Protein secondary structure detection in intermediate-resolution cryo-EM maps using deep learning

Abstract

An increasing number of protein structures have been solved by cryo-electron microscopy (cryo-EM). Although structures determined at near-atomic resolution are now routinely reported, many density maps are still determined at an intermediate resolution, where extracting structure information is still a challenge. We have developed a computational method, Emap2sec, which identifies the secondary structures of proteins (α helices, β sheets, and other structures) in an EM map of 5 to 10 Å resolution. Emap2sec uses a 3D deep convolutional neural network to assign secondary structure to each grid point in an EM map. We tested Emap2sec on 6.0 and 10.0 Å resolution EM maps simulated from 34 structures, as well as on 43 maps determined experimentally at 5.0 to 9.5 Å resolution. Emap2sec was able to clearly identify the secondary structures in many maps tested, and showed substantially better performance than existing methods.

Haruspex: A Neural Network for the Automatic Identification of Oligonucleotides and Protein Secondary Structure in Cryo‐Electron Microscopy Maps

“…Forschungsartikel tal maps with aw ell-matching model for training and testing such an etwork are difficult to obtain. This obstacle has previously been faced by Si et al [33] (SSELearner), Li et al [23] and Subramaniya et al [25] (Emap2Sec) who developed machine learning approaches for protein secondary structure prediction in cryo-EM maps,but not oligonucleotides, [23] and consequently resorted partly to simulated maps generated with pdb2mrc. [34] These simulated maps lack the error structure and processing artefacts found in experimentally derived reconstruction densities, [4][5][6] as they assume aperfectly processed data set of ah omogenous sample where all atoms interact with the electron beam as if they were uncharged and unbound.…”

Extraction of Protein Dynamics Information Hidden in Cryogenic Electron Microscopy Maps Using Deep Learning

“…We utilized one of the deep-learning methods, 3D-CNN [10][11][12] , which is widely used to detect or classify threedimensional objects constructed from several resources, such as a video data [12][13][14] and magnetic resonance imaging [15][16][17] . Moreover, 3D-CNN has been shown to exhibit remarkable performance on the three-dimensional cryo-EM maps to recognize several structural patterns, such as secondary structures, amino acids, and the local map resolutions [18][19][20][21] .…”

Advances in Structure Modeling Methods for Cryo-Electron Microscopy Maps

“…At this range of the resolution, some fragments of secondary structure elements (SSE), α-helices and β-sheets are barely visible, but ML can significantly improve identification. RENNSH is a method which identifies α-helices in a density map by applying nested K-nearest neighbors (KNN) classifiers with spherical harmonic descriptors [101,102]. SSELearner, uses another classification method, support vector machines (SVM), to identify both α-helices and β-sheets in EM maps [103].…”

“…Our group has developed Emap2sec, a deep learning-based method, which uses 3-D CNN for detecting secondary structures of a protein (α-helix, β-sheets, and other structures) in cryo-EM maps of 5 to 10 Å [101]. Emap2sec first scans a cryo-EM map with a voxel of size 11 Å. Emap2sec consists of a two-phase stacked network architecture.…”

“…(b) Eschirichia coli replicative DNA polymerase complex (EMD-3201 of resolution 8.34 Å; PDB 5FKU). This figure was adapted from the EMap2sec paper[101].…”