FastHer: Fast Local Heritability Estimation from GWAS Summary Statistics

FastHer is an open-source R package for rapid and accurate estimation of local SNP heritability using genome-wide association study (GWAS) summary statistics and linkage disequilibrium (LD) matrices. The package implements a restricted maximum likelihood (REML) framework. Computational efficiency is achieved through an analytical reformulation of the likelihood function using eigen-decomposition of the LD matrix, which reduces the per-iteration complexity from cubic [O(m³)] to linear [O(m)] time relative to the number of SNPs (m). This enables the analysis of extended genomic regions (tens of thousands of SNPs) within practical timeframes, making genome-wide scans feasible for biobank-scale datasets. The package provides functions for data input, maximum likelihood optimization, and output of heritability estimates with analytical standard errors. It supports standard file formats and includes documentation, tutorial examples, and demo datasets.