multipheno.OBrien: Multi-phenotype 1 df test for association
In tobyjohnson/gtx: Genetics ToolboX

Description Usage Arguments Details Value Author(s)

For a set of phenotypes, given summary results for association tests for each single phenotype, over a large fixed set of marker genotypes (e.g. GWAS results), this function calculates a 1 degree of freedom multi-phenotype association test, which has a close correspondence to O'Brien's test.

1	multipheno.OBrien(z, cor.use = NULL, cor.method = "spearman")

`z`	a matrix of association Z statistics, with rows corresponding to markers and columns corresponding to phenotypes.
`cor.use`	a logical vector of length nrow(z), indicating a subset of markers to use to calculate the correlation matrix.
`cor.method`	a method to calculate the correlation matrix.

The function is named for the close correspondence with O'Brien's test.

It is the user's responsibility to ensure that the columns of “z” are marginally well calibrated, i.e. over a set of null markers, each column of “z” should be standard normal marginal to the other columns of “z”.

A list with three elements. The first element, “rmatrix”, is the estimated test statistic correlation matrix. The second and third elements are both of length nrow(z); “OB” contains the test statistics, which are chi-squared with 1 d.f. under the null, and “pval” contains the P-values.

Toby Johnson Toby.x.Johnson@gsk.com

tobyjohnson/gtx documentation built on Aug. 30, 2019, 8:07 p.m.