Job: Reconstruct 2D Classes
Last updated
Last updated
Reconstruct 2D class averages from a particle stack.
During a job, particle images are rotated, aligned, and classified such that each particle
is in a class with particles it is most similar to, and
is rotated and translated to be in the same orientation as the other particles in the class.
These particle images are then averaged together to create the class average, which typically has a significantly better signal-to-noise ratio than any individual particle image.
Reconstruct 2D Classes uses existing 2D alignments from a 2D Classification job to produce these 2D class averages without performing the alignment step. This can be useful for a variety of reasons, including:
Producing larger versions of the class average images
Assessing how classes change as the particles are filtered via other means (such as Heterogeneous Refinement)
Particles must have existing 2D alignments from a prior 2D Classification job. If you see an error including the message:
AssertionError: Non-optional inputs from the following input groups and their slots are not connected: particles.alignments2D. Please connect all required inputs.
the particles do not have 2D alignments and must be run through a 2D Classification job first.
These two parameters set the reconstruction box size (that is, the number of pixels on each side of the 2D class averages generated by the job). At least one of these parameters must be set for the job to run. If both are set, the box size is set to be the smaller (equivalently, the lower-resolution) of the two.
Note that changing these parameters does not change the size of the output image — the class averages will have more or fewer reconstruction pixels, but the image you download will always have the same number of pixels.
Maximum resolution (A)
When this parameter is set, class averages are reconstructed at a box size such that this resolution is near the Nyquist resolution. Lower numerical values for this parameter (i.e., higher resolutions) result in 2D class average with more pixels, but increases job runtime. Note that increasing the resolution of the reconstruction will not necessarily increase the quality or resolution of the class average.
Class box size
This parameter directly sets the number of pixels on a side instead of selecting a maximum resolution.
Turning this parameter off removes the white scale bar which is typically included in the first class average of every other row.
These class averages can be used in the same way as class averages produced directly by 2D Classification or Select 2D Classes.
This job is often used to investigate how 2D Classes change as a result of some processing pipeline, but the class averages can be used in .