Interactive Job: Select 2D Classes

At a Glance

Select particles to include or exclude based on the quality of their 2D class average.

Description

In general, it is exceedingly difficult to assess the quality of individual single particle cryo-EM images; low signal level, high noise, and unknown particle pose make it challenging to determine if an intact particle is present in each image. However, when particles have been grouped and aligned in a 2D Classification job, the quality of those class averages can be more readily apparent.

Select 2D Classes takes advantage of the significantly improved signal-to-noise ratio of class averages by allowing you to interactively keep or discard entire classes of particles based on the quality of their class average.

Inputs

Particles

Particles must be 2D classified, and should come from the same job as the 2D class averages.

2D Class Averages

Class averages will be used to select particles, and should come from the same job as the Particles.

Commonly Adjusted Parameters

Auto Thresholds

Typically, Select 2D Classes will be run as an interactive job (in which case these parameters should be left blank. However, in some cases (such as when included in a Workflow) it can be beneficial to automate the class selection process based on simple thresholds. The Auto Thresholds enable this workflow. Note that setting either of these parameters will skip the interactive mode.

Setting Classes where resolution better than keeps only particles which belong to classes for which the Fourier ring correlation estimated resolution is better than (i.e., a lower number than) the given threshold.

Setting Classes where particle count higher than keeps only particles which belong to classes which contain greater than the given number of particles.

Setting both parameters keeps only classes when both conditions are met. Put another way, the thresholds are AND, not OR.

Interaction

When the interactive panel loads, the main panel displays the 2D class averages. If there are too many to fit on the screen, you can scroll through this panel to view them all. Above this is a toolbar with options to filter, sort, and select the class averages. The toolbar also contains a count and percentage of the particles currently selected.

Above each class average, the following information is displayed:

• The number of particles in each class.

• The Fourier ring correlation resolution of each class, in Å. This is the resolution for which the correlation between half sets is 0.5.

• The Effective Classes Assigned (ECA) for each class. The Effective Classes Assigned for a given 2D Class average is the median Effective Sample Size (ESS) for particles in that class (see 2D Classification for an explanation of ESS). A higher ECA means that particles assigned to this class have, on average, less certain assignments than for a class with a lower ECA.

Clicking anywhere on an unselected class average selects it:

while clicking anywhere on a selected class average un-selects it:

Buttons in the top toolbar allow you to sort the class averages by number of particles, resolution, or ECA in ascending or descending order:

Right-clicking a class average opens a context menu in which you can select or un-select classes with better or worse values for particle number, resolution, and ECA. In combination with sorting, these can be a powerful tool to quickly select a large number of classes:

If most of the particles in the 2D Classification job were good, there may be only a few junk classes. In this case, it can be useful to use the selection tools in the toolbar to select all classes then unselect the bad classes, or to select the bad classes and invert the selection.

Once all classes you wish to include are selected, click the green Done button to finish the job and produce the outputs.

Outputs

Particles selected

Particles assigned to the selected classes are in this output. The particle information is unmodified — this is merely a subset of the input particles.

Templates selected

The selected class averages are in this output.

Particles excluded

Particles assigned to an unselected class are in this output. The particle information is, again, unmodified.

Templates excluded

Unselected class averages are in this output.

Common Next Steps

The selected particles are often carried forward for downstream analysis, either more 2D classification or Ab initio reconstruction to move into 3D. The selected templates can be provided to Template Picker to re-pick particles and potentially find a greater number of high-quality particles.

Recommended Alternatives

When multiple datasets are collected on the same target, 2D class selection can be automated with Reference Based Auto Select - 2D, available in CryoSPARC v4.5 and later. To automate class selection in this way, one could follow this general workflow:

1. The first time a dataset is collected, the classes are selected manually as usual.

2. Once a 3D reference is available, the first 2D Classes and the final reference are used to determine the appropriate selection mode and threshold values for the target

3. In subsequent datasets, the reference and thresholds in step (2) can be used to automatically select class averages.

References

1. Kumar, K. et al. Structure of a Signaling Cannabinoid Receptor 1-G Protein Complex. Cell 176, 448-458.e12 (2019).

2. Campbell, M. G., Veesler, D., Cheng, A., Potter, C. S. & Carragher, B. 2.8 Å resolution reconstruction of the Thermoplasma acidophilum 20S proteasome using cryo-electron microscopy. eLife 4, e06380 (2015).