VAF considered by HAPPY sensitivity and specificity

[Belle-AM]

We are using hap.py to compare various sequencing runs processed via the DRAGEN pipeline. In the DRAGEN output vcf, any variant with a VAF of at least 0.01 is reported. However, any variant with less than a VAF of 0.05 is marked as lowvaf in its filter column.

When hap.py reportsthe "PASS" results for concordance, I assume that the comparison excludes variants with the lowvaf filter, and thus sensitivity and specificity are based on only those variants with at least 0.05 VAF. Can this be confirmed?

[tetedange13]

Hi @Belle-AM,

I am only a user of hap.py, not one of the developper

But I can tell you that by default hap.py calculates performances (sensitivity, specificity and so on) both :

• On "all" variants (= regardless of the FILTER field in input VCF)
• Only on variants with "FILTER=PASS"

You should be able to confirm that on your own "summary.csv" outputted by hap.py
=> Because this file has 2 rows for SNP (= "ALL" and "PASS") and same for inDel

To complete, hap.py check chrom + pos + ref + alt + genotype, but not VAF (som.py can do that)

Hope this helps !
Kind regards,
Felix.