Phelippeau, H. | bioRxiv Science

bioRxiv2024

DeepOrientation: Deep Orientation Estimation of Macromolecules in Cryo-electron tomography

Orientation estimation of macromolecules in cryo-electron tomography (cryo-ET) images is one of the fundamental steps in applying subtomogram averaging. The standard method in particle picking and orientation estimation is template matching (TM), which is computationally very expensive, with its performance depending linearly on the number of template orientations. In addition to conventional image processing methods like TM, the investigation of crowded cell environments using cryo-ET has also been attempted with deep learning (DL) methods. These attempts were restricted to macromolecule localization and identification while orientation estimation was not addressed due to a lack of a large enough dataset of ground truth annotations suitable for DL. To this end, we first generate a large-scale synthetic dataset of 450 tomograms containing almost 200K samples of two macromolecular structures using the PolNet simulator. Utilizing this synthetic dataset, we address the problem of particle orientation estimation as a regression problem by proposing a DL-based model based on multi-layer perceptron networks and a six-degree-of-freedom orientation representation. The iso-surface visualizations of the averaged subtomograms show that the predicted results by the network are overly similar to that of ground truth. Our work shows that orientation estimation of particles using DL methods is in principle possible provided that ground truth data is available. What remains to be solved is the gap between synthetic and experimental data. The source code is available at https://github.com/noushinha/DeepOrientation.

molecular biology↗

bioRxiv2023

Simulating the cellular context in synthetic datasets for cryo-electron tomography

Cryo-electron tomography (cryo-ET) allows to visualize the cellular context at macromolecular level. To date, the impossibility of obtaining a reliable ground truth is limiting the application of deep learning-based image processing algorithms in this field. As a consequence, there is a growing demand of realistic synthetic datasets for training deep learning algorithms. In addition, besides assisting the acquisition and interpretation of experimental data, synthetic tomograms are used as reference models for cellular organization analysis from cellular tomograms. Current simulators in cryo-ET focus on reproducing distortions from image acquisition and tomogram reconstruction, however, they can not generate many of the low order features present in cellular tomograms. Here we propose several geometric and organization models to simulate low order cellular structures imaged by cryo-ET. Specifically, clusters of any known cytosolic or membrane bound macromolecules, membranes with different geometries as well as different filamentous structures such as microtubules or actin-like networks. Moreover, we use parametrizable stochastic models to generate a high diversity of geometries and organizations to simulate representative and generalized datasets, including very crowded environments like those observed in native cells. These models have been implemented in a multiplatform open-source Python package, including scripts to generate cryo-tomograms with adjustable sizes and resolutions. In addition, these scripts provide also distortion-free density maps besides the ground truth in different file formats for efficient access and advanced visualization. We show that such a realistic synthetic dataset can be readily used to train generalizable deep learning algorithms.

bioinformatics↗