Morning! Let's start the day! SVDSS: structural variation discovery in hard-to-call genomic regions using sample-specific strings from accurate long reads by Luca Denti. #RECOMB2023

Talk from @BonizzoniPaola starting the #ShonanMeetingPangenomics

SVDSS combines multiple strategies to improve detection of structural variation using low-error long-read sequencing data. @l_denti @ParsoaKhorsand @DavisCompGen @BonizzoniPaola @PangaiaProject @RayanChikhi https://t.co/EY2znczRBb

Code is freely available at

For now SVDSS can be used to discover insertions and deletions. We have plans to extend its functionalities to other SVs, such as inversions and translocations. Other extensions include somatic SVs discovery and SVs genotyping.

Analysis of SVs inside medically-relevant genes shows that SVDSS outperforms state-of-the-art tools in calling heterozygous SVs.

Even on CHM13 HiFi reads (provided by the T2T consortium, https://t.co/DGUyFwMYqZ), SVDSS achieves superior accuracy. But the homozygous nature of CHM13 makes SV calling relatively easier.

SVDSS achieves overall superior SV discovery performance outside Tier 1 regions of the genome and even at lower coverage (10x).

While most approaches focus on HG002 GIAB Tier 1 callset, we built our own truthsets using dipcall ( https://t.co/LLXhMBOF2T) and starting from HG002/HG007 assemblies. Truthsets are publicly available at

SVDSS achieves significant improvements in calling SVs in repetitive and hard-to-call regions of the genome.

SVDSS finds potential SV sites using SFS and performs local POA of clusters of SFS to produce accurate SV predictions.

SVDSS is based on our recent notion of sample-specific strings (SFS, from https://t.co/faGaEMJon8) and effectively leverages alignment-free, mapping-based, and assembly-based methodologies.

Our work on Structural Variations discovery from accurate long reads (@PacBio HiFi) is out ( https://t.co/EmfM2xunJo). Joint work with @l_denti @ParsoaKhorsand @BonizzoniPaola @RayanChikhi @DavisCompGen

Pangenome Graph Construction from Genome Alignment with Minigraph-Cactus https://t.co/SIukrRJpTT #biorxiv_bioinfo

In this new Bioinformatics Application Note, we introduce a file format named KFF for efficiently storing kmers on disk. https://t.co/20gTzasFuD

First steps in #pangenomics? Check out our new tutorial! https://t.co/VlOzRfuZzl Thanks @jasmijnbaaijens Paola Bonizzoni Christina Boucher @giandellavedova Raffaella Rizzi @jltsiren @PangaiaProject

Code is freely available at

SFS strings are more expensive to compute than k-mers. In the paper we introduce an heuristic algorithm that finds non-overlapping SFS strings efficiently.

There are intrinsic similarities between SFS strings and other stringology concepts such as maximal exact matches, however we have not yet fully elucidated them.

This makes sense because SFS strings can better capture variations in repeated regions where k-mers would be confused by many identical instances.