Metalink: Joint taxonomic and functional profiling of metagenomes and metatranscriptomes across resolution scales

Abstract

Metagenomic and metatranscriptomic analyses remain largely fragmented: taxonomic and functional profiling are performed separately, often at a single resolution (reads, contigs, or MAGs) yielding separate species or function lists rather than mechanistic insights. Here, we developed Metalink, a tool that unifies taxonomic and functional profiling into coupled outputs across all three resolution levels. Metalink is built on three innovations. First, we constructed a Last Common Function-based functional Hierarchy (LCFH) from ~424 million proteins, organizing KEGG Orthologs into a tree structure that enables LCA-like functional assignment. Second, a unified framework integrates the taxonomy of Genome Taxonomy Database (GTDB), the International Committee on Taxonomy of Viruses (ICTV), and NCBI for consistent classification across reads, contigs, and MAGs. Third, a multi-level integration strategy with overlap-based deduplication balances accuracy and completeness. At the read level, Metalink exceeds both HUMAnN3 and eggNOG-Mapper in functional assignment accuracy while running ~30-fold faster and 1,000-fold faster, respectively; at the contig level, its functional profiles remain highly consistent with eggNOG-Mapper. Multi-level integration further improves taxonomic resolution over single-level analyses and outperforms leading taxonomic profilers on the CAMI II benchmarks. Metalink's joint outputs move the analysis of metagenomes and metatranscriptomes from cataloging species and functions toward mechanistic inference of species-function-environment triadic associations. By enabling these integrated analyses across resolution scales, Metalink provides a general framework for linking community composition to its functional potential across diverse ecosystems.