(A) Classification workflow for the simultaneous generation of 30 3D models from the complete dataset of particles using RELION. Models were built using DNA-dependent protein kinase catalytic subunit low-pass filtered to 60 Å as an arbitrary reference model.
(B) Top 3 models generated using RELION. Models were scored based on their rotational-translational error (a measure of the internal consistency of the model; see STAR Methods). The distribution of model error scores was compared to models generated using random particles from our template-picked data.
(C) 30 classes generated using RELION from the complete template-picked dataset of particles with the reference model shown in gray. Models are colored by their rotational-translational error and are unrelated to colors in (A) and (B).
See also Figure S3.