public tools to distinguish germline variants from somatic ones in NGS data without matched normal

Question

[J.F.Jiang]

Hi all,

Distinguishing germline variants, especially those pathogenic ones, are of great importance, since these variants might increase the cancer susceptibility in other family members.

It is quite challenge when blood sample/matched normal is absence, leaving a lot of variants with zygosity/Allele frequency around 0.5.

Is there any public tool to determine the origin of the variants?

Thanks

Junfeng

[German.M.Demidov]

Dear Junfeng, I believe people simply exclude frequent variants from GnomAd. There is no reliable way to distinguish somatic and germline variants without the 2nd tissue sequenced.

[J.F.Jiang]

We indeed use gnomAD to annotate the VCF file. However, pathogenic variants are not so frequently. Even ClinVar and HGMD database are applied, it is very hard to determine those variants not existed in database.

[German.M.Demidov]

It seems for me that it is impossible to distinguish having a tumor tissue only. You can suspect a presence of a germline pathogenic variant if you have an unusual clinical presentation such a bilateral nature of tumor, early onset, family history, but otherwise you kind of need to sequence a second tissue. Even having a normal blood sequenced may be not of much help since blood also tends to accumulate age- and therapy-related somatic variants.

If you have a low tumor content (say, 60%) and see some mutation with allele frequency close to 0.5 and sufficient coverage - it is another good predictor for the germline origin.

[J.F.Jiang]

YES, it is very hard without the matched normal. Thanks for explaination.