README.md

In liudoubletian/ZIMED: The effect decomposition of mediation analysis for zero-inflated models in microbiome data

ZIMED (The effect decomposition of mediation analysis for zero-inflated models in microbiome data)

Introduction

The characteristics of microbiome data are complicated: sparsity and over-dispersion. However, there are few existing causal mediation methods specifically designed to handle sparse and high dimensional microbiome data. We develop a novel zero-inflated mediation effect decomposition model (ZIMED) to estimate and test the mediation effects of the microbiome utilizing the zero-inflated negative-binomial (ZINB) regression model.

Installation

install.packages("devtools")  
devtools::install_github("liudoubletian/ZIMED")  
library(ZIMED)  

Basic Usage

mediation analysis for microbiome data

zimed(M_mat,Treat,Outcome,method="joint",ci.method="delta)

• M_mat : an OTU table with n rows (samples) and m columns (taxa)
• Treat : a n-vector of group indicators
• method : joint, HDMT or DACT
• ci.method : bootstrap or delta method

it returns a list of results: NIE : the estimated natural indirect effect NIE.p : the calculated p value for NIE NIE.ci : the calculated confidence interval for NIE NIEA : the estimated natural indirect effect through changes of abundance NIEA.p : the calculated p value for NIEA NIEA.ci : the calculated confidence interval for NIEA NIEP : the estimated natural indirect effect through changes of absence/presence NIEP.p : the calculated p value for NIEP * NIEP.ci : the calculated confidence interval for NIEP

liudoubletian/ZIMED documentation built on April 11, 2022, 12:05 a.m.