The abstract is read at the publisher; the summary is JClub's.
Local assembly–based consensus strategies for human long-read SV calling show strong input-type dependence, with Sawfish’s strategy working well on HiFi and recent ONT but failing on older, noisier ONT reads.
- Why it matters: Overall SV-caller benchmarks hide which processing steps drive performance differences, limiting targeted improvements. Step-level evidence is needed to choose assembly strategies matched to read characteristics.
- What they did: Five local assembly consensus computation strategies, similar to those in popular SV callers, were implemented and compared on extensive simulated and real datasets. Performance was assessed across different read types and error profiles, including effects of read preprocessing.
- The result: Some strategies are more effective for particular input data types, while others are more robust overall despite changes in error profiles and preprocessing. These results provide guidance for developing improved future SV callers.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.