Gilda

Gilda grounds named entities in biomedical literature by scored string matching and machine-learned disambiguation to map mentions to identifiers in biomedical ontologies.

Key Features:

• Scored string matching algorithm: Employs a scored string matching algorithm to identify names and synonyms across ontology entries.
• Machine-learned disambiguation models: Uses machine-learned models that leverage contextual information from surrounding text to resolve ambiguous entity names.
• Species-prioritization: Applies species-prioritization to refine disambiguation based on biological context.
• Ontology coverage: Grounds entities across biomedical ontologies covering genes, proteins (including families and complexes), small molecules, biological processes, and diseases.
• Identifier selection: Selects the correct identifiers when multiple potential matches exist for a given string.

Scientific Applications:

• Biomedical text mining: Automates grounding of entity mentions to support large-scale text mining and annotation of literature.
• Data extraction and curation: Facilitates reliable extraction and annotation of entities from high volumes of publications where manual curation is impractical.
• Normalization for analysis: Provides normalized identifiers for genes, proteins, small molecules, biological processes, and diseases to enable downstream data integration and analysis.

Methodology:

Implements a scored string matching algorithm combined with machine-learned disambiguation models that leverage surrounding text and species-prioritization to map mentions to ontology identifiers.