I’ve seen some old archived posts asking about the status of helical reconstruction in cryoSPARC. I saw one reply saying it was in the works, but that was back in '17. I was just seeing if I opened a new thread if there was any new updates.
We did start working on helical reconstruction but did not find any improvements in our attempts over existing software so we decided to postpone the project.
We may return to it eventually but it’s not a high priority currently.
@apunjani Which software packages did you compare it to? For us the user experience alone of cryosparc is better than any other package right now. So it would be interesting to hear what you compared it to. We’ve had much better luck lately with our data sets in 2D classification in cryosparc than other packages too. If the helical isn’t a priority at the moment is there any development on a better cs to star converter to make it easier to get the data into relion. We’ve been using pyem, but the star file it makes needs a lot of modification to get it to work and is missing some key values when it comes to helical reconstruction.
Thanks @hansenbry for the praise - user experience is quite important to us.
We were particularly working on trying to automatically detect/search/find helical symmetries and do ab-initio reconstruction simultaneously, but the extremely high degree of helical symmetry is very difficult to deal with in that context. We didn’t venture much into high-resolution helical refinement or compare with other programs that do that.
Which fields are missing in the cs-to-star files for helical reconstruction?
@apunjani thanks for the info. Our current workflow that we’ve been trying to develop has been to particle pick in Relion then go to cryosparc for 2D classification. When we do this we’ve noticed that _rlnAngleTiltPrior and _rlnAnglePsiPrior don’t get passed through cryosparc and into the star file. Which yields a “helix.cpp::updatePriorsForHelicalReconstruction: Labels of helical priorinformation are missing!” error in Relion. I also had to modify the star file that pyem outputs from the cs file on traditional data too because it doesn’t have the pixel size and mag column that is needed for relion. I also had issues because relion wanted the files to be .mrcs instead of .mrc that change though wasn’t anything that cryosparc could have changed.