Scaling coalescent-based species tree inference to 100,000 taxa with STELAR-X.

STELAR-X infers species trees from 100,000 taxa and 1,000 gene trees in under 35 minutes, marking a new scale for coalescent-based phylogenetic methods.

• Why it matters: Accurately reconstructing large-scale species trees is computationally challenging, limiting existing methods to smaller datasets despite their statistical reliability, which hampers comprehensive phylogenomic studies.
• What they did: The authors developed STELAR-X, a triplet-based algorithm utilizing compact encoding, bipartition precomputation, and GPU parallelism, achieving asymptotically optimal memory complexity of O(nk) for large datasets.
• The result: STELAR-X demonstrated unprecedented scalability, analyzing massive datasets efficiently and producing results consistent with established phylogenies, enabling large-scale evolutionary insights previously out of reach.