Focused refinement?

Sure, I will try your strategy as well. If I got it correctly, you are advising me to use the cryosparc star file as input in 3d classification without alignment, with a full complex as input reference map, and a focused mask (only covering the flexible region) as input reference mask in Relion. I haven’t tried this way yet, but will it still output full complex 3d classes? Or the output would be just the flexible region? If it’s the latter, how would I merge it with the rigif part (the sharpened map from cryosparc)?


Yes, that’s right. No, it will output masked maps by default, but you can generate unmasked maps for each class using relion_reconstruct, or by subsequently running a refinement job (in either relion or cryosparc) using the particles belonging to the classes of interest.

Thanks a lot Oli for the help. Though, I haven’t used relion_reconstruct till now, but will surely try the steps you suggested.
Thanks again. Hoping to see cryosparc V2 (with focused refinement) release soon!


Good luck! Cryosparc v2 sadly does not have focused refinement at present, though it has plenty of other useful features!


I see. I thought that they might be implementing it in their new release. Yes, v2 beta has a lot of new tools. I tried it once. So much faster!
Anyway, your suggestion seem to work perfectly! But as you said, now I have the density for the flexible region only. How should I use relion_reconstruct. I assume that I do a subset selection of good 3d classes, and then run it on the star file? Any other parameters (relion_reconstruct seem to have a lot of them!) required?


Yes, that’s right.

Something like this should work:

mpirun -n 20 relion_reconstruct_mpi --i --o t100_class2.mrc --maxres 4.5 --ctf >& log1 &


Thanks yet again for a nice run-down!