Figure 7. (A) Comparison of an isolated helix from the previously published reconstruction (Campbell, 2015) on the left, and the reconstruction using exposure-filtered data on the right.
The two maps appear almost identical after scaling the amplitudes using diffmap (http://grigoriefflab.janelia.org/diffmap), suggesting that in this case exposure filtering performs as well as the weighting based on B-factors implemented in Relion (Scheres, 2014). (B) Plot of FSC curves for the various proteasome reconstructions. The exposure-filtered reconstruction has a resolution of ∼2.8 Å, matching the resolution previously obtained. (C) Plot of the average particle signal-to-noise ratio (SNR) as a function of resolution. The exposure-filtered particles have equal or higher SNR than the other data sets at all resolutions. (D) Plot of FSC curves from the signal-limited data set. In this case, the exposure-filtered reconstruction is of better quality than those derived from the other data sets. Recalculating the non-filtered reconstructions with particle alignment parameters obtained for the exposure-filtered data set increases the resolution to that of the filtered data set (curves labeled ‘EF Alignment’), demonstrating that the loss in resolution was due to particle misalignments.