SAFE-LD: A novel method for the estimation of linkage disequilibrium from summary statistics

Genome-wide association studies (GWAS) have advanced understanding of complex traits, but downstream analyses require accurate in-sample LD, which is difficult to share due to privacy restrictions. SAFE-LD (Shrinkage and Anonymisation Framework for LD Estimation) generates pseudo-genotypes that reproduce exact in-sample LD while discarding individual-level data, stored in VCF/PGEN formats for seamless use with standard pipelines. SAFE-LD achieves fine-mapping accuracy comparable to internal LD and outperforms external LD, providing a scalable, privacy-preserving, and efficient alternative. An extension, SAFE-LDss, enables the same from published GWAS summary statistics, supporting precise fine-mapping without compromising privacy.