lotman: Low-throughput mediation analysis (Lotman)
In jdreyf/Hitman: High-throughput mediation analysis

lotman

R Documentation

Low-throughput mediation analysis (Lotman)

Description

Low-throughput mediation analysis to test if rows of M mediate the effect of exposure E on outcome Y. Before applying Lotman, you should know a priori the direction that E changes Y and verify that E changes Y significantly and in the same direction here.

Usage

lotman(
  E,
  M,
  Y,
  covariates = NULL,
  fam = "gaussian",
  reorder.rows = TRUE,
  fdr.method = c("BH", "BY"),
  verbose = TRUE,
  check.names = TRUE
)

Arguments

`E`	A numeric vector of exposures.
`M`	A numeric matrix-like data object with one row per feature and one column per sample of mediators.
`Y`	A numeric vector of `length(E)` of outcomes. Only continuous, normally distributed outcomes currently supported.
`covariates`	Numeric vector with one element per sample or matrix-like object with rows corresponding to samples and columns to covariates to be adjusted for.
`fam`	Character string of family to use in generalized linear model of `Y`. The default `"gaussian"` reduces to the usual linear regression model. See stats::family.
`reorder.rows`	Logical, should rows be reordered by p-value?
`fdr.method`	Character string; either "BH" for Benjamini-Hochberg or "BY" for Benjamini-Yekutieli. See stats::p.adjust.
`verbose`	Logical; should messages be given for lack of association between `E` & `Y` and filtering?
`check.names`	Logical; should `names(E)==colnames(M) & colnames(M)==names(Y)` be checked?

Details

E and Y cannot have NAs. M may have some NAs, but rows that have less non-missing values than 5 + ncol(covariates) will be filtered out, and if verbose=TRUE, a message will be written with the number of rows filtered out.

Larger chi-square values are more significant.