Multiple Phenotype Genome-Wide Association Studies


DISSECT can conduct individual association tests on each variant as in genome-wide association analysis, on multiple phenotypes, in an optimized way to improve resource usage and time.


This analysis has to be performed in different steps:

1) Compute the diagonalized GRMs (see here for other options) you want to correct for:

dissect --make-grm --bfile genotypes --diagonalize --out results

2) Compute the residuals:

dissect --mpresiduals --grm grm --pheno individuals.phenos --covar covars.dat --qcovar covars.dat --out residuals

The covariates files have to use the same format as in other analysis. The phenotypes file uses also the same format, but it must contain a header, the two first columns containing “FID” and “IID”, respectively. The remaining columns must contain all the phenotypes which will be analyzed, and all the individuals in the GRM have to have a phenotype. To use a Leave-One-Chromosome-Out (LOCO) approach, 22 residuals have to be computed, each one using a GRM computed by keeping out the corresponding chromosome. Other approaches are also possible.

3) Perform the GWAS:

dissect --mpgwas --bfile-residuals-list genotyped.residuals.list --redist-meth2 --out results

Where the file genotyped.residuals.list contains a list of paired file names for the genotype and residuals pairs one wishes to test. The first column has to contain the genotype file, the second column, the residuals which will be tested against the corresponding genotype file. For instance, for a Leave-One-Chromosome-Out (LOCO) approach, the variants in the chromosome 6, have to be tested against residuals computed using a GRM not containing the chromosome 6. The results for all phenotypes will be stored in the same output file.