Both gnomAD SVs and CNVs v4.1.0 show individuals with some sort of copy number alteration. My understanding is that gnomAD SVs will include individuals with as small as a partial exon deletion (or perhaps at least one exon deleted) but gnomAD CNVs requires at least 3 exons to be deleted. I also know that gnomAD SVs relies on WGS data whereas gnomAD CNVs relies on WES data.
Are there any other differences between these cohorts?
Does SVs only show intragenic events whereas CNVs only shows multigene events?
Could any of the individuals who had WGS for inclusion in SVs also had WES for inclusion in CNVs?
The gnomAD v4.1.0 SVs and CNVs are separate callsets that were generated from different data with different methods, so they have quite a few differences. I recommend consulting the blog posts for detailed information: CNVs in v4, SVs in v4.
Briefly, the SVs were generated from the gnomAD v4 genome data (~63k genomes) using the GATK-SV pipeline, so they include balanced SVs in addition to CNVs, the calls are genome-wide and at all frequencies, the resolution is higher, and smaller events are included (minimum 50bp). The CNVs were generated from the gnomAD v4 exome data (~464k exomes) using the GATK-gCNV pipeline, so the CNVs are for the exome only, and the resolution is more limited and the underlying data is noisier, so only rare (AF<1%) autosomal CNVs impacting 3 or more exons are included. The SVs can span multiple genes, and the CNVs can be intragenic. The gnomAD exome samples and genome samples do not overlap; you can read more in the v4 blog post.