CNV detection using beakdancer and SVDetect

Question

[kk.mahsa]

hi

i have whole genome of a mmalian in paired end aligned on reference genome using BWA.

i used breakdancer to CNV detection but i do not know how interpret the results. where are CNVs in output??

also i used SVDetect cnv but it is hard to interpret too.

manual of programs are not rich.

anybody can help me?

[Samarth Kulshrestha]

Hi Mahsa,

Can you please provide more details about the output generated by those SV callers? Generally these callers reports different types of Structural Variants (SV) sites in the genome. There are variety of CNV callers are available for CNV detection,you can look for Control-FREEC, BICseq.

Hope this helps.

Answer 1

[d-cameron]

BreakDancer and SVDetect do not use read depth to call copy number changes. They use discordant read pairs to detect breakpoints and call structural variants (SVs) from that.

BreakDancer and SVDetect are also quite old and there are better programs using more sophisticated algorithms that have much better performance. I strongly recommend you do not use BreakDancer. It is designed from 2x36bp reads and is outperformed by every SV tool I have compared it against (10+) on data from current sequencers (e.g. 2x100bp). Longer read lengths actually make breakdancer's performance worse! My benchmarking of SV detection in Illumina short read sequence data leads me to recommend GRIDSS^ or manta. Both GRIDSS and manta report variants using the VCF file format.

If you want to call copy number changes, you should use a copy number caller. There are many of these available include sequenza, ABSOLUTE, absCN-seq, FACETS, and many others.

^ Disclaimer: I am the author GRIDSS.