ORGANISMS

ORGANISMS performs taxonomy-aware retrieval of biomedical literature by identifying NCBI Taxonomy terms in PubMed abstracts to enable extraction of organism–related associations.

Key Features:

• Named Entity Recognition (NER): Dictionary-based NER implemented via SPECIES that targets NCBI Taxonomy terms in PubMed abstracts and identifies species and other taxa with high precision and recall.
• Performance: SPECIES achieves speed more than an order of magnitude faster than existing tools while maintaining comparable accuracy.
• Evaluation datasets: Performance was evaluated using a manually annotated gold-standard corpus and a newly annotated set of 800 abstracts.
• Corpus coverage: The evaluation corpus includes abstracts from journals representing diverse taxonomic categories, providing insights into detection difficulty across taxa.
• Automated association extraction: Automated text mining extracts organism–disease and organism–tissue associations from literature.
• Medline-scale tagging: Organism names are tagged across the entire Medline database to support large-scale taxonomy-aware literature analysis.
• Update frequency: The index is updated on a weekly basis.
• Software and resources: The SPECIES component is open source and associated with dictionary files and a manually annotated gold-standard corpus.

Scientific Applications:

• Taxonomy-aware literature retrieval: Enables retrieval of literature based on taxon mentions for taxonomy-focused searches in PubMed/Medline.
• Organism–disease association analysis: Facilitates extraction and exploration of organism–disease associations from biomedical abstracts.
• Organism–tissue association discovery: Supports discovery and analysis of organism–tissue relationships via automated text mining.
• Large-scale taxonomy studies: Supports large-scale tagging and comparative analyses across Medline for studies in taxonomy and related fields.
• Domain research support: Supports research areas such as genomics, epidemiology, and systems biology that require precise identification of biological entities in literature.

Methodology:

Dictionary-based named entity recognition via SPECIES targeting NCBI Taxonomy terms in PubMed abstracts, automated text mining to extract organism–disease and organism–tissue associations, and evaluation against a manually annotated gold-standard corpus and an annotated set of 800 abstracts.