Can anyone tell me the frequency percentage given by varscan2 is for individual or global?

Question

[ivivek_ngs]

Can anyone tell me the difference between the frequency output provided by the VarScan? I know one is for normal and other is for the variant allele in tumor. But is this frequency for the Variant allele likely said in website of the tool the global VAF or is it the VAF for the individual in the matched normal to control. Am a bit confused with that. If I have to consider for the somatic and germline events and its corresponding frequency of variant allele I should consider this frequency but can I term it as global VAF or is it population VAF? If not global VAF how should I calculate the global VAF? If I considered a published literature datasets with 400 datasets where the same variants is mentioned with a VAF value will that be global VAF and can be compared to my dataset whose VAF should be for the individual population? I am a bit confused but since am not getting some important cancer genes from my samples (which is 1 N/T for low grade cancer and 1 N/T for high grade cancer) I have to device a strategy to list out the driver and passenger mutations from this using both germlline and somatic variations. I would appreciate any suggestions given by the community.

Thanks

Answer 1

[Jordan]

Varscan gives you variant allele frequency of that particular sample. Allele frequency is calculated by varscan2 as,

reference depth/allele depth

for both normal and tumor.

I'm not sure what exactly global allele frequency means here. Is it the maf value reported by dbSNP?

[ivivek_ngs]

Thanks a lot for the reply. I am aware that this calculation is done by the way your have written. Only problem is that I just have 2 samples from which based on the VAF value I want to device a strategy to give driver and passenger mutations. Problem is that Tumor VAF is the usual one which I should be using but am not aware what is the global variant allel frequency, the one which is given by VarScan2 is for my samples but if I have to prioritize the candidate genes I have to see the relevance of these in global population right? From there it is important to calculate the MAF value of those mutations which have high variant allele frequency in my tumor samples. I want to know is this a right strategy or not. And the MAF value in dbSNP database can be used to prioritize for global populations or not?

[Jordan]

I see, if I'm understanding right you want to pick a handful of variants which you think are the causal variants for that tumor.

What we normally do in our lab is:

1. Filter only the somatic variants
2. Pick variants with at least 20% allele frequency in tumor.
3. Annotate the variants with dbSNP and COSMIC.
4. Remove the ones which are present in dbSNP. If you are looking for novel variants, remove the variants which are present in COSMIC.
5. You can also annotated them with thousand genome variants and remove the ones which are reported as the general variants in the population.

Note that what I have said is a very general strategy and there might be a few caveats depending on what kind of analysis you are looking for.

Does this help?

[ivivek_ngs]

Thanks a lot for the advice,