FASTRAL

FASTRAL improves scalability and efficiency of species tree reconstruction from multi-gene phylogenomic datasets by reducing the constraint space used in quartet-based coalescent inference under the multi-locus coalescent model (MSC).

Key Features:

• Constraint-space construction: Implements a novel technique for constructing the constraint space that substantially reduces its size relative to ASTRAL.
• Statistical properties: Maintains statistical consistency under the multi-locus coalescent model (MSC).
• Computational complexity: Operates in polynomial time.
• Speed: Achieves up to approximately 800-fold speedup compared to ASTRAL without compromising reported accuracy.
• Accuracy under ILS: Matches or often surpasses ASTRAL in species tree topology accuracy, particularly under high incomplete lineage sorting (ILS).
• Empirical validation: Performance demonstrated on both biological and simulated datasets.

Scientific Applications:

• Species tree reconstruction: Estimating species trees from hundreds to thousands of gene trees under the MSC framework.
• Large-scale phylogenomic analyses: Enabling reconstruction of species trees for datasets with extensive numbers of genes and species.
• Analyses with high ILS: Improving topology accuracy in conditions with substantial incomplete lineage sorting.

Methodology:

Builds upon ASTRAL's dynamic programming approach, employs a novel constraint-space construction to reduce the constraint set, operates in polynomial time, and preserves statistical consistency under the MSC.