Defining a unified interface to implement simulations of microbiome relative abundance data, based on previous simulation designs. The goal of this package is to:
- Enable quick simulations for testing different statistical packages from an implementation standpoint
- Collet implementations of popular simulation designs
- Modular design to extend simulations for methodological research
Count distribution models
- Dirichlet-Multinomial distributions
- Gamma-Poisson distributions (used in ANCOM and ANCOMBC)
- Negative Binomial and Zero Inflated Negative Binomial distributions
Simulation tasks
- Differential Abundance
- Simple inflated counts (McMurdie & Holmes)
- Inflated counts with compensation (Hawinkle et al. & Weiss et al.)
- Simple inflated means
- Clustering
- Tree-based structures (Chen et al. 2013)
- Simple clustering
- Prediction
- Regression
- Binary classification
Additional considerations
- Flexible correlation structures
- Standard structures (independence, exchangable, AR1, unstructured)
- Tree-based correlation structures
- Network-baesd correlation structures
- Provenance tracking and reproducibility