#29 · ML / AI methods

Curated databases structure microbial habitat, antigen, lysin and biosynthetic gene data

The dominant thread is curated, queryable resources that organize microbial sequence and functional annotation for reuse. Microntology provides a 148-term controlled vocabulary for microbial habitats and lifestyles, with annotations for over 300k public metagenomic datasets. Other entries are domain-specific: GUTAID catalogs 73 gut microbial antigens against nine experimentally supported autoimmune mechanisms, PhaLP 2.0 extends a phage lysin database with a SUBLYME pipeline for metagenomic discovery, and ARTS-DB 2.0 supports target-directed genome mining of biosynthetic gene clusters across bacteria and fungi. PaVarDB builds a missense variant database for Pseudomonas aeruginosa through a Nextflow/Docker pipeline, and SynVectorDB applies BGE-M3 embeddings with a three-level hierarchical classification to synthetic biology parts. What is new is the combination of standardized vocabularies, reproducible pipelines and searchable interfaces, which structures sequence-derived knowledge for reuse.