Extended Data Fig. 6: CryoEM 3D class distribution of particles.
a, Representative processing pipeline of hairpinD-U7 showing the presence of both vase and prong conformation. b, Standardized data processing workflow to compare 3D class distribution of particles (Methods). Good particles picked by WarpEM’s neural network-based picker first underwent 2D classification in cryoSPARC. Particles from classes with high estimated resolution and particle number were selected for ab initio reconstruction with 4 to 5 classes (Methods). The resulting initial models were then used as starting references in a heterogeneous refinement with the full set of good particles. The resulting refinements were evaluated for their similarity to the prong or vase conformation and the proportion of particles belonging to each class plotted in c. c, Comparison of particle distribution in different datasets. Numbers #1 and #2 indicate repeat data collections of the same complex. Controls with EDTA (100 µM) were done for the hairpinD-U8 and -U5 complexes. Effect of hairpin GC content on conformational distribution of Dis3L2 was evaluated for a -U7 overhang. X-axis: % of particles in vase (pink) or prong (blue) conformation, y-axis: individual datasets RNA-free or hairpin RNA-bound Dis3L2, numbers denote the length of the 3′ overhang. The deeper color indicates higher quality 3D-reconstructions, gray indicates particles that did not contribute to a meaningful reconstruction. d, Control processing pipeline excluding 2D classification. e, 3D maps from heterogeneous refinement of hairpinC-U12 dataset in CryoSPARC with the proportion of particles in a vase conformation indicated.
