Hello,
I am working on reproducing results from TCGA data on prostate cancer using MuTect2. I will state the main differences between their parameters and mine, and also some results:
- They used GRCh38 (GRCh38.d1.vd1.fa) as a reference genome was modified by adding decoy and viruses sequences while I used hg19.
- They decided to exclude the Chr7 intervals (intervals=[chr7:1-30000000]) whereas I excluded all intergenic and intragenic regions of the genome since we are working only on exome (by substracting the intervals of the library kit from the genome).
- standard_min_confidence_threshold_for_calling = 30.0 (default in stable version is 10).
Other than that, I used the same dbsnp version, an newer cosmic version. As for the alignment files, I had to convert them back to FastQ and realign them again against hg19.
In one of the samples, they report a 31 that passed the filters only. As for mine, while the job is still running on the cluster and still on Chr7, the number of variants that passed the filters are 359 !
Any suggestion or opinion on why such difference would be appreciated, thanks in advance !
Alaa