Support for complex SVs

[d-cameron]

Are there any plans to support symbolic SVs, or SVs in breakend notation? To properly compare such variants requires haplotype reconstruction as equivalent representations can be quite different. For example, the CHM1 and CHM13 truth sets from Huddleston 2016 report duplication events as INS, whereas most short read SVs will report the same event as a DUP. In the case of tandem duplication expansions, the nominal variant positions can be non-overlapping (even hundreds of base pairs apart) but still result in the same haplotype thus are just different representations of the same variant.

[pkrusche]

We are working on a simple comparison engine for structural variants which works using overlap / distance-based comparisons, which we hope to make available soon. One part of this is that we worked out a few additions to the GA4GH intermediate formats here: https://github.com/ga4gh/benchmarking-tools/blob/master/doc/ref-impl/intermediate.md to be able to handle SVs, we can share this once the comparison engine is out.

If there are other tools out there that can do a haplotype-based comparison we could also integrate this by producing a GA4GH intermediate file as described above -- I am not aware of any tool that does this currently though -- making such a tool would certainly be very interesting, but isn't on our list of priorities in the near term.

Also, we will release a separate set of tools (later this week hopefully) to jointly genotype in multiple samples. I'll post a link here once this is out (see here for the abstract we presented at ASHG: https://ep70.eventpilotadmin.com/web/page.php?page=IntHtml&project=ASHG17&id=170122227 )

[pkrusche]

Just a quick update: we have released some tools to help with SV typing / validation: https://github.com/illumina/paragraph -- these tools don't implement SV assessment but they do include some information on variant validation. Also, we have released variant calls validated with this method here: https://github.com/Illumina/Polaris