VarScan2. Strange behaviour of somatic-p-value

[slois]

Hi community of SEQanswers,

I'm running varscan2 for somatic mutation calling following the commands of its documentation: mpileup + varscan somatic

In order to test the effect of --somatic-p-value parameter in the number of variants called by varscan somatic, I have tried different p-values and logically I obtain different number of variants in the final output. However, checking the differents outputs I see that the value of the field somatic p-value (col. 15 in the tabulated output or value SPV of col. INFO in the VCF format) is greater than the specified threshold. I expected that all the somatic variants of the output has a somatic-p-value below my threshold.

Is there any reason to explain this behaviour? Am I doing something wrong? I would like to know whether --somatic-p-value parameter is filtering variants according its value or not.

Thank you in advance for you answers,

[slois]

Hi,

According the documentation:
--p-value - P-value threshold to call a heterozygote [0.99]
--somatic-p-value - P-value threshold to call a somatic site [0.05]

Are these parameters filtering the variants identified according its corresponding values?

Different tests applied using different somatic p-values: (outputs are modified to fit in this post). Last two columns are the variant-p-value and somatic-p-value:

Results obtained for --somatic-p-value = 1
result1.snp:chr1 1019753 A G 10 0 0% A 8 2 20% R Somatic 1.0 0.2368
result1.snp:chr1 1111270 C T 36 2 5,26% C 52 17 24,64% Y Somatic 1.0 0.0090
result1.snp:chr1 1111272 C T 35 2 5,41% C 55 16 22,54% Y Somatic 1.0 0.0182

Results obtained for --somatic-p-value = 0.05 (Default)
result2.snp:chr1 1019753 A G 10 0 0% A 8 2 20% R Somatic 1.0 0.2368
result2.snp:chr1 1111270 C T 36 2 5,26% C 52 17 24,64% Y Somatic 1.0 0.0090
result2.snp:chr1 1111272 C T 35 2 5,41% C 55 16 22,54% Y Somatic 1.0 0.0182

Results obtained for --somatic-p-value = 0.01
result3.snp:chr1 1019753 A G 10 0 0% A 8 2 20% R Somatic 1.0 0.2368
result3.snp:chr1 1111270 C T 36 2 5,26% C 52 17 24,64% Y Somatic 1.0 0.0090

Results obtained for --somatic-p-value = 0.001
result4.snp:chr1 1019753 A G 10 0 0% A 8 2 20% R Somatic 1.0 0.2368

Thank you

[vd4mindia]

I also have a same question,

I have also ran the VarScan2 on my normal/tumor pair with p-value as 0.1 somatic p-value 0.001 and then ran the somatic filter with parameters -min-strands2 2 -min-avg-qual 30. I get very less variants that are significant. I get around 250 variants. Can anyone tell me what is the standard threshold for the p-value and somatic pvalue? Which is the paramter you have been supplying?