The evolution of SPHIRE-crYOLO particle picking and its application in automated cryo-EM processing workflows

Abstract

Particle selection is a crucial step when processing electron cryo microscopy data. Several automated particle picking procedures were developed in the past but most struggle with non-ideal data sets. In our recent Communications Biology article, we presented crYOLO, a deep learning based particle picking program. It enables fast, automated particle picking at human levels of accuracy with low effort. A general model allows the use of crYOLO for selecting particles in previously unseen data sets without further training. Here we describe how crYOLO has evolved since its initial release. We have introduced filament picking, a new denoising technique, and a new graphical user interface. Moreover, we outline its usage in automated processing pipelines, which is an important advancement on the horizon of the field.The crYOLO particle picking procedure A major goal of electron cryo microscopy (cryo-EM) is to obtain high-resolution threedimensional (3D) reconstructions of proteins and protein complexes to gain novel biological insights. This process involves the selection of thousands to millions of noisy two-dimensional (2D) particle projections, a number that only keeps increasing with recent advances in hardware and software development.In our recent work in Communications Biology 1 we introduced the "crYOLO" particle picking procedure. It is based on a deep neural network and the You Only Look Once (YOLO) object detection framework 2 . This approach enables the automated picking of particles within cryo-EM micrographs with a low signal-to-noise ratio requiring minimal human supervision or intervention. CrYOLO is easy to configure and train on a specific data set. It is fast and can process up to six micrographs per second. As crYOLO sees the complete micrograph, it is able to learn the

Atomic structures of respiratory complex III2, complex IV, and supercomplex III2-IV from vascular plants

“…After motion correction and CTF correction, a set of 8541 micrographs was selected for further processing. Automated particle picking using crYOLO ( Wagner et al, 2019 ; Wagner and Raunser, 2020 ) resulted in ~1.5 million particles. The particles were extracted using 400 2 pixel box binned two-fold and sorted by reference-free 2D classification in Relion using (--max_sig 5), followed by re-extraction at 512 2 pixel box.…”

Atomic structure of a mitochondrial complex I intermediate from vascular plants

“…Contrast transfer function (CTF) parameters of the corrected micrographs were estimated using Ctffind4 ( Rohou and Grigorieff, 2015 ) and refined locally for each particle in RELION. Automated particle picking using crYOLO ( Wagner et al, 2019 ; Wagner and Raunser, 2020 ) resulted in ~1.5 million particles. The particles were extracted using 400 2 pixel box binned two-fold and sorted by reference-free 2D classification followed by re-extraction at 512 2 pixel box.…”

Atomic structures of respiratory complex III2, complex IV and supercomplex III2-IV from vascular plants

“…After motion correction and CTF correction, a set of 8,541 micrographs was selected for further processing. Automated particle picking using crYOLO [96,97] resulted in ~1.5 million particles. The particles were extracted using 400 2 pixel box binned two-fold and sorted by reference-free 2D classification in Relion using (--max_sig 5), followed by re-extraction at 512 2 pixel box.…”