Nice to chat at CSHL last week. I missed the chance to tell you in person what a valuable resource these videos have been, I think I've seen every one. Even for well-known topics, the clarity of presentation can really sharpen intuition. Thanks for the great work.

In addition to CNV recall and breakpoint accuracy improvements, this update boosts precision for large, unbalanced SVs and simplifies analysis by harmonizing all depth and breakpoint evidence into one consistent call set. The full update is now available in the most recent sawfish2 release on github

This example is hard to understand from e.g. IGV/Ribbon (see Fig1) but pretty simple in SVTopo: 4 blocks deleted (B,D,F,H), 2 inverted (E,G), 1 re-ordered (C)

Finally, a huge thank-you to all sawfish co-authors and the broader @pacbio.bsky.social team, our anonymous reviewers, SV community members, and users for their contributions, feedback and many helpful suggestions!

Also worth noting that a substantial new sawfish CNV integration feature will be coming as a preview release on GitHub later this month, which I’ll also be detailing as a poster presentation at ESHG. Looking forward to sharing more about this soon.

Since the preprint version of sawfish itself, we’ve released the full source code and now output all assembled contig alignments underlying the SV calls, in addition to several other stability and usability improvements. Sawfish binaries, documentation and source can be found on GItHub:

Since the preprint we’ve added several analyses including comparison to SV calling from global assembly, direct assessment of SV haplotype quality and SV context analysis, among others. To improve clarity, all SV calling and assessment results are now provided on Zenodo: