Combining/merging particles into new datasets, plus re-stacking

The latest version of cryoSPARC (v0.5.6) implements a new feature allowing for merging/combining particles across multiple datasets (perhaps from 2D classifying multiple sessions of data) into a new dataset. This feature can also optionally re-stack the selected particles on disk into a new particle stack, so as to save space after cleaning/classifying particles.

To use the new feature, select a dataset from the Datasets page, then go to the experiments page. Select particle sets you wish to combine. Then open the Datasets page in a new tab, and select the second dataset from which you wish to combine particles. Then go back to the first tab, and the second dataset will now be selected, but the selection cart will contain particles from the first dataset. Now you can select particle sets from the second dataset. Repeat this with as many datasets as desired, but be careful not to navigate away from the experiments page in the first tab, because this will clear your selection cart.

Once you’ve selected the desired particles, click on the New Dataset from Particles button above the selection cart. This will start a new job that will create the new dataset. You must provide a name for the new dataset (can be changed later) in the job parameters, as well as specifying if you wish the particles to be restacked (they will be written out to the output directory of the new dataset job) or if not (in which case the job will only create new csv files with particle metadata, pointing to the original location of the particles on disk). In case you made selections of particle sets that included assignment probabilities (for example from an ab-initio reconstruction), you can also specify the assignment probability threshold (or a new option, to take only particles that had their max assignment probability in the selected classes). Once parameters are set, Launch the job and the new dataset will be created. File copy progress in the case of restacking shows in the streamlog, and can take some time depending on disk speeds and how many particles you have.