Fig. 2

The BOF results for simulated and experimental pure noise data. a A schematic flow diagram showing that “particles” were picked by FLC from pure-noise micrographs, using a single projection of the HIV-1 envelope glycoprotein (Env) trimer as a template. The picked particles were subjected to MLE alignment, using different starting references. b-d The FLC-picked particle set, derived from the simulated Gaussian-noise micrographs was aligned by MLE, starting from a noise image randomly chosen from the particle set (b), a Gaussian circle (c), or the average of the picked particles (d). The starting reference for MLE optimization is shown in the first column. Each row shows the history of the MLE-aligned class averages at the indicated iterations of optimization, ending with the respective converged class averages in the far-right column. e-g The FLC-picked particle set derived from the experimental ice-noise micrographs and aligned using MLE, starting from a noise image randomly chosen from the particle set (e), a Gaussian circle (f), or the average of the picked particles (g). The averages shown in (d) and (g) appear as an FLC-generated replica of the 2D template used for particle picking