Illumina / Pisces

Somatic and germline variant caller for amplicon data. Recommended caller for tumor-only workflows.
GNU General Public License v3.0
96 stars 16 forks source link

WRONG VF FOR LOW FREQUENCY VARIANTS #74

Open AriadnaColmeneroCobo opened 1 year ago

AriadnaColmeneroCobo commented 1 year ago

Good afternoon,

I am using the following code to report variants for some targeted sequencing analysis samples.

After applying the following code, I could see how the low frequency variants were misreported. That is to say, in the other 3 variant callers I was getting a VAF, for example, of less than 0.01, while PISCES, of more than 0.2. At this point, I noticed that the VF was being reported with a ",", which is making me doubt that it is really OK, since the other variant callers report the VF with ".". I'm using PISCES/5.3.0.0-FORK. Also, when I add the --MinVF 0.001 argument, it stops working. I understand why in this case it has a . and not a ",",....


## A.1) Gemini
bashArguments = "GeminiMulti -bam "+cleanBqsrBam+" -genome "+refGenomeFolder+" -samtools "+samtoolsPath+" --outFolder "+outDir+"/pisces/gemini/"+sample+" -numProcesses "+cpus+" -exepath "+geminiExePath
## A.2) Pisces

bashArguments = "Pisces -bam "+bamGemini+" -g "+refGenomeFolder+" -i "+targetBedFileIntervalList+" -UseStitchedXD TRUE -CallMNVs TRUE -MinBQ 30 -MinVF 0. 001 -MinMQ 10 --OutFolder "+outDir+"/pisces/pisces/"+sample+" --MaxNumThreads "+cpus

The line obtained taking into account that the VF is reported with a comma:

GT:GQ:AD:DP:VF:NL:SB 0/1:43:155,9:164:0,55:20:-14,2244

In addition, I should add that the QUAL of these variants is variable, some of them reaching 100.

It's a bit of a strange situation and I would need your help and advice.

Thank you so much,

tamsen commented 1 year ago

Hi there,

Thank you for using pisces!

Please move to the most recent version of Pisces at https://github.com/tamsen/Pisces. The illumina version is no longer supported. If you still have an error with the most recent version, please report the issue on that site.

As a side note, if you are seeing commas where you should see decimal points, this might have to do with your text editor. European-style text editors sometimes take "0.55" and make it "0,55". To verify this, open your raw output vcf in directly in vi or emacs or something and see if your decimal points come back. And for a crashing program, it never hurts to double check your inputs. Ie, check for white space or spontaneous commas etc appearing in your input as well as your output.

Good luck!

Best, Tamsen