Hi all, I am looking to benchmark germline CNVs from WES data generated by the DRAGEN v4.4 pipeline. I am using HG002 truth sets (NIST SV v0.6, CMRG v1.00, and the latest v5.0), along with benchmarking tools such as Wittyer and Truvari. So far, I have benchmarked small variants using hap.py and structural variants using Wittyer and Truvari. However, I am wondering whether it is appropriate to use an SV truth set to benchmark CNV call data specifically. Are these truth sets suitable for evaluating CNVs from WES, or is there a more appropriate reference for this purpose?

maybe if you convert CNV call data to SV (in BND format), benchmarking may work. But I dont have any experience with it.