Template-Based and Template-Free Approaches in Cellular Cryo-Electron Tomography Structural Pattern Mining

Abstract

Cryo-electron tomography (Cryo-ET) has made possible the observation of cellular organelles and macromolecular complexes at nanometer resolution in native conformations. Without disrupting the cell, Cryo-ET directly visualizes both known and unknown structures in situ and reveals their spatial and organizational relationships. Consequently, structural pattern mining (a.k.a. visual proteomics) needs to be performed to detect, identify and recover different sub-cellular components and their spatial organization in a systematic fashion for further biomedical analysis and interpretation. This chapter presents three major Cryo-ET structural pattern mining approaches to give an overview of traditional methods and recent advances in Cryo-ET data analysis. Template-based, supervised deep learning-based and template-free approaches are introduced in detail. Examples of recent biological and medical applications and future perspectives are provided.

Deep learning improves macromolecule identification in 3D cellular cryo-electron tomograms

“…CNNs have recently been investigated for learning high-level generic features in electron microscopy. Several algorithms based on deep learning techniques have been developed for 2D particle pick-1 ing in single-particle cryo-EM 19 , including DeepPicker 20 , AutoCry-oPicker 21 , crYOLO 22 , Topaz 23 , and Warp 24 . In cellular cryo-ET, the algorithm proposed by Chen et al, was implemented for supervised segmentation of 2D slices from 3D volumes 25 , but it was not designed to handle complex environments (for example, crowded cells) and requires an additional classification step to achieve satisfactory results.…”

Deep Learning Improves Macromolecule Identification in 3D Cellular Cryo-Electron Tomograms

“…CNNs have recently been investigated for learning high-level generic features in electron microscopy. Several algorithms based on deep learning techniques have been developed for 2D particle picking in single-particle cryo-EM 25 , including DeepPicker 26 , AutoCryoPicker 27 , crYOLO 28 , Topaz 29 , and Warp 30 . In cellular cryo-ET, the algorithm proposed by Chen et al ., was implemented for supervised 2D segmentation 31 , but it was not designed to handle complex environments (e.g., crowded cell) and requires an additional classification step to achieve satisfactory results.…”

Compressed sensing for electron cryotomography and high-resolution subtomogram averaging of biological specimens

“…Finding macromolecules in cells and tissues is notoriously difficult ( Melia and Bharat, 2018 ; Wang et al., 2011 ) and often requires additional experiments, such as immunolabeling or cryo-fluorescence microscopy. Increased SNR in tomograms may reduce the reliance on these additional steps, allowing more straightforward identification of target macromolecules by visual inspection of the density or by utilizing a template matching approach ( Frangakis et al., 2002 ; Wu et al., 2019 ). The revealed molecular detail, enabled through the increase in contrast while maintaining high-frequency resolution information, allows direct interpretation of the data without the need for additional filtration or denoising procedures.…”