Confusion with Varscan somatic filters

Question

[mkaushal]

Hello,

I am running Varscan somatic to identify somatic mutations in normal tumor samples. These are the filters I am using:

--min-coverage 25
--min-avg-qual 30
--min-var-freq 0.05
--somatic-p-value 0.05
--normal-purity 1
--tumor-purity 0.5
--min-reads2 3

Please note though I have used --min-var-freq of 0.05 , I am still getting variants with vaf < 5%. Example:

chr1    762273  .       G       A       .       PASS    DP=284;SOMATIC;SS=2;SSC=13;GPV=1E0;SPV=4.0376E-2        GT:GQ:DP:RD:AD:FREQ:DP4 0/0:.:156:152:0:0%:37,115,0,0   0/1:.:128:121:4:**3.2%**:27,94,1,3

Can someone please help me understand what I am missing here. Is this because of some other filter I am using? Or did I get the usage of --min-var-freq wrong?

Thanks,
Madhurima

Answer 1

[sergio.lois]

Hi Madhurima,

Last days I have been playing with this parameters and I think that this situation is possible because you are considering a tumor purity of 0.5. In this scenario:

min-var-freq = min-var-freq x tumor-purity

So, the new cut-off for variant allelic frequency is 2.5% (0.05 x 0.5). Is it possible?

Thanks for sharing it!

Sergio

[Chirag Nepal]

Hi Madhurima and mbhattacharjee,

Could you please explain what filtering criteria you used for called snv from varscan2. I tried to use two of the scripts from varscan, with very little overlap between them. I have posted my quesion here, Varscan Filtering scripts: very little overlap between output of varscan filtering scripts

Thanks in advance.