Hi,
The align 3D job seems to work very effectively for very different orientations of asymmetric molecules, and it is fantastic that it has the option to update the particle orientations!
However, it still seems to often fail for local alignment - even when the two maps are close enough that fitmap in chimera works effectively.
It would be useful if this worked more robustly, as the option to update particle alignments is very useful for combining slightly misaligned classes separated by 3D classification.
Cheers
Oli
Hi @olibclarke,
Thanks for the report! Very interesting – is the failure very obvious (i.e. making things much worse and the volume clearly rotating shifting beyond where it should be) or subtle (slightly off rotation/shift)?
Do you find the maximum shift (A) and maximum search angle (degrees) parameters helpful to prevent this?
Best,
Michael
Subtle - it’s basically not moving it at all, and I tried restricting the search to no avail… happy to share the pair of volumes if helpful?
Hi @olibclarke,
Sorry for the delay – we’d definitely appreciate the Align 3D job, perhaps you could upload an exported version of this job? Let me know if you need access to our remote upload system, or if you still have access from a prior upload ![:slight_smile: :slight_smile:](https://discuss.cryosparc.com/images/emoji/apple/slight_smile.png?v=12)
Best,
Michael
Yep will do! Please resend upload credentials if you can, not sure where they are ![:slight_smile: :slight_smile:](https://discuss.cryosparc.com/images/emoji/apple/slight_smile.png?v=12)
Thanks Oli for providing data. We have tested out the alignment algorithm on your volumes, and identified an avenue for improvement of the Align 3D algorithm. I will triage this!
Best,
Michael
1 Like