PipeMaster

PipeMaster implements simulation-based inference using approximate Bayesian computation (ABC) and supervised machine learning (SML) to infer population divergence, divergence time, effective population size, and migration rate from Sanger-type and Next-Generation sequencing data.

Key Features:

• Simulation Capabilities: Simulates summary statistics under the coalescent framework for multiple demographic models and supports Sanger-type and Next-Generation sequencing data, including single-locus data for hierarchical demographic models.
• Species Tree Simulation: Simulates species trees with one horizontal connection to model evolutionary relationships and gene flow events.
• Approximate Bayesian Computation (ABC): Compares observed data to simulated datasets to approximate posterior probabilities for hypothesis testing without explicit likelihood calculations.
• Supervised Machine Learning (SML): Employs SML for model selection and parameter estimation, demonstrated to require fewer simulations and often outperform ABC in computational efficiency and inference accuracy, particularly with large numbers of loci.

Scientific Applications:

• Muller’s Termite Frog (South America): Supported a divergence model without migration and indicated a recent population bottleneck in one population.
• Cottonmouth Snakes (North America): Provided evidence for a divergence scenario with ongoing migration and recent expansion.
• Copperhead Snakes (North America): Identified a model involving divergence with migration and a recent bottleneck event.

Methodology:

Simulates summary statistics under the coalescent framework, simulates species trees with one horizontal connection, and integrates approximate Bayesian computation (ABC) and supervised machine learning (SML) to compare observed and simulated datasets and estimate posterior probabilities.