Cryo-electron tomography (cryo-ET) is a powerful tool for visualizing macromolecular complexes in their native conformations at sub-nanometer resolutions and for revealing their spatial and organizational relationships. Particle picking, i.e., locating recognition, is a crucial step in this technological process. However, the application of existing automatic picking methods is limited by various factors, such as the large amount of manual annotation, high computational cost, and poor particle quality.

Researchers from the Institute of Biophysics (IBP) of the Chinese Academy of Sciences (CAS), in collaboration with scientists from the Institute of Automation of CAS, have used artificial intelligence technology to empower in-situ structural biology and proposed a fast and accurate particle picking method based on weakly supervised deep learning, named DeepETPicker.

The study was published in Nature Communications.

According to the researchers, DeepETPicker requires only a small amount of manual annotation of particles for training to achieve fast and accurate three-dimensional automatic particle picking. It optimally simplifies labels to replace real labels and employs a more efficient model architecture, richer data augmentation techniques, and overlap partitioning strategies to improve model performance with small training sets.

To improve particle localization speed, DeepETPicker employs GPU-accelerated average pooling and non-maximum suppression post-processing operations, achieving a several-fold increase in picking speed compared to existing clustering post-processing methods.

In a comprehensive performance evaluation of particle picking quality based on six quantitative metrics, DeepETPicker can achieve fast and accurate particle picking on both simulated and real datasets. Its overall performance is significantly better than existing methods, and the resolution of the structure reconstruction of biomacromolecules reaches the level of manual expert particle picking for structure reconstruction, further demonstrating the practical value of DeepETPicker in in-situ high-resolution structural analysis.

Currently, in order to facilitate users, the research team has released open-source software with simple operation and user-friendly interface to assist users in performing image preprocessing, particle annotation, model training, and inference operations.

The workflow for picking particles from cryo-electron tomograms using DeepETPicker. (Image by IBP)