Inquiry regarding expected Recall performance for Single-Sample benchmarks (Illumina WGS)

[lmtani]

Hello,

I am running the GATK-SV pipeline in single-sample PCR-free mode using Illumina paired-end 2x150bp reads. I benchmarked the results against the GIAB HG002 Tier 1 v1.1 truth set using truvari.

Observed Results:

• PASS variants: 5.792
  • Precision: 95.80%
  • Recall: 25.50%
• All variants (unfiltered): 8.389
  • Precision: 72.60%
  • Recall: 27.40%

While my precision (~96%) aligns with the reference paper, observed recall remains low (~25%). Is this recall range expected for single-sample Illumina inputs?

Thank you.