The plots here are designed to allow browsing overall trends in the SNP heritability and intercepts results from LD Score regression of the Neale Lab GWAS. Individual phenotype results can be viewed in the browser.
All plots below are restricted to results with at least some confidence in the LDSR results, and report SNP heritability on the liabulit scale. See the Methods section for more information on our selection of GWAS phenotypes, confidence ratings for the LDSR results, and significance thresholds.
In addition to the plots below, additional visualizations are available for considering:
Previous plots from the Round 1 results can be found in the Archive section.
The average SNP hertiability estimate across all 4178 primary GWAS is 0.065 (median = 0.058). Average estimates increase when restricting to any confidence (mean = 0.078, median = 0.059) or high
confidence (mean = 0.088, median = 0.062). Strong departure from the null hypothesis of \(h^2_g=0\) is observed across phenotypes, especially among high confidence results.
Note: The displayed 95% confidence bands assume tests are independent across phenotypes and do not account for multiple testing across annotations.
Note: The displayed 95% confidence bands assume tests are independent across phenotypes within confidence rating and do not account for multiple testing across annotations. Read more about the confidence ratings here.
Relative GWAS confounding or LDSR model misspecification is limited across the phenotypes, with a median intercept ratio of NA. Intercepts remain statistically significant for many phenotypes, including many high
confidence phenotypes. Outlier intercept values > 1.2 are observed for home and birth location, height, and certain biomarkers.
Note: Ratio estimates restricted to phenotypes with mean \(\chi^2 > 1\). Clearer results are observed by restricting to high
confidence results and/or higher effective sample size using the filter on the right. The LDSR ratio estimate is:
\[\frac{\text{intercept}-1}{\text{mean}(\chi^2)-1}\]
Note: The displayed 95% confidence bands assume tests are independent across phenotypes and do not account for multiple testing across annotations.
Note: The displayed 95% confidence bands assume tests are independent across phenotypes within confidence rating and do not account for multiple testing across annotations. Read more about the confidence ratings here.
Among medium
and high
confidence phenotypes, there’s a positive relationship between the \(h^2_g\) estimate and the intercept estimate. This trend may indicate some degree of model misspecification in the paritioned LDSR model. Sample size may also be a factor if \(h^2_g\) is biased at some sample sizes, but comparison of \(h^2_g\) to the intercept ratio shows a similar trend towards non-zero ratios where \(h^2_g\) is large.
Note: Clearer results are observed by restricting to high
confidence results, continuous variables, and/or higher effective sample size using the filter on the right. Colors indicate our confidence rating as described in the Methods section.
Note: Ratio estimates restricted to phenotypes with mean \(\chi^2 > 1\). Clearer results are observed by restricting to high
confidence results, continuous phenotypes, and/or higher effective sample sizes using the filter on the right. Colors indicate our confidence rating as described in the Methods section. The LDSR ratio estimate is:
\[\frac{\text{intercept}-1}{\text{mean}(\chi^2)-1}\]