FIGURE 9. Bead model, ensemble, localization probability and localization volume.

a, Top: two representative bead models of the NPC (excluding the FG-repeat regions) from the ensemble of 1,000 superposed structures satisfying all restraints (Fig. 8b). The eight positions of three sample proteins (Nup192, Nup57 and Nup85) on the cytoplasmic side are shown, with a detailed view of the bead representation of one copy of Nup85 at the bottom. b, The localization probability for each protein type is obtained by converting the ensemble into the probability of any volume element being occupied by the protein. Shown are contour maps of the cross-sections in the plane parallel to the equatorial plane that contains the maximum value of the protein localization probability. c, The localization volume of the sample proteins, derived from the localization probability. The volume elements are first sorted by their localization probability values. The localization volume then corresponds to the top-ranked elements, the volume of which sums to the protein volume, estimated from its molecular mass. The localization volume of a protein reveals its most probable localization. Because of the limited precision of the information used here, the localization volume of a protein should not be mistaken for its density map, such as that derived by cryo-EM.