HoMiStats

Statistics tools for host-microbiome dual transcriptomics

Installation

Bioconductor (preferred)

Pending Bioconductor approval HoMiStats can be installed vi:

if (!require("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install("HoMiStats")

GitHub installation (not preferred)

To install HoMiStats with the latest updates from GitHub:

devtools::install_github("sterrettJD/HoMiStats")

Package capabilities

Statistical modules

HoMiStats consists of two main modules:

1. Metatranscriptomics differential expression (mtxDE implemented in run_mtxDE())

   • Zero inflated beta regression, motivated by this review and using a model inspired by Peng et al., 2016
     • Implemented via GAMLSS BEZI and ZIBR (supports random effects)
   • Linear regression and linear mixed-effects regression

2. Construction of correlation network between host and microbial transcripts (HoMiCorr implemented in run_HoMiCorr())

   • Also supports both zero inflated beta regression (gamlss or ZIBR for random effects) and linear regression or linear mixed-effects regression

3. Targeted analysis of host-microbe crosstalk

   • GO_targeted_for_each_KO_within_GMM() allows targeted differential expression analysis, where only microbial genes belonging to a specific KO are tested for associations with host genes belonging to certain GO terms.
   • This improves power to detect crosstalk by only assessing biologically plausible interactions of interest.

Utilities

HoMiStats contains tools for transforming metatranscriptomic data, as well as tools for filtering genes based on low expression, prevalence, or variance.