project_sparse: A streamlined function to project a trait onto a sparse basis...
In ollyburren/cupcake: PCA for GWAS

A streamlined function to project a trait onto a sparse basis project_sparse

1	project_sparse(beta, seb, pids)

`beta`	a vector of beta estimates
`seb`	a vector of standard error of the beta estimates
`pids`	a vector of primary identifiers for SNPs with the same order as beta and seb

a data.table with the following columns#'

PC - principal component label
var.proj - Variance of the projection score.
delta - The difference between projection score and pseudo control score.
p.overall - The p value over all components for the projection score.
z - z score for projection score
p - p value for projection score.

This function assumes that the following objects are defined in the current environment

rot.pca - Matrix of rotations, usually obtained from PCA via prcomp.
beta.centers - Vector of basis SNP beta centres, labelled by pid.
shrinkage - Vector of basis SNP shrinkage values, labelled by pid.
LD - Matrix of covariance between basis SNPs

This function assumes that the order snps in arguments is the same. Whilst missing SNPs are allowed this will degrade the projection a warining is issued when more than 5

ollyburren/cupcake documentation built on July 30, 2020, 9:58 a.m.