The abstract is read at the publisher; the summary is JClub's.
A session-aware evaluation shows AFM knot-type classification drops to 27–32% balanced accuracy (near 25% chance) despite 58–67% on training sessions across 20 imaging sessions.
- Why it matters: Automated DNA knot identification in AFM is unreliable because models can exploit session-specific cues, and dense crossings limit accurate strand pairing and crossing labeling.
- What they did: They built a session-holdout protocol (scoring four size-indistinguishable classes) and tested classifiers plus a session probe; they also introduced an energy-based strand pairing that traced 84% of knots/catenanes correctly.
- The result: Crossing-resolution tests found correct upper-strand labeling at 173 crossings without manual annotation and 79% correct ordering by the width rule, but knot-type accuracy remained 6–8% due to 72% losing crossings when merged.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.