When estimating heritability using Haseman-Elston or PCGC Regression and including covariates, it is necessary to first produce adjusted kinship matrices, obtained by regressing the original kinship matrices on the covariates. This is because, whereas REML regresses the phenotype on the covariates and kinships jointly, HE and PCGC Regression instead adjusted phenotypes on adjusted kinships. Note that it is not necessary to separately obtain adjusted phenotypes (i.e., regress these on the covariates), because this is included within the HE and PCGC routines.
Always read the screen output, which suggests arguments and estimates memory usage.
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
The main argument is --adjust-grm <outfile>.
This requires the options
--grm <kinfile> - to provide a kinship matrix. Note that if plan to perform HE or PCGC Regression using multiple kinship matrices, you must adjust each one separately.
--covar <covarfile> - to provide covariates (in PLINK format).
If you will perform HE or PCGC regression using only a subset of samples, you should ideally use the same filtering now (e.g., use --keep <keepfile> to specify samples, and/or --pheno <phenofile> to consider only samples with non-missing phenotypes).
The adjusted kinship matrix will be saved with the stem <outfile> (see Kinship Formats for details of how kinship matrices are stored).
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Example:
Here we use the covariates covar.covar from the Test Datasets, and the kinship matrix with stem LDAK-Thin created in the example for Calculate Kinships.
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
./ldak.out --adjust-grm LDAK-Thin.covar --grm LDAK-Thin --covar covar.covar
The adjusted kinships are saved with stem LDAK-Thin.covar.
