BDMMA: Bayesian Dirichlet-Multinomial approach for meta-analysis of...

In BDMMAcorrect: Meta-analysis for the metagenomic read counts data from different cohorts

Description

Bayesian Dirichlet–Multinomial approach for meta-analysis of metagenomic read counts

Usage

BDMMA(Microbiome_dat, abundance_threshold = 5e-05, burn_in = 5000,
  sample_period = 5000, bFDR = 0.1, PIPcut = 0.5)

Arguments

Microbiome_dat	A SummarizedExperiment object that includes the taxonomy read counts, phenotypes and batch labels.
abundance_threshold	The minimum abundance level for the taxa to be included (default value = 5e-05).
burn_in	The length of burn in period before sampling the parameters (default value = 5,000).
sample_period	The length of sampling period for estimating parameters' distribution (default value = 5,000)
bFDR	The false discovery rate level to control (default value = 0.1).
PIPcut	The threshold to cut the posterior inclusion probabilities (PIPs). By default, PIP is thresholding at 0.5.

Value

A list contains the selected taxa and summary of parameters included in the model.

selected.taxa	A list includes the selected taxa fesatures that are significantly associated with the main effect variable.
parameter_summary	A data.frame contains the mean and quantiles of the parameters included in the model. Each row includes a parameter's distribution summary and the parameter name is labeled in the first row. alpha_g: the baseline intercept of g-th taxon; betaj_g: the association strength between the g-th taxon and j-th input variables; deltai_g: the batch effect parameter of batch i, taxon g; L_g: the posterior selection probability of g-th taxon; p: the proportion of significantly associated taxa; eta: the standard deviation of the spike distribution (in the spike-and-slab prior).
PIP	A vector contains the PIPs of selected microbial taxa.
bFDR	The corresponding bFDR under the selected microbial taxa.

References

Dai, Zhenwei, et al. "Batch Effects Correction for Microbiome Data with Dirichlet-multinomial Regression." Bioinformatics 1 (2018): 8.

Examples

require(SummarizedExperiment)
data(Microbiome_dat)
## (not run)
## output <- BDMMA(Microbiome_dat, burn_in = 3000, sample_period = 3000)

BDMMAcorrect documentation built on Nov. 8, 2020, 5:50 p.m.