Best approach to quantify large-scale structural variants (INV/DUP/TRA) from a Minigraph-Cactus pangenome

[Sumit1331]

Hi all,

I built a pangenome graph with Minigraph-Cactus for a haploid fungal species and I’m trying to quantify large-scale SVs (inversions, duplications, translocations).

What’s the recommended approach to get counts of these SV classes from a Minigraph-Cactus graph?

-> Should I be aiming to generate a VCF with SVTYPE (INV/DUP/TRA) (e.g., using vg giraffe → vg pack → vg call) and count by SVTYPE?

-> Or is there a more “graph-native” way to summarize/classify these events directly from the graph (e.g., snarls/bubbles), especially for TRA/DUP?

-> If translocations are represented as BND rather than SVTYPE=TRA, what’s the best practice for counting them consistently?

Any pointers to recommended tools/commands/papers for this specific setup would be appreciated.

Thank you for your support.

[adamnovak]

This isn't a bug exactly (and might want to be moved over to the minigraph/cactus repo?), but @glennhickey I think has a recommended pipeline for using vg deconstruct to inventory the possible structural variants that a graph describes.

But I think that only handles insertions/deletions/inversions; I'm not sure that any tools exist that can recognize that a graph motif might be well-described as a translocation.