Comparison of our current primary results from partitioned LDSR of the Round 2 Neale Lab GWAS to previous results from paritioned LDSR of Round 1 GWAS. Both results use the Baseline LD v1.1 model. The Round 2 GWAS results use a slightly broader sample set than ‘white British’ and incorporate additional covariates (more detail on github).
Since the sample sizes are different, we focus here on comparison of the ratio. Smaller ratios in the current results from Round 2 GWAS suggest reduced confounding, likely due to the expanded set of covariates. P-values for the intercept estimates are similarly less significant in the Round 2 results.
Note: Comparison limited to phenotypes with at least some confidence in the Round 2 results. Trend is clearer when filtering to higher effective sample size using the slider on the right.
SNP heritability results are highly concordant between Round 1 and Round 2. SNP heritability is increased for a handful of phenotypes where the updated GWAS covariates are likely to have strong effects. P-values for the Round 2 SNP heritability results are slightly stronger in the increased sample size.
Note: Comparison limited to phenotypes with at least some confidence in the Round 2 results. Trend is clearer when filtering to higher effective sample size using the slider on the right.
We can compare the standard errors (SEs) of the estimates between Round 1 and Round 2 to evaulate how much precision was gained from the expanded sample size in Round 2. We observe generally smaller SEs for the SNP heritability and slightly increased SEs for the intercept.