SumHer is our new tool for analyzing summary statistics. It has four aims:

Estimate the SNP heritability of a trait
Estimate the amount of confounding bias in a GWAS
Estimate heritability enrichments
Estimate genetic correlations

(in short, it's a version of LDSC which allows the user to specify the heritability model).

There are two ways to run SumHer; either manually, or using the SumHer Server.

To run SumHer manually, you require a (well-matched) Reference Panel and (correctly-formatted) Summary Statistics from a GWAS. Regardless of the aim, there are two steps: the first is to compute a Tagging File, which provides the (relative) expected heritability tagged by each predictor under the chosen Heritability Model; the second is to regress the summary statistics onto these.

At the moment, SumHer is designed only for common predictors (MAF>0.01). When we extend the method to accommodate rare predictors, we will let you know.