Particle picking and centering in cryo-electron micrographs via a vanishing radial kernel

A training-free cryo-EM method finds particle centers by convolving micrographs with a vanishing radial kernel, enabling 3.68 Å and 4.30 Å reconstructions for Plasmodium falciparum 80S ribosome and β-galactosidase.

• Why it matters: Particle picking and centering are difficult because particles are weaker than noise and centering must work without assuming symmetric particles or relying on training.
• What they did: The approach performs one convolution with a compact radial kernel that increases with radius, saturates in an annulus, and vanishes beyond it; local response minima define centers, then grid seeds are updated, nearby endpoints merged, and 2D classification applied.
• The result: For the Plasmodium falciparum 80S ribosome and β-galactosidase data sets, reconstructions reach 3.68 Å and 4.30 Å (vs 3.77 Å and 4.36 Å with reference coordinates), with picking taking 0.20–0.26 s per micrograph.

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