bumpingr: Run a local bump-hunting algorithm

In brentp/clustermodelr: Models on clusters of correlated data

Description

This performs a similar task to the Bump-Hunting algorithm, but here the meth argument is a putative bump. The residuals of the null model are shuffled added back to the null model and the beta's of that simulated data are repeatedly stored and finally compare to the observed coefficient. Due to the shufflings, this is much slower than the other functions in this package.

Usage

bumpingr(formula, covs, meth, weights = NULL, n_sims = 20, mc.cores = 1)

Arguments

formula	an R formula containing "methylation"
covs	covariate data.frame containing the terms in formula except "methylation" which is added automatically
meth	a matrix of correlated data.
weights	optional weights matrix of same shape as meth
n_sims	this is currently used as the minimum number of shuffled data sets to compare to. If the p-value is low, it will do more shufflings
mc.cores	sent to mclapply for parallelization

Value

list(covariate, p, coef) where p and coef are for the coefficient of the first term on the RHS of the model.

brentp/clustermodelr documentation built on May 13, 2019, 5:11 a.m.