ASTRAL

ASTRAL estimates species trees from unrooted gene trees under the multi-species coalescent to account for incomplete lineage sorting and gene tree discordance.

Key Features:

• Statistical Consistency: Grounded in the multi-species coalescent model, providing statistically consistent species tree estimates in the presence of incomplete lineage sorting (ILS).
• Scalability and Efficiency: ASTRAL-II increased processing speed enabling analysis of datasets up to 1,000 species and genes, and ASTRAL-III further optimizes running time via dynamic programming while constraining bipartition growth to be linear in n and k to ensure polynomial-time performance.
• Handling Gene Tree Discordance: Focuses on maximizing agreement of quartet topologies between input unrooted gene trees and the species tree to manage discordant evolutionary histories across genomes.
• Polytomy Management: ASTRAL-III implements polytomy handling that leverages similarities among gene trees to reduce unnecessary search space.
• Noise Reduction through Branch Contraction: Supports contracting low-support branches in gene trees (e.g., below 10%), which reduces noise and can improve accuracy in large phylogenomic datasets such as avian analyses with 14,000 genes.

Scientific Applications:

• Phylogenetic Reconstruction: Reconstructing species trees for evolutionary inference, biodiversity studies, and gene–species co-evolution analyses under the multi-species coalescent.
• Genome-Scale Analyses: Analyzing genome-scale datasets where gene tree discordance and computational scaling are primary challenges.

Methodology:

Uses dynamic programming to search for the species tree that maximizes shared quartet topologies with input gene trees, constrains the search using a set of allowed bipartitions, optionally contracts low-support branches in gene trees, and in ASTRAL-III restricts bipartition constraints to grow linearly with n and k while providing polytomy handling to improve efficiency.