RobPicker: a meta-learning framework for robust identification of macromolecules in cryo-electron tomograms.

RobPicker, a meta-learning framework for cryo-ET particle picking, substantially outperforms state-of-the-art methods, especially for small and rare macromolecules under noisy labels and imbalanced species distributions.

• Why it matters: Deep learning for cryo-electron tomography can fail when training annotations contain label errors and when macromolecule species are unevenly represented.
• What they did: RobPicker automatically learns to correct label errors and increases emphasis on smaller and rarer macromolecule species during training and evaluation across diverse cryo-ET datasets.
• The result: Its robustness and efficiency improve 3D density maps of protein structures, enabling higher-resolution cellular structural biology analysis.

AI summary of the abstract · may contain errors · not medical advice · read the paper.