Filtering germline with VAF instead of matched normal

[katestan]

I want to filter germline variants using VAF instead of a matched normal sample (either undiluted nanoseq or standard wgs). For this I think all I have to do is construct a sample file that looks like:

id,d_fastq1,d_fastq2,n_fastq1,n_fastq2
P10,P10_R1.fastq.gz,P10_R2.fastq.gz,P10_R1.fastq.gz,P10_R2.fastq.gz

Where the d_fastq and n_fastq are the exact same file. I think randomreadinbundle will then construct a neat.cram for the n_fastq and use this to filter variants - in this case it would just be the deduplicated cram for the exact same nanoseq sample. I know that var_z sets the minimum coverage in the matched normal, but is there a parameter that allows me to set the minimum VAF in the matched normal? Or minimum number of supporting reads?

From reading other issues I should also set -b to 0 if using a nanoseq normal?

Is there documentation where each of these parameters are clearly defined?
From NanoSeq_main.nf:
params.var_a = 50
params.var_b = 0
params.var_c = 0.02
params.var_d = 2
params.var_f = 0.9
params.var_i = 1.0
params.var_m = 8
params.var_n = 3
params.var_p = 0
params.var_q = 45
params.var_r = 144
params.var_v = 0.01
params.var_x = 8
params.var_z = 15

Any help greatly appreciated!

[fa8sanger]

Hi Kate, see the attached excel for an ongoing attempt to describe all the parameters.
What is the depth of coverage after deduplication in your sample? To use one sample as its own matched normal you need: high depth and that the sample is highly polyclonal.

You can set -b to 0 for everything

Var_v controls the maximum vaf in the matched normal

Let me know if there's anything unclear in the attached file

NatProtocol_NanoSeq_parameters.xlsx

[katestan]

Thanks so much for the xlsx with defined params!

We have 30X whole-genome nanoseq data. We expect the sample to be highly polyclonal. Is using VAF appropriate, here?

Using Var_v 0.01 as a VAF cutoff is confusing me. Wouldn't we want to set the minimum VAF in the matched normal, not the maximum VAF in order to filter for germlime? That is, we want to filter out any variants that are above let's say 0.1 or 0.3 VAF in the matched normal as germline.

[fa8sanger]

If you have an independent matched normal, I would go for var_v of 0.01. With this, basically any somatic mutation seen also in the matched normal with VAF > 1% will be filtered. This, in my experience, gives the cleanest results. But feel free to go for 0.1. If you go for 0.3 with 30X you would get lots of SNPs leaking below 0.3.

[katestan]

When using the same sample as the matched normal (as in exactly the same fastq files as tissue/normal) and a cutoff of 0.01 all of my true somatic variants are filtered out. I know they are true somatic variants because we have the matched blood reference that we used in a separate run. Would you agree that this strategy of using VAF to filter for germline is really only possible with the higher coverage targeted nanoseq data (and polyclonal tissue)?

[fa8sanger]

Yes, if you want to use the same sample as the matched normal you need high coverage and relax the cutoff to at least 0.1