My sample is long filaments entire length around 2.3 micron, and we usually reconstruct 5 crown (which is 725A), so we pick a box size that can fit 725A. we are actually picking segments of filaments instead of entire filament. When I use the filament tracer to pick the particles (in this case segments), I space them like 1 crown (145A) which is the separation between picks in terms of diameter in the filament tracer job, so I put 0.725 coz my particle diameter is 200A, so if I want it to place each box around 145A spacing, it is 0.725 (in terms of diameter).
Which is not leaving out the neighboring particles that might be sitting right next to each other. Since I am reconstructing 725A worth of spacing of the filament, and I am only separating them at 145A, along the filament there’s multiple picks in same filament, like distance between one box center to next box center is 145A, and my reconstruction is 725A worth of space, so there’s definitely repeat, I wish I could only pick particles that are 725A apart that means no repeat in 5 crown. But when I set the separation parameter spacing as 725A, it doesn’t consider 725A distance in one direction, along the same filament, it considers 725A radially, like a big circle so the neighboring filament is not getting counted even if they are part of another filament.
So my question, is there any way I can pick filament segments specifying that I want each box to be placed at 725A distance from each other along the filament instead of considering that distance radially which excluding my neighboring filaments which might just sitting in close proximity?
If Cryosparc don’t have that feature, is there any other filament picking software, where I can direct the software to pick in every 725A distance in one direction (along the filament length) instead of considering 725A distance in all direction?
Could you post (or DM me, if you prefer) the “Diagnostic Plot” for the first 3 micrographs from each job (the 725Å spacing one, and the 145Å spacing one)? It would help us get a sense of what might be going wrong in the greater-spacing job.
Hi, I actually meant for remove duplicate job rather than filament tracer. However I am posting both for filament tracer, and remove duplicate plots as well.
First three is for 145A spacing
for 725A separation in remove duplicate job (after running 2D class)
you can see some good neighboring filaments are not kept due to 725A radial distance (that’s what I think, not sure entirely), I would like to keep the segments at 725A in each filament, so I would like it to count the 725A distance in one direction (along the filament), not in radial distance.
As you can see, it got rid of a loot of good particles, and from 800k particles, it ended up keeping 96k only.
Remove Duplicates is not usually recommended to use with filaments, partially for the reason you stated, and also partially as it isn’t necessary: CryoSPARC’s helical refinement ensures that particles belonging to the same filament don’t end up in the same half-set (as long as picking metadata is available).
Instead, the recommended approach if you desire particles spaced by 725Å is to simply run the Filament Tracer with an appropriate “Separation distance between segments (diameter)” parameter, as you did already. To my eye, the results from tracing at 725Å look great; would you be able to proceed by extracting these particles without an additional remove duplicates job?
Yes, I will do that. Thank you. The whole reason I even ran remove duplicates coz my FSC wasn’t reaching zero while I ran Homogeneous refinement, which I suspect due to duplicate particles, so I ran remove duplicate, then it hit me how much distance I wanna keep each segments and all.
Anyway, I will use 725A separation on filament tracer and then I will do helical refinement. Thanks.
That sounds good! Yes – Homogeneous Refinement isn’t suggested for filaments, for this reason. If you want to run a refinement without helical symmetry, you can indeed still use Helical Refinement, and simply leave the symmetry parameters empty.
