Partitioned SNP heritability results from LD score regression with the baseline-LD v1.1 model (Finucane et al. 2015; Gazal et al. 2017). For each annotation, this yields two potential parameters of interest:

  • coefficient \(\tau\): increase in per-SNP heritability associated with the given annotation conditional on all other annotations in the model
  • enrichment: marginal increase in per-SNP heritability for SNPs in the given annotation (i.e.┬ásummed across any overlapping annotations)

For easier comparison of coefficient tau across phenotypes, we report here the value after normalization for the total estimated SNP heritability of the phenotype, \(\tau^*\). The full results file available for download reports the unstandardized value. Separate results are reported for the continuous annotations, as described below.

Reported partitioning results are limited to phenotypes with strongly significant \(h^2\) after Bonferroni correction (\(z > 7\)) and medium or high confidence. See the Methods pages for more detail on the confidence ratings and significance thresholds. From the annotations in the Baseline-LD model, we omit results for the base annotation, 10 MAF bin annotations, and the 26 buffer annotations (extending 500-bp around the baseline functional annotations) since they are primarily intended as statistical controls rather than for direct inference. Results for these annotations are included in the full results file available for download.

Binary Annotations

Coefficient p-values

Note: For efficiency only p-values < .1 are displayed. The displayed 95% confidence bands assume tests are independent across phenotypes and do not account for multiple testing across annotations.