Supporting data for "Large-scale whole-genome sequencing and comprehensive analysis identifies novel genes associated with systemic lupus erythematosus"

Systemic lupus erythematosus (SLE) is a prototypic autoimmune disease characterized by clinical heterogeneity and multisystem involvement. Its genetic basis is complex, encompassing both rare monogenic forms and common polygenic risk. Although numerous genes and susceptibility loci have been identified, they explain only a fraction of the heritability estimated from family and twin studies, highlighting the problem of “missing heritability.” Most genetic studies have relied on SNP-arrays and imputation, approaches that are limited in their ability to capture rare variants and variations in complex genomic regions, underscoring the need for more comprehensive genomic strategies to better define the genetic architecture of SLE.Here, we aim to overcome these limitations by performing whole-genome sequencing (WGS) in a cohort comprising 1,532 patients with SLE and 11,357 population controls. This comprehensive approach enables interrogation of both common and rare variants across the entire genome without reliance on imputation. We examined the ability of WGS to identify novel common variant signals, assess the contribution of rare variants to SLE heritability, and improve genetic risk prediction beyond SNP-array approaches.We first validated the WGS dataset by replicating established SLE risk loci and performing direct genotype‑level concordance analyses between WGS and SNP‑array platforms. Using WGS, we identified a previously unrecognized SLE‑associated signal at <i>A2ML1-PHC1</i>, a locus characterized by low imputation accuracy in prior SNP array studies. In addition, a meta‑analysis integrating our WGS data with existing SNP array‑based GWAS results uncovered four additional novel susceptibility loci, <i>PANTR1</i>, <i>SPINK14-SPINK6</i>, <i>IFFO1</i>, and <i>FXR2</i>, highlighting the value of combining WGS data with large‑scale array cohorts to enhance discovery power in SLE genetics.We also conducted a gene‑based rare‑variant association study (RVAS). This analysis yielded a rare‑variant signal upstream of <i>LOC105379476</i>, which remained stable under stringent leave‑one‑variant‑out (LOVO) sensitivity tests. Pathway‑level gene‑set analyses revealed nominal enrichment of pathogenic rare variants in several adaptive immune pathways, although significance was not retained after multiple‑testing correction. These findings suggest a role for rare variants in SLE susceptibility, while indicating that larger sample sizes will be required for definitive assessment.We evaluated polygenic risk score (PRS) performance using a 2×2 design crossing GWAS training data (WGS‑derived vs SNP array‑derived) with test genotypes (WGS vs SNP array). Across both PRSice‑2 and lassosum, models trained on WGS‑based GWAS and tested on WGS genotypes showed the best performance (Receiver Operating Characteristic Area Under the Curve (ROC AUC): 0.756-0.771), whereas all other combinations performed less well (AUC 0.605-0.681). These results indicate that WGS improves PRS accuracy, likely through enhanced effect‑size estimation and higher‑quality genotyping.In summary, our study demonstrates that WGS is a powerful tool for addressing the missing heritability of SLE. WGS complements SNP array-based approaches by recovering genetic signals in regions with poor imputation performance, reveals an aggregate contribution of rare variants to immune dysregulation, and provides a more robust foundation for polygenic risk prediction. Together, these findings highlight the value of WGS for achieving a more complete and accurate understanding of the genetic architecture of SLE.