Structural variant filtering

Question

[dominique.massau]

Hello everyone, I would appreciate some advice regarding the filtering of structural variants. I am planning to use a total of three variant callers for discovery (Delly, Manta, and Pindel) using whole-genome sequencing data (~20x coverage) across multiple samples.

I have been thinking about the best way to filter structural variants, but I am quite unsure what the best way is.

• Should I run the three SV callers on default settings and use some software for posterior filtering or is it advisable to already define more strict thresholds when using the SV callers? An issue here is that the three SV callers have different options for filtering (Pindel offering the most options while Delly has quite few). Also, is there any software that is a must-use in case of filtering the VCF files?
• What are attributes that are very important/effective to select for filtering? As for now I would like at least 2 out of 3 callers to support a SV, I will consider the amount of reads supporting an SV and I also take the quality and filter into consideration.

I am just getting started in the structural variant area, and while interesting, so as I rookie I could really use some advice from experienced users.

Thanks in advance!