Admittedly this is a niche request: for some particles, when oriented correctly the 2D slices on job card are sufficient to know the results of the jobs. +/- domains, conformational variability, complex formation, whatever. It would be cool to have a jobtype that loads a volume, allows you to specify position in the box, or even just orientation relative to the gui, so we can get the 2D slice views that are most interesting for all downstream jobs. Like, generate templates - select template consistent with favorite forward view? or load to 3D viewer and set “preferred forward face”?
Use example: run ab initio, slices are uninformative, load to the job type and 1) flip z or don’t 2) forward orient to user-defined view, use as reference for all remaining 3D jobs and from now on see which classes have conformation A immediately without downloading 100 maps for 3D inspection.
Could also somehow be of use for helical reconstructions I’m sure. Obviously this can already be done by resampling within chimera and re-uploading the volume and to a lesser degree with the volume alignment tool.
wow. did not know that existed, thanks!. yes it’s less error prone but still manual. I’m thinking of finding 10 classes out of 200 that have desired blurb in the projection/slice for instance.
I see, when I want to do that I re-do the classification with the correct starting volume, then I know the classes where I assigned that reference are the ones I want. E.g. I gave interesting class as volume 0 in input list, I know class 0 will have the particles for that reference at the end.
same, sometimes. but I’m talking 3D class PCA start no reference. alignments establish the orientation, but there is no “best reference” class that will have the desired particles at the end. Basically once ab initio is done, the next 500 children 3D jobs will have variations of the same projection displayed on the card, and sometimes I get lucky and it’s informative. there are whole workspaces where I can follow my logic retrospectively without opening a single card. and whole workspaces where I’m looking at an uninformative view. Now, put that way it might be of value personally to just resample ab initio to the box of a best case orientation from prior processing.
slightly unrelated: ab initio completely changes the orientation in the box / slices / projections at the final iteration? no mask generation prior to completion I guess… intended?