Abstract
<title>Abstract</title> <p>Genomic structural variants (SVs) constitute a major source of human genetic diversity and disease susceptibility. Yet, population-scale SV genetics lacks a viable mechanism to synthesize massive individualized callsets into reliable cohort resources. Traditional heuristic tools force a compromise between memory exhaustion and precision degradation, severely limiting their scalability. To resolve these bottlenecks, we present KGGSV, an end-to-end framework that conceptually unifies secure data governance with high-throughput genomic computing. KGGSV couples PACKER—a storage architecture enabling zero-copy logical subsetting and on-the-fly streaming decryption—with PICO, an order-independent, density-based parallel clustering engine that maintains high merging fidelity under strict memory boundaries. Utilizing KGGSV, we harmonized a population catalog of 3.68 million SVs across 490,276 UK Biobank genomes on a single computing node in just 13 hours. A comprehensive phenome-wide association study (PheWAS) across 877 complex traits in ~389,000 unrelated individuals identified 154,506 significant SV-trait associations. Crucially, we isolated unique structural risk loci—including 35.8% of asthma-associated SVs—that are uncoupled from local single-nucleotide variant (SNV) architecture. Furthermore, trans-chromosomal network partitioning unmasked authentic pleiotropic axes, capturing shared mucosal and epithelial vulnerabilities between respiratory and digestive disorders. Our study establishes a highly scalable and secure framework for translating individualized structural variation into population-scale genetic discoveries.</p>