Calling CNVs without a normal

Question

[vctrm67]

Hi all,

Is there a way to call CNVs without a matched normal? I have a tumor sample and a panel of normals and a reference, but no matched normal. I understand that the CNVs won't be somatic but is there a way to just get CNVs with respect to a reference?

[Pierre Lindenbaum]

try manta, lumpysv, etc...

[vctrm67]

These seem like they're for SV analysis, not CNV?

[Pierre Lindenbaum]

CNV= DEL= SV CNV= DUP/INS = SV

[German.M.Demidov]

idk, for me calling CNVs in cancer automatically mean "clonal decomposition" also - and SV callers dont do this (as well as inference of actual copy-numbers), also SV callers struggle with non-tandem duplications

however, without matching normal this task become close to impossible (without B-allele frequencies mainly) - it is not realistic to decompose CNVs into 2 alleles with different copy-numbers and estimate clonal cell fraction for each event...

Answer 1

[jared.andrews07]

CNVkit works well with a reference sample (or pool of normals to create one from). Yes, they aren't somatic, but it does what you want. Also very easy to use.