search

Illumina / canvas

Canvas - Copy number variant (CNV) calling from DNA sequencing data
Other
121 stars 20 forks source link

Get purity estimate in quiet sample #14

Closed RichardCorbett closed 8 years ago

RichardCorbett commented 8 years ago

Hi again,

I'm doing some test regarding the ploidy and tumour purity estimation. In the design document I see this:

If we did not assign many CNVs (.....), then this SNV-estimated purity is what gets reported to the purity output VCF file

and I see this in my SomaticCNV-a.stdout.txt file

Loaded 3423 somatic variants; saved 3368 somatic SNV frequencies Estimated tumor purity of 0.410252243280411 from somatic SNV calls Purity estimates: 1.0000 from CNVs, 0.4103 from SNVs Fraction abnormal CNV is 0.0407

but my VCF file says

EstimatedTumorPurity=1.00

Am I missing something or should my tumour purity estimate be 0.4103 in the VCF file?

sivakhno commented 8 years ago

Richard Hi, Purity estimated from somatic SNV is used when purity estimate from CNVs is not available. In you case "Purity estimates: 1.0000 from CNVs" was reported, so while Canvas also reports purity estimate from SNVs - this is not used in VCF output. Reason is that SNV data is less accurate as HET / HOM ALT discrimination is hard leading to purity overestimation, hence only CNV-derived purity is used as a default.