Once you have a Tagging File, you can estimate SNP heritabililty using the command
which requires the options
--tagfile <tagging_file> - specifies the tagging file
--summary <sum_file> - specifies the file containing the summary statistics.
It is possible to use --extract and --exclude here, but ideally all filtering should be performed when constructing the tagging file (and if using the LDAK Model, also when computing predictor weightings).
LDAK will report an error if it finds predictors for which the alleles in the summary statistics file are not compatible with those in the tagging file. If only a few predictors are affected, it should be OK to override this error by adding --check-sums NO.
If you suspect that the summary statistics were subjected to genomic control, come from mixed-model association analysis, or are confounded due to population structure or relatedness, you should add --genomic-control YES (see estimating Confounding Bias for more details).
If the summary statistics come from analysing a binary phenotype, then you can use --prevalence and --ascertainment to specify the proportion of cases in the population and in the GWAS; LDAK will then also report estimates of varianc explained on the liability scale.
The estimate of SNP heritability is saved in <output>.hers.
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
To regress the summary statistics onto the tagging file (assuming no confounding), use
../ldak.out --sum-hers sumldak --tagfile sumldak.tagging --summary height.txt
This script fails for me because "168 predictors have inconsistent alleles" (i.e., the alleles in height.txt are not compatible with those in ref.bim). Ideally I would have identified and removed these predictors when formatting the summary statistics; however, as there are so few, it will suffice to instead instruct LDAK to ignore this error.
../ldak.out --sum-hers sumldak --tagfile sumldak.tagging --summary height.txt --check-sums NO
The estimate of SNP heritability is in sumldak.hers.