multi-dice

multi-dice performs comparative population genomic inference under hierarchical co-demographic models to test independent single-population size changes across multiple taxa using the aggregate site frequency spectrum (aSFS).

Key Features:

• Hierarchical co-demographic models: Implements hierarchical co-demographic models for direct testing of multi-taxa hypotheses and independent single-population size changes.
• Data compatibility: Accepts mitochondrial barcode datasets and reduced-genome polymorphism datasets including thousands of SNPs from RAD-seq and GBS.
• Aggregate site frequency spectrum (aSFS): Utilizes the aSFS to summarize population genomic variation across multiple taxa.
• Inferential frameworks: Supports approximate Bayesian computation (ABC), random forest, and composite likelihood optimization for parameter inference and model comparison.
• Simulation integration: Acts as a wrapper that integrates existing simulation software within the R environment to generate simulated datasets for inference.
• Co-demographic pulse controls and hyperpriors: Enforces minimal thresholds around co-demographic pulses and allows specification of flexible hyperprior distributions.
• User-defined model specification: Permits user customization of model parameters, thresholds, and hyperprior settings for tailored comparative analyses.

Scientific Applications:

• Comparative phylogeography: Tests community-scale phylogeographic hypotheses about synchronous or asynchronous demographic responses to environmental change.
• Population genomics: Infers demographic histories and population size changes across multiple taxa using mitochondrial and reduced-genome data.
• Multi-taxa evolutionary dynamics: Assesses evolutionary dynamics across species within a shared analytical framework using hierarchical models and the aSFS.

Methodology:

Integrates reduced-genome and mitochondrial data via the aggregate site frequency spectrum (aSFS); integrates external simulation software within R; applies approximate Bayesian computation, random forest, and composite likelihood optimization for inference; enforces minimal thresholds around co-demographic pulses and permits specification of flexible hyperprior distributions.