AFT-LDSC Summary Statistics for 508 UK Biobank Phecodes

This dataset contains summary statistics derived from a survival-based association analysis of 508 phecode-defined phenotypes in the UK Biobank, formatted for use with AFT-LDSC. Associations were computed on ~1.16 million HapMap3 SNPs across 381,315 unrelated European-ancestry participant. For each phenotype, three parametric accelerated failure time (AFT) models were fit — log-normal, log-logistic, and Weibull — and the best-fitting model was selected per phenotype via goodness-of-fit testing; the summary statistics and alpha values provided here correspond to that best-fitting model. The shared data includes: sumstats.tar, an archive of 508 per-phenotype summary-statistic files ({phecode}.sumstats.gz) with columns SNP, A1, A2, Z, N, with alleles harmonized to the UK Biobank imputed genotypes; alphas.tar, an archive of 508 {phecode}.alpha files, each storing the scalar alpha parameter of the best-fitting AFT model for the corresponding phenotype; and metadata.csv — a table mapping each phecode to its human-readable phenotype name. These files are directly compatible with the AFT-LDSC software and can be used for heritability estimation, genetic correlation analysis, and partitioned heritability applications for time-to-event phenotypes.