Hello! I am trying to merge particles deriving from 2D classifications made by me and a colleague. The workflows (comprising preprocessing) were applied on the same dataset but in separate instances. I’ve imported his last “select 2D classes” in my project and then used “remove duplicate particles” to merge his particles with mine. (I guess) due to preprocessing being done independently the job was failing when inputting my imported micrographs so I’ve simply inputted the particles and specified the pixel size. Is this strategy reasonable? I ended up with less rejected particles then expected (almost 0 redundancy).
In general when using “remove duplicate particles” I always have the feeling the job is working only partially. Today I’ve tried to merge eight particles classes from four very similar ab-initio reconstructions done on the same initial set of particles (around 1 million). Merging around 1.8 millions particles with expected high redundancy (they were from very similar volume classes) I’ve re-obtained almost 850’000 particles. Given the initial number of particles and the results of the ab initio reconstructions themselves indicating that more than 500’000 particles should belong to junk classes I found the result quite strange.
I am sorry if my question is a bit convoluted and thanks in advance for your help!!!
Hi @FNai!
Let me be sure that I understand the situation correctly:
- Your colleague processed the micrographs and produced a particle stack
- You processed the same micrographs in an entirely separate CryoSPARC project (i.e., re-imported the movies, motioncorrected, etc.) to get your own particle stack
- You merged your colleague’s and your particle stacks and used the remove duplicate particles job
If that is the case, you’re right that there will be many duplicates that haven’t been removed yet. This is because CryoSPARC detects duplicate particles using the micrograph’s UID, which is randomly generated during motion correction. This means that particles picked from two different motion correction jobs on the same micrographs will not be treated as duplicates, no matter how close they are!
You can first use the Reassign Particles to Micrographs job to associate your colleague’s particle stack with your micrographs, so that all particles share the same micrograph UIDs. Then Remove Duplicates should be able to correctly handle the two stacks.
If this doesn’t work, or you have any other questions, please let me know!