Lagrange-NG

Lagrange-NG computes ancestral biogeographic ranges using the Dispersion Extinction and Cladogenesis (DEC) model with optimized numerical methods for large-state analyses.

Key Features:

• DEC model implementation: Implements the Dispersion Extinction and Cladogenesis (DEC) model for ancestral range inference on phylogenetic trees.
• State-space scaling: Addresses the exponential increase in state count relative to the number of regions, enabling analyses of larger region sets.
• Matrix exponential acceleration: Uses a Krylov subspace method to compute large matrix exponentials and reduce the primary runtime bottleneck.
• Multithreading and performance: Provides multithreading with reported speedups up to 49× over the original implementation and up to 26× in single-threaded comparisons.
• Benchmark capacity: Demonstrates the capacity to analyze datasets with up to 12 regions in approximately 18 minutes.
• Range-comparison metric: Introduces a novel metric for assessing differences between range inferences on phylogenetic trees.

Scientific Applications:

• Ancestral biogeography inference: Infers ancestral ranges for taxa across geographic regions under the DEC model.
• Comparative phylogenetic analysis: Compares and validates alternative ancestral range inferences on phylogenetic trees using the introduced metric.

Methodology:

Uses the Dispersion Extinction and Cladogenesis (DEC) model, computes large matrix exponentials via a Krylov subspace method, and applies multithreading for performance scaling.