Darling

Darling identifies frequent sentence-based associations between diseases and human-related biomedical entities in PubMed literature to support literature-based discovery and construction of biomedical entity association networks.

Key Features:

• Sentence-based association extraction: Identifies co-mentions of biomedical entities and diseases within individual sentences of PubMed abstracts.
• Named Entity Recognition (NER): Uses NER to detect human-related biomedical terms in PubMed disease-related articles.
• Database cross-referencing: Cross-references extracted entities with OMIM (Online Mendelian Inheritance in Man), DisGeNET, and the Human Phenotype Ontology (HPO).
• Entity coverage: Extracts and represents genes, proteins, chemicals, functions, tissues, diseases, environments, and phenotypes.
• Association network generation: Builds a biomedical entity association network where nodes correspond to entities and edges represent sentence co-mentions.
• Search modalities: Supports queries by identifiers, terms or entities, and free-text to retrieve annotated PubMed abstracts.
• Frequency detection: Detects and reports frequently co-mentioned entity associations across PubMed articles.
• Abstract annotation: Produces entity annotations within PubMed abstracts to provide contextual information for associations.

Scientific Applications:

• Disease–entity association discovery: Identify associations between diseases and genes, proteins, chemicals, and phenotypes from sentence-level co-mentions in PubMed.
• Hypothesis generation: Support generation of hypotheses based on recurring sentence-level associations found in the literature.
• Literature curation and integration: Assist curation of disease-related abstracts with entity annotations cross-referenced to OMIM, DisGeNET, and HPO.
• Network analysis of biomedical relationships: Enable analysis of interconnections among genes, proteins, chemicals, functions, tissues, diseases, environments, and phenotypes.

Methodology:

Applies Named Entity Recognition (NER) to disease-related PubMed abstracts to extract human biomedical terms and detect sentence-level co-mentions, cross-references entities with OMIM (Online Mendelian Inheritance in Man), DisGeNET, and the Human Phenotype Ontology (HPO), and constructs a biomedical entity association network with nodes representing genes, proteins, chemicals, functions, tissues, diseases, environments, and phenotypes.