r1j5PRIbijW2L2A-_82It82HymejyOf2MLmuWUYHo$  <https://urldefense.com/v3/__https://github.com/3dem/relion/issues/530__;!!Mih3wA!GdOSCrM397i8nVaqvkqjA0gF9XHLTLvYS4O-up-fozGgTt6r1j5PRIbijW2L2A-_82It82HymejyOf2MLmuWUYHo$>   ). However, the solutions provided did not alleviate the issue. The command we used, with the suggested modifications, is:
> 
> 
> `which relion_refine_mpi` --o Class3D/job134/run --i JoinStar/job028/join_particles.star --ref Refine3D/job034/run_half1_class001_unfil.mrc --firstiter_cc --ini_high 60 --dont_combine_weights_via_disc --pool 30 --pad 2  --skip_gridding  --ctf --iter 25 --tau2_fudge 4 --particle_diameter 315 --fast_subsets  --K 10 --flatten_solvent --zero_mask --oversampling 1 --healpix_order 2 --offset_range 5 --offset_step 2 --sym C1 --norm --scale  --j 2 --gpu ""  --pipeline_control Class3D/job134/?
> 
> 
> I appreciate the help.
> 
> 
> Best wishes,
> Reza
> 
> Reza Khayat, PhD
> Associate Professor
> City College of New York
> Department of Chemistry and Biochemistry
> New York, NY 10031
> 
> _______________________________________________
> 3dem mailing list
> 3dem at ncmir.ucsd.edu
> https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem