powmic: Power assessment in microbiome case-control studies

In lichen-lab/powmic: powmic: an R package for power assessment in microbiome case-control

Power assessment in microbiome case-control studies

Description

Perform simulation-based power assessment for differential abundance analysis of metagenomic sequencing data. Microbiome data can be generated from distributions included Negative Binomial (NB), Zero-Inflated Negative Binomial (ZINB), Zero-Inflated Poisson Distribution (ZIP) and Dirichlet-Multinomial Distribution (DM). The parameters of these distributions as well as the microbial correlation network can be estimated from real datasets. The estimated parameters are used toe generate synthetic datasets for the power assessment. TPR and FDR are reported for different sample sizes. The users can choose a sample size with desired TPR and FDR in the study design.

Usage

powmic(n1s=c(20,40,60), n2s=c(20,40,60), nsims=10, params,
                    distrib=c('NB','ZINB','ZIP','DM'),
                    DAmethod=c("edgeR", "DESeq2","metaGenomeSeq","wilcox"), 
                    prevalence=0,verbose=TRUE) 

Arguments

n1s	Sample sizes in case group
n2s	Sample sizes in control group
nsims	Number of simulations
params	Estimated parameters for simulation from real datasets. Check how to set params in setParams and estParams
distrib	Distribution assumption of data generation
DAmethod	Differential abundance analysis methods
prevalence	Non-zero precentage of taxon
verbose	Print out implemented steps if TRUE

Value

An object "powmic" containing:

pvalues	A three-dimensional matrix of pvalue (#Taxon by #sample by #simulation)
fdrs	A three-dimensional matrix of fdr (#Taxon by #sample by #simulation)
x.abunds	A three-dimensional matrix of abundance (#Taxon by #sample by #simulation)
x.prevs	A three-dimensional matrix of prevalence (#Taxon by #sample by #simulation)
DAids	A list of vectors. Length of list is #simulation. Each vector contains the number of true Differential Abundant Taxa

See Also

setParams,assess,summaryAssess,plotStrata,plotStrataAll

Examples

## Not run: 
library(powmic)
data(params)
distrib='NB'
setParams=switch(distrib,ZIP=setParams.ZIP, NB=setParams.NB, ZINB=setParams.ZINB)
lmu0=log(params$mu)
lphi0=log(params$phi)
Sigma=params$Sigma
params=setParams(nTaxa=1000,p.DA=0.05,Sigma=Sigma,lmu0=lmu0,lphi0=lphi0,lfc.mu=2,
                        depth=1,sim.seed=123456)
powmic.out=powmic(n1s=c(20,40,60), n2s=c(20,40,60), 
                 params=params,distrib=distrib,DAmethod='edgeR')

assess.out = assess(powmic.out, alpha.type='fdr',alpha.level=0.1,stratify.type='prevalence')
sum.out=summaryAssess(assess.out,assess.type='overall')

## End(Not run)

lichen-lab/powmic documentation built on April 7, 2023, 4:40 p.m.