Hi,
A reviewer for my paper had concerns that CNVs determined from exome data in gnomAD v4.0 may have false positives that are not seen on CNVs from microarray data. Is there a publication from gnomAD that compares or validates the exome CNV to microarray CNV?
Thank you.
The methods for exome CNV detection used for gnomAD-v4 CNVs (GATK-gCNV) were published in Babadi et al, Nature Genetics 2023 (GATK-gCNV enables the discovery of rare copy number variants from exome sequencing data | Nature Genetics) which included extensive benchmarking against microarrays and WGS.
Thank you Dr Hoffman,
In the Babadi et al paper, they mentioned 4 filtering examples in table 1, each with different sensitivity and specificity. Can I check that gnomAD v4 uses the “recommended” QS filtering approach mentioned in Table 1?
Thanks,
Shuxiang Goh
This blog post has all of the details: Rare coding CNVs from exome sequenced individuals in gnomAD v4 | gnomAD browser. I probably should have just linked that in my first reply! You’ll see that the gnomAD CNV team applied the same filters as in Babadi et al, as well as additional filters to account for the increased cohort size and heterogeneity. They then repeated the same benchmarking on a subset of the gnomAD CNV samples to show gnomAD CNV attained >90% precision for CNVs > 2 exons as well.