This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Merge vcf files of structural variants

Hi,

I have many single sample vcf files containing structural variants. I'm trying to merge all the vcfs to one multi-sample vcf, however as the breakpoints of structural variants are only estimates, the position of the same structural variant in different samples would be slightly different. Is there a way to merge individual vcfs containing structural variants into one multi sample vcf? taking account for example a percentage of overlap to call structural variants in different samples as the same variant.

Thanks

structural_variants vcf merge
• 8,332 views • link
updated by Qingxiang 2 • | written by ma17 2

3 answers

For that purpose I use SURVIVOR merge, which has a distance setting specifying the maximal difference between pairs of breakpoints (start1 vs start2 and end1 vs end2). As such it doesn't use a percentage of overlap, but it is also not requiring exact breakpoints.

About the percentage overlap the author of SURVIVOR said that he doesn't think it is correct that large SVs easier reach that minimal percentage, and therefore he just uses a distance. I believe he uses a distance of 1kb, but I commonly use 500bp.

• 1 views • link
written by WouterDeCoster 4.8K

Thanks! Exactly what I was looking for

• 1 views • link
written by ma17 2

When an overlap between vcf files A and B is found, which SV representation (i.e. which breakpoints) is reported by SURVIROR?

• 0 views • link
written by bi_Scholar 1

A newer option for this is OctopuSV. We developed it to merge SVs across samples while allowing for imprecise breakpoints.

It uses distance-based matching and also supports optional interval Jaccard filtering for DEL/DUP/INV, which may be useful if you want overlap to be part of the matching criteria.

For a multi-sample merge, for example:

octopusv merge -i sample1.svcf sample2.svcf sample3.svcf --mode sample --union -o cohort.svcf

• 0 views • link
written by Qingxiang 2

Please include a link to the software repository when mentioning a software package. There can be multiple packages with similar names that may do different things.

Looks like the repo is here: https://github.com/ylab-hi/OctopusV

• 0 views • link
written by GenoMax 16K

There is one recently published method that could address this question: SV-Pop https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-019-2718-4

• 0 views • link
written by Benbird 0

Log in to answer this question.

More posts like this