cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination

Abstract

Single-particle electron cryomicroscopy (cryo-EM) is a powerful method for determining the structures of biological macromolecules. With automated microscopes, cryo-EM data can often be obtained in a few days. However, processing cryo-EM image data to reveal heterogeneity in the protein structure and to refine 3D maps to high resolution frequently becomes a severe bottleneck, requiring expert intervention, prior structural knowledge, and weeks of calculations on expensive computer clusters. Here we show that stochastic gradient descent (SGD) and branch-and-bound maximum likelihood optimization algorithms permit the major steps in cryo-EM structure determination to be performed in hours or minutes on an inexpensive desktop computer. Furthermore, SGD with Bayesian marginalization allows ab initio 3D classification, enabling automated analysis and discovery of unexpected structures without bias from a reference map. These algorithms are combined in a user-friendly computer program named cryoSPARC (http://www.cryosparc.com).

Structure of the insulin receptor–insulin complex by single-particle cryo-EM analysis

“…Particle picking was done using Gautomatch (http://www.mrc-lmb.cam.ac.uk/kzhang/) and resulted in 1,206,222 particles. All subsequent data processing (from 2D classification to final reconstruction) was done using Cryosparc 43 . After several cycles of 3D classification, 151,409 particles in one class were used to generate a 4.6 Å map with C2 symmetry (and a 5.6 Å map with C1 symmetry).…”

“…After a round of heterogeneous refinement, 147,436 particles were used to generate a 4.3 Å map with C2 symmetry (and a 4.7 Å map with C1 symmetry)(Extended Data Figure 1). The resolution of each reconstruction was determined by the gold standard Fourier shell correlation criterion in Cryosparc 43 and Relion PostProcess 44 . The completeness of the maps was assessed by measuring the Euler angle orientation distribution (as calculated by Cryosparc 43 ) and by evaluating each map 3D FSC 14 .…”

“…d , Plot of the global half-map FSC (solid red line) and spread of directional resolution values (+/− 1σ from mean, green dotted lines; the blue bars are a histogram of 100 such values evenly sampled over the 3D FSC 14 . e , Local resolution distribution (as calculated in Cryosparc 43 ). f , selected areas from the Class 1 C2 map at a resolution of 4.3 Å (map contoured at 12σ).…”

“… a , Gold standard FSC curve from Cryosparc 43 . b , Euler angle orientation distribution from Cryosparc 41 .…”

Hybrid Structure of the RagA/C-Ragulator mTORC1 Activation Complex

“…2D classes were selected based on high population and visual inspection. Ab-initio classification and reconstruction was performed on these particles in Cryosparc (Punjani et al, 2017) producing a reference volume e2 Molecular Cell 68, 835-846.e1-e3, December 7, 2017 from 24,061 particles which were used for further refinement. Refinement in Cryosparc converged on a map with a determined resolution of 16.2 A (gold-standard FSC 0.143 criterion).…”