Dear Takanori and Randy,
Thanks a lot for your answer and suggestions.
Indeed, running a round of 3D classification w/o alignment sounds like a clever way to generate 3D reference out of a set of particles although getting a reference is less of a problem in my case (I could also simply scale the volume obtained with binned data to get the starting reference). I am more worried about the lost alignments of the particle images. The subtracted/binned particles were extensively classified and refined focusing on the region of interest. Since I have around 6 to 8 structures at this level I don't want to start the alignments all over and do the global angular search. I was hoping that I can un-bin the particles and use the poses found with the binned data to do a quick local angular search using un-binned data to push the resolution a bit further.
Takanori, as you suggested I had a look at the problematic star files, compared them with original particle metadata and do a few tests. The only differences that I can find between the problematic subtracted particle star files and non-subtracted ones are the rlnOrignXAngs and rlnOrignYAngs columns. If I re-extract at the full box size a set of binned particles that before that went through signal subtraction job then the map reconstructed with those particles looks pretty much nonsense. I managed to make it look normal if I copied the rlnOrignXAngs and rlnOrignYAngs from the original star file (ie., the one used as an input to the subtract job). I noticed that even if I revert the particles_subtracted.star to original.star using the --revert option, the rlnOrignXAngs and rlnOrignYAngs stay the same ie., they are not reverted to the original values. Is this intentional? It seems like this is the cause of my problems.
I am curious what is be the best or commonly used strategy to work with signal subtracted binned particles? What is the proper way to un-bin them once the heterogeneity is solved and particles have been separated into more homogenous subsets?
Best regards,
Piotr
______________________
Piotr Draczkowski, PhD
Academia Sinica
Institute of Biological Chemistry
128, Sec. 2, Academia Road, Nankang, Taipei 115
TAIWAN
Faculty of Pharmacy with Division of Medical Analytics
Medical University of Lublin
POLAND
########################################################################
To unsubscribe from the CCPEM list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCPEM&A=1
This message was issued to members of www.jiscmail.ac.uk/CCPEM, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
|