PURPLE germline CNV calls

[silkrp]

Hi, I’m running PURPLE as part of the oncoanalyser pipeline in tumour–normal mode and would like to examine the germline CNVs detected in the normal sample.

The output includes the *.purple.cnv.somatic.tsv file, but there doesn’t appear to be an equivalent *.purple.cnv.germline.tsv file. Is there an existing output or recommended method to identify the CNVs present in the normal sample?

Thanks,
Ryan

[p-priestley]

PURPLE is primarily a somatic variant caller although germline copy number is evaluated internally. We do write a file 'TUMOR.purple.germline.deletion.tsv' which presents a list of genes impacted heterozygous and homozygous deletions (relevant to reporting), but not a complete germline copy number profile

The 'TUMOR.purple.segment.tsv' does segment the whole genome. Each region is marked with a germlineStatus (DIPLOID, HET_DELETION, HOM_DELETION, AMPLIFICATION or NOISE). Note that only regions of more than 1kb will be detected. The 'observedNormalRatio' column also the relative germline copy number adjusted for GC. Not sure if these might be helpful

[silkrp]

Thanks for the quick response. Would it be unwise to treat the normal as the tumour sample when run in tumour-only mode to get germline CNV calls?

[DarioS]

To get the best results, you would want GRIDSS-PURPLE, which utilises split reads and discordant pairs.