tax2tree

tax2tree maps taxonomic group names from reference taxonomies onto de novo phylogenetic trees to improve taxonomic resolution and annotation of bacterial and archaeal sequence datasets.

Key Features:

• Taxonomy transfer methodology: Implements a "taxonomy-to-tree" approach that transfers group names from reference taxonomies such as Greengenes, NCBI, and cyanoDB onto the topology of de novo phylogenetic trees, enabling integration of large datasets (e.g., 408,315 sequences).
• Explicit rank information: Incorporates explicit rank designations from the NCBI taxonomy by prefixing group names with rank information to standardize rank assignment.
• Improved classification accuracy: Enhances taxonomic resolution, improving classification for 75% of sequences by one or more ranks relative to the original NCBI taxonomy.
• Candidate phyla assessment: Evaluates NCBI-defined candidate phyla and provides recommendations to consolidate redundantly named groups, including 34 identified consolidations.
• Pipeline components: Uses tree inference, jackknifing, and taxonomy transfer processes as core computational steps within the pipeline.

Scientific Applications:

• Large-scale microbial sequencing projects: Supports taxonomic annotation in megasequencing efforts such as the Human Microbiome Project and the Earth Microbiome Project.
• Marker gene and metagenomic surveys: Improves interpretation of marker gene and metagenomic surveys, particularly for under-classified environmental sequences.
• Microbial ecology and evolution: Provides an improved taxonomy framework to facilitate studies of bacterial and archaeal ecology and evolutionary relationships.

Methodology:

Computational steps include de novo tree inference and jackknifing, followed by a taxonomy-to-tree transfer of group names from Greengenes, NCBI, and cyanoDB onto the tree topology, incorporation of NCBI rank prefixes into group names, and evaluation of candidate phyla with consolidation recommendations; the approach was demonstrated on a dataset of 408,315 sequences.