somatic CNV PoN
I'm using GATK 4.3.0.
I was wanting to create my own PON for CNV analysis. I have no matched normals and I'm using a different sequencing technology to Illumina.
I was wondering if it was possible for me to use some freely available GIAB (i.e. NA12878 - sequenced with the same technology) to create the PON. I think I will only have two maybe three available to me to create the PON.
I know you would recommend a minimum of ~ 40 samples but is do you think I could still get useable information from this?
And just to check, can you do CNV analysis if you have tumour only by using this PON?
These are whole-genome samples? I don't have great intuition here because I've done much more work on the germline side, but I think it's worth trying both approaches: with the Illumina PoN and with the small number of samples on the same technology. Any chance the library prep is "PCR-free"? If they're not, then I would lean toward the smaller, more similar PoN giving better results because it will better capture sequencing biases that might otherwise appear as copy number events. Using allelic counts in the ModelSegments step will help to improve your results. High tumor purity (which you probably have no control over at this point) may be required for a project with so few samples. Your results will also potentially depend on how "active" your cancer type is -- fewer events like thyroid and pancreatic will be harder to find amidst noise the PoN couldn't successfully filter out.
Best of luck!
Please sign in to leave a comment.