DeepHEMNMA: ResNet-based hybrid analysis of continuous conformational heterogeneity in cryo-EM single particle images

Abstract

Single-particle cryo-electron microscopy (cryo-EM) is a technique for biomolecular structure reconstruction from vitrified samples containing many copies of a biomolecular complex (known as single particles) at random unknown 3D orientations and positions. Cryo-EM allows reconstructing multiple conformations of the complexes from images of the same sample, which usually requires many rounds of 2D and 3D classifications to disentangle and interpret the combined conformational, orientational, and translational heterogeneity. The elucidation of different conformations is the key to understand molecular mechanisms behind the biological functions of the complexes and the key to novel drug discovery. Continuous conformational heterogeneity, due to gradual conformational transitions giving raise to many intermediate conformational states of the complexes, is both an obstacle for high-resolution 3D reconstruction of the conformational states and an opportunity to obtain information about multiple coexisting conformational states at once. HEMNMA method, specifically developed for analyzing continuous conformational heterogeneity in cryo-EM, determines the conformation, orientation, and position of the complex in each single particle image by image analysis using normal modes (the motion directions simulated for a given atomic structure or EM map), which in turn allows determining the full conformational space of the complex but at the price of high computational cost. In this article, we present a new method, referred to as DeepHEMNMA, which speeds up HEMNMA by combining it with a residual neural network (ResNet) based deep learning approach. The performance of DeepHEMNMA is shown using synthetic and experimental single particle images.

ContinuousFlex: Software package for analyzing continuous conformational variability of macromolecules in cryo electron microscopy and tomography data

“…Then, the results from HEMNMA (on the training set) and DeepHEMNMA (on the inference set) are combined and analyzed in terms of animations and 3D reconstructions in the low-dimensional conformational space, as when using HEMNMA alone. HEMNMA and DeepHEMNMA give very similar results but the estimated total number of computing hours needed by DeepHEMNMA for obtaining normal-mode amplitudes, angles, and shifts is several tens of times smaller compared to the number of computing hours needed by HEMNMA (e.g., of the order of 40 times for 1 million synthetic images of size 128 × 128 pixels and 3 normal modes, (Hamitouche and Jonic, 2022)).…”

“…This is important for small complexes, such as Pol α ̶ B complex, but also for large complexes, such as TBSV virus (Jin et al, 2014). Yet, in some cases, localized motions of smaller parts of complexes can be detected, like the motion of bound to 80S ribosome (article in press (Hamitouche and Jonic, 2022)). TomoFlow (not based on normal modes but on optical flow) involves a large degree of data smoothing, which may hinder the detection of motions of very small parts of complexes.…”

“…Nevertheless, the computational demand is overwhelming for cryo-EM single particle datasets containing hundreds of thousands to millions of images nowadays. Therefore, we have recently combined HEMNMA with deep learning to accelerate it and this method is referred to as DeepHEMNMA (Hamitouche and Jonić, 2021;Hamitouche and Jonic, 2022). The deep learning neural network of DeepHEMNMA is based on ResNet-34, which is a residual convolutional neural network with 34 layers (He et al, 2016), trained on HEMNMA estimations of rigid-body and elastic alignment parameters for single particle images (angles, shifts, and normal-mode amplitudes) on a small subset (training set) of an input single particle dataset.…”

“…The software of other methods has been developed and added to ContinuousFlex later. Besides, ContinuousFlex currently also offers a deep learning extension of HEMNMA, named DeepHEMNMA, the software of a very recent method (article in press (Hamitouche and Jonic, 2022)). In this…”

MDSPACE: Extracting Continuous Conformational Landscapes from Cryo-EM Single Particle Datasets Using 3D-to-2D Flexible Fitting based on Molecular Dynamics Simulation

“…Continuous conformational variability is a term introduced about ten years ago and currently largely used in cryo-EM to refer to the concept according to which biomolecular complexes can be considered to follow continuous trajectories of conformational changes and the conformational states contained in the images are samples on these trajectories [1][2][3][4][5]. This concept has been the basis for the development of methods to obtain the conformational landscapes of the complexes in vitro, by single particle analysis [1,2,[6][7][8][9][10][11][12][13][14][15]. Similar methods are currently also being developed for analyzing continuous conformational variability of the complexes in situ, by cryo electron tomography [16,17].…”