Two particle-picking procedures for filamentous proteins: SPHIRE-crYOLO filament mode and SPHIRE-STRIPER

Abstract

<jats:p>Structure determination of filamentous molecular complexes involves the selection of filaments from cryo-EM micrographs. The automatic selection of helical specimens is particularly difficult, and thus many challenging samples with issues such as contamination or aggregation are still manually picked. Here, two approaches for selecting filamentous complexes are presented: one uses a trained deep neural network to identify the filaments and is integrated in <jats:italic>SPHIRE-crYOLO</jats:italic>, while the other, called <jats:italic>SPHIRE-STRIPER</jats:italic>, is based on a classical line-detection approach. The advantage of the <jats:italic>crYOLO</jats:italic>-based procedure is that it performs accurately on very challenging data sets and selects filaments with high accuracy. Although <jats:italic>STRIPER</jats:italic> is less precise, the user benefits from less intervention, since in contrast to <jats:italic>crYOLO</jats:italic>, <jats:italic>STRIPER</jats:italic> does not require training. The performance of both procedures on Tobacco mosaic virus and filamentous F-actin data sets is described to demonstrate the robustness of each method.</jats:p>

High-resolution structures of the actomyosin-V complex in three nucleotide states provide insights into the force generation mechanism

“…Data sets were automatically pre-processed on-the-fly during the data acquisition using TranSPHIRE (Stabrin, Schoenfeld, Wagner, Pospich, Gatsogiannis, & Raunser, 2020a). Pre-processing included drift correction and dose weighting by MotionCor2 (Zheng et al, 2017), CTF estimation using GCTF (Zhang, 2016) and particle picking with crYOLO (Wagner et al, 2020; 2019) (filament mode, box distance 26-27 px equivalent to one rise of 27.5 Å, minimum number of boxes 6) for all data sets. The latest version of TranSPHIRE (Stabrin, Schoenfeld, Wagner, Pospich, Gatsogiannis, & Raunser, 2020b), which was used for the processing of the AppNHp data sets, also supported automatic, on-the-fly particle extraction (box size 320 px, filament width 200 px) as well as batch-wise 2D classification (batch size 13k, filament width 200 px, radius 150 px, 60-100 particles per class), 2D class selection and 3D refinement using software of the SPHIRE package (Moriya et al, 2017).…”

PspA adopts an ESCRT-III-like fold and remodels bacterial membranes

“…A subset of the resulting micrographs was subjected to manual filament tracing using E2HELIXBOXER (Ludtke, 2016). Manually traced filaments were used for neural network training and subsequent automated tracing with crYOLO 1.5 (Wagner et al, 2019(Wagner et al, , 2020. This way, a total of 55,800 rods were detected.…”

TranSPHIRE: automated and feedback-optimized on-the-fly processing for cryo-EM

“…While the actual processing is fully-automated, some preparation is still needed when using the TranSPHIRE pipeline to process filaments. Specifically, crYOLO needs to be trained to pick filaments 34 , as these look fundamentally different from the single particle complexes known to its default general model. Additionally, Cinderella 25 also needs to be trained with 2D class averages of the filament in question.…”