e2spt_refinemulti_new question

[Madeline Rollins]

Hi all,

I would like to use the multi-reference classification program (e2spt_refinemulti_new) to get three classes from my subtomogram average. I am currently using the aliptcls3d_xx.lst file as particles without providing a reference map:

 

e2spt_refinemulti_new.py --ptcls=spt_11/aliptcls3d_02.lst --niter=5 --maxres=20 --nref=3 --loadali3d –parallel=thread:12:/home/mgrollins/tmp

 

However, after five iterations, all of my classes look very similar and have a similar number of particles. In order to better separate out the different classes, would the best approach be to increase the number of iterations? Or would it be to include a reference (i.e., the subtomogram average from that run) and mask out the region of the reference where I am expecting the most heterogeneity?

 

Also, is there any advantage to using the multi-reference classification program over the orthogonal projections/k-means classification (e2spt_classify_byproj) program?

Thank you,

Madeline

[Muyuan Chen]

Increasing iterations generally won't help, since it converges quickly. If you specify --ptcls from spt_xx and --loadali3d, it essentially use the information from subtomogram average already, so adding a reference shouldn't make a difference either. Masking obviously helps though, if you know where to mask. 

Alternatively, the initial model generation program now does classification too, and often with better convergence. Maybe it is worth a try.

e2spt_sgd_new.py spt_00/aliptcls3d_03.lst  --refine --skipali  --res 30 --ncls 2 --classify

If you are dealing with a large complex with symmetry, maybe projection based classification would be good too.

--
--
----------------------------------------------------------------------------------------------
You received this message because you are subscribed to the Google
Groups "EMAN2" group.
To post to this group, send email to em...@googlegroups.com
To unsubscribe from this group, send email to eman2+un...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/eman2

---
You received this message because you are subscribed to the Google Groups "EMAN2" group.
To unsubscribe from this group and stop receiving emails from it, send an email to eman2+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/eman2/480f9288-6cc4-4564-a524-b39437792e47n%40googlegroups.com.

[Madeline Rollins]

Thanks so much! I will try masking as well the initial model generation program. My complex doesn't have any symmetry, but I can try the projection-based classification as well.