Question

PureCN.R _variants.csv file prior.somatic and somatic variant count

0

Entering edit mode

twtoal ▴ 10

@twtoal-15473

Last seen 14 months ago

United States

In the PureCN.R output file *_variants.csv, am I correct to think that I can get a quick count of the number of somatic variants that PureCN.R saw in the input VCF file by adding the number of entries > 0.9 in column "prior.somatic"? (A log message said that variants marked SOMATIC were found and their prior probability was set to 0.999).

PureCN.R is not seeing most of my somatic variants, or at least is not putting the prior value to 0.999. For one sample, which should have about 120 somatic variants, only 1 variant has prior probability > 0.9.

What filtering might PureCN be applying to my somatic variants? I didn't see log file messages indicating that a bunch of them had been filtered out.

PureCN Somatic Mutations • 1.2k views

ADD COMMENT • link updated 6.0 years ago by markus.riester ▴ 130 • written 6.0 years ago by twtoal ▴ 10

score 0 · Answer 1 · 2018-05-03

0

Entering edit mode

markus.riester ▴ 130

@markusriester-9875

Last seen 22 months ago

United States

The filtering is done by the filterVcfBasic function.

You can load your VCF in R like PureCN does with:

vcf <- PureCN:::.readAndCheckVcf(vcf.file, "hg39")

Are you sure your VCF is fine? Have you compared it to Mutect 1 or Mutect 2 with --genotype-germline-sites? Do you get largely overlapping somatic calls?

ADD COMMENT • link 6.0 years ago markus.riester ▴ 130

0

Entering edit mode

So far I haven't found anything wrong with the VCF. However, I noted when trying to run the vcf object through filterVcfBasic() that in one place it attempts to get the depths from the DP genotype field, which does not exist in my VCF. When I run it by hand, it gets a fatal error. Yet that isn't happening when I run PureCN.

I'm wondering if I should add a FORMAT/DP field?

Later -- I went ahead and added a DP field, and then when I ran it, PureCN crashed with the following output:

    INFO [2018-05-03 18:05:55] CG-HFLLA-JG-74T3 is tumor in VCF file.
    Error in if (sum(homozygous)/length(homozygous) < 0.001) { :
    missing value where TRUE/FALSE needed
    Calls: runAbsoluteCN -> getSexFromVcf

ADD REPLY • link 6.0 years ago twtoal ▴ 10

0

Entering edit mode

Did you load the VCF with PureCN:::.readAndCheckVcf(vcf.file, "hg38")? This should take care of necessary standardizations like adding DP.

ADD REPLY • link 6.0 years ago markus.riester ▴ 130

0

Entering edit mode

No, ok, thx.

However, the crash occurred when I was running PureCN normally.

ADD REPLY • link 6.0 years ago twtoal ▴ 10