VCF output of speedseq(lumpy) SVtype BND are translocations? about svtools HOT 5 CLOSED

BND is a catch-all for a generic breakpoint, so you can't assume all to be translocations

When LUMPY reports a BND with both sides on the same chromsome, it is indicating misoriented reads (++ or --). When we see reciprocal evidence (both ++ and -- from the same event), LUMPY calls this an inversion. Howevere, when only one of these orientations are observed, it is designated a BND. Some of these "one-sided inversion" BNDs might be true inversions where we miss one of the sides, and others may be parts of complex variants or artifacts.

from svtools.

In that example of a simple deletion the chromosome name is chr and the numbers are positions. Perhaps not the best choice, but it is intended to show BND entries on the same chromosome. Sorry for the confusion.

In terms of Lumpy output, I am not as confident as to the correct answer. I do think you are correct that if both breakends are on the same chromosome then the interpretation is that Lumpy was unable to determine that it belonged to any of the other classes (DEL, INV, DUP see https://github.com/arq5x/lumpy-sv/blob/7d60167d17347ceadb297e9a66c36df8013d1b83/src/lumpy/SV_VcfVariant.cpp#L110-L135). I would defer to @ryanlayer or @cc2qe on that though as I'm just guess from the source.

from svtools.

Thanks for clarifying this. So, a BND with both sides on the same chromosome can be an INV. Can it be a DUP or DEL?

from svtools.

LUMPY never represents breakpoints with DEL or DUP orientations as BNDs. Only one sided inversions and interchromosomal events

from svtools.

Great to know this!

Ming

from svtools.