Low-pass filtering revisited: ramped phase randomization for cryo-EM map comparison and data augmentation

Ramped phase randomization for cryo-EM map comparison reduces higher-resolution fine detail while preserving FSC behavior, matching lower-resolution independently determined maps across three published map pairs.

• Why it matters: Conventional low-pass filtering can leave original fine-scale detail beyond the target resolution or, with sharp cutoffs, introduce ringing artifacts that complicate fair map comparison.
• What they did: The method extends low-pass filtering by adding phase offsets that ramp smoothly with spatial frequency, with phase levels set using correlation to an ideal noiseless target-resolution map and a correction for finite input resolution.
• The result: Across three published map pairs, filtered maps’ fine detail is reduced to levels similar to independently determined lower-resolution maps, and FSCs closely resemble those of experimental maps, enabling both comparison and neural-network data augmentation.

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