search

AstraZeneca-NGS / VarDict

VarDict
MIT License
187 stars 62 forks source link

I have true sets, which vardict call high FP in tumor-only #124

Closed wentgithub closed 5 years ago

wentgithub commented 5 years ago

thanks for making this software calling the complex variant, this super all others.

but I found another problem, for tumor-only, even for paired, has high FP. these days I also test a paired standard sample.

you know like gatk, it can supply a germline resource to filter gemline vatrianrts, can vardict also use this?

PolinaBevad commented 5 years ago

Hi @wentgithub,

Sorry, but we don't have such functionality as you mention for GATK.

For this tumor-only case you have to process VarDict VCF results in some downstream software or make a script to filter germline variants: if you have VCF file from germline DB (dbSNP, for example) then go through variants from tumor VCF and check if they are in dbSNP.

VarDict is a very sensitive variant caller, but you can "tune" the parameters to decrease FP rate: filter low mapping quality reads by -Q 10, remove QCfail reads with -F 0x700 and so on. It can be done simpler by using VarDict in bcbio pipeline (https://github.com/bcbio/bcbio-nextgen), all the tunings are already included there. Please, take a look at it!