VeryFastTree

VeryFastTree accelerates phylogenetic tree inference for ultra-large sequence alignments by applying optimized parallelization and vectorization while preserving FastTree-2 methods and heuristics.

Key Features:

• Performance Optimization: Constructs phylogenetic trees from ultra-large alignments (e.g., a 330k alignment) in approximately 4.5 hours on a standard server, achieving ~7.8× speedup over sequential FastTree-2 and ~3.5× faster than the best prior parallel implementation.
• Deterministic Output: Produces consistent, deterministic results for identical input datasets across runs.
• Methodological Consistency with FastTree-2: Preserves the core phases, methods, and heuristics of FastTree-2 to maintain comparable topological accuracy.

Scientific Applications:

• Large-scale phylogenetics: Generation of phylogenetic trees from ultra-large multiple sequence alignments.
• Evolutionary biology: Rapid inference of evolutionary relationships among large numbers of sequences.
• Comparative genomics: Scalable tree inference for comparative analyses across extensive genomic datasets.

Methodology:

Optimizes computational strategies through parallelization and vectorization, distributing tasks across multiple processors and leveraging advanced hardware capabilities while retaining FastTree-2 core phases, methods, and heuristics.