Marginal Mediation Effects with Zero-Inflated Compositional Mediator

A way to estimate and test marginal mediation effects for zero-inflated compositional mediators. Estimates of Natural Indirect Effect (NIE), Natural Direct Effect (NDE) of each taxon, as well as their standard errors and confident intervals, were provided as outputs. Zeros will not be imputed during analysis. See Wu et al. (2022) .

MarZIC

Overview

This package provides a way to estimate and test marginal mediation effects for zero-inflated compositional mediators.

Installation

## From GitHub: 
devtools::install_github("quranwu/MarZIC")

Usage

Detailed instructions can be found in the vignette file.

Example

The example data was built from scratch, with 200 observations and 10 taxon.

library(MarZIC)

## A make up example with 1 taxon and 100 subjects.
set.seed(1)
nSub <- 200
nTaxa <- 10
## generate covariate of interest X
X <- rbinom(nSub, 1, 0.5)
## generate mean of each taxon. All taxon are having the same mean for simplicity.
mu <- exp(-5 + X) / (1 + exp(-5 + X))
phi <- 10

## generate true RA
M_taxon<-t(sapply(mu,function(x) dirmult::rdirichlet(n=1,rep(x*phi,nTaxa))))

P_zero <- exp(-3 + 0.3 * X) / (1 + exp(-3 + 0.3 * X))

non_zero_ind <- t(sapply(P_zero,function(x) 1-rbinom(nTaxa,1,rep(x,nTaxa))))

True_RA<-t(apply(M_taxon*non_zero_ind,1,function(x) x/sum(x)))

## generate outcome Y based on true RA
Y <- 1 + 100 * True_RA[,1] + 5 * (True_RA[,1] > 0) + X + rnorm(nSub)

## library size was set to 10,000 for all subjects for simplicity.
libsize <- 10000

## generate observed RA
observed_AA <- floor(M_taxon*libsize*non_zero_ind)

observed_RA <- t(apply(observed_AA,1,function(x) x/sum(x)))
colnames(observed_RA)<-paste0("rawCount",seq_len(nTaxa))
## Construct SummerizedExperiment object
CovData <- cbind(Y = Y, X = X, libsize = libsize)
test_dat <-
  SummarizedExperiment::SummarizedExperiment(assays = list(MicrobData = t(observed_RA)), colData = CovData)

## run the analysis
res <- MarZIC(
  Experiment_dat = test_dat,
  lib_name = "libsize",
  y_name = "Y",
  x_name = "X",
  num_cores = 1,
  mediator_mix_range = 1
)

Once the analysis is done, res is a list with four elements, each for NIE$_1$, NIE$_2$, NDE, NIE, respectively. The NIE$_1$, for example, could be extracted as:

NIE1 <- res$NIE1_save

And the significant result could be extracted as:

subset(NIE1,significance == TRUE)

Reference

Wu et al.(2022) MarZIC: A Marginal Mediation Model for Zero-Inflated Compositional Mediators with Applications to Microbiome Data. Genes 2022, 13, 1049.