Can Breakdancer be applied on Paired end Exome Sequencing data for germline copy number/sv detection?

Question

[gsr9999]

Dear Biostar Leaders,

We have a Paired End Sequencing Reads from Whole Exome Sequencing run generated on a cell line that is known to contain 15kb deletion on chrX. This 15kb region includes part of exon 4 of MECP2 gene. I have run Pindel and DELLY tools on it and this 15kb deletion is successfully detected by both the tools.

I am under the impression that Breakdancer can be applied only on Whole Genome Sequencing data, and not the Whole Exome data (WES). Please correct me if I am right or wrong.

Is it wise to run Breakdancer on a paired end WES data ?

Thanks, gsr9999

Answer 1

[d-cameron]

Is it wise to run Breakdancer

It is not wise to run Breakdancer unless you have 2x50bp reads or shorter. Even for WGS, breakdancer is a poor choice of SV caller and is outperformed by most alternative tools. It's designed for very short reads (2x36bp) and does not perform well with longer (>=100bp) reads. Methods papers love comparing to and citing BreakDancer in part because it's so easy to do better than in.

I have run Pindel and DELLY tools on it and this 15kb deletion is successfully detected by both the tools

BreakDancer would probably detect this variant, although I question why you would want to use BreakDancer given you could instead use lumpy, manta, or gridss.

I learnt that CNV detection in Exome data by SV detection tools that are based on Discordant Read Pairs like Breakdancer may not be able to detect CNVs in every case

Fundamentally, split read, discordant read pair, and assembly-based callers are all breakpoint callers and are unable to detect SVs whose breakpoints fall outside the targeted region (exome). There are a number of specialised WES copy number based tools

[gsr9999]

Hi Cameron,

Thank you for the details/information on how breakdancer is best suited for short read sequencing <2*50 bp.

Thank u for pointing me to other tools like lumpy, manta and gridss.

Answer 2

[gsr9999]

After reading review papers, I learnt that CNV detection in Exome data by SV detection tools that are based on Discordant Read Pairs like Breakdancer may not be able to detect CNVs in every case. These tools require at least one breakpoint to be in exonic regions, and when real breakpoint is outside of the exonic region, tools like Breakdancer may not detect them.

It appears to me the read depth based tools, like CoNIFER or XHMM, are appropriate for CNV detection in WES data.