CryoPPP: A Large Expert-Labelled Cryo-EM Image Dataset for Machine Learning Protein Particle Picking

Abstract

Cryo-electron microscopy (cryo-EM) is currently the most powerful technique for determining the structures of large protein complexes and assemblies. Picking single-protein particles from cryo-EM micrographs (images) is a key step in reconstructing protein structures. However, the widely used template-based particle picking process is labor-intensive and time-consuming. Though the emerging machine learning-based particle picking can potentially automate the process, its development is severely hindered by lack of large, high-quality, manually labelled training data. Here, we present CryoPPP, a large, diverse, expert-curated cryo-EM image dataset for single protein particle picking and analysis to address this bottleneck. It consists of manually labelled cryo-EM micrographs of 32 non-redundant, representative protein datasets selected from the Electron Microscopy Public Image Archive (EMPIAR). It includes 9,089 diverse, high-resolution micrographs (~300 cryo-EM images per EMPIAR dataset) in which the coordinates of protein particles were labelled by human experts. The protein particle labelling process was rigorously validated by both 2D particle class validation and 3D density map validation with the gold standard. The dataset is expected to greatly facilitate the development of machine learning and artificial intelligence methods for automated cryo-EM protein particle picking. The dataset and data processing scripts are available at https://github.com/BioinfoMachineLearning/cryoppp

Accurate cryo-EM protein particle picking by integrating the foundational AI image segmentation model and specialized U-Net

“…After CryoSegNet was trained and validated on the training/validation, we blindly benchmarked it on a test dataset consisting of thousands of labeled cryo-EM micrographs of 7 different protein types from the CryoPPP 4 dataset. The particles picked by CryoSegNet were compared with the ground truth coordinates of the expert-labeled particles.…”

“…Thus, it is imperative to determine the protein structure for understanding protein function and interaction, studying their roles in the diseases, and accelerating the design of drugs. X-ray crystallography, nuclear magnetic resonance (NMR), and cryo-EM 4,5 are three main experimental techniques to determine protein structures. Among them, cryo-EM is the cutting-edge technique for solving the structure of large protein complexes.…”

“…With advancements in hardware and software tools [8][9][10][11][12] , numerous semi-automated or automated approaches varying from traditional computational methods to modern deep learning techniques have been proposed to streamline the cryo-EM processing and particle picking. Conventional computer vision methods like edge detection, blob detection and template matching 4 are still widely used for particle picking. However, due to the low SNR of cryo-EM micrographs, these techniques are susceptible to picking ice patches, carbon areas and aggregated particles, resulting in a high number of false positives.…”

“…Fig 4. Comparison results for viewing direction, resolution, and 3D density map of particles picked by crYOLO, Topaz and CryoSegNet.…”

Improving protein function prediction by learning and integrating representations of protein sequences and function labels

“…Predicting rare GO terms is analogous to the few-shot learning problems [6] in various domains like computer vision[7, 8, 9], and natural language processing(NLP). For example, in the classification task of named entity typing[10, 11] in NLP, assigning rare entity types to entity names pose a similar challenge, due to the increasing size and granularity of entity types.…”