pubmedKB

pubmedKB extracts semantic relationships among genetic variants, genes, diseases, and chemicals from PubMed abstracts to support literature-scale biomedical knowledge discovery.

Key Features:

• Semantic Relationship Extraction: Extracts semantic relations among genetic variants, genes, diseases, and chemicals from the PubMed corpus.
• Corpus-scale Coverage: Processes over 33 million PubMed abstracts and reports more than 2 million identified semantic relations between entity pairs.
• Supporting Sentence and Snippet Extraction: Identifies supporting sentences and text spans (snippets) from abstracts that justify extracted relations.
• Literature-scale Mining and Indexing: Performs large-scale literature mining to index and enable retrieval of relation-level evidence across PubMed abstracts.

Scientific Applications:

• Genetic research: Facilitate identification and interpretation of variant–gene and variant–disease associations across the literature.
• Disease mechanism exploration: Support elucidation of relationships among genes, variants, chemicals, and diseases for mechanistic hypotheses.
• Drug discovery: Enable literature-based linking of chemicals to genes and diseases to inform target and compound prioritization.
• Personalized medicine: Assist aggregation of variant–disease evidence to inform genotype–phenotype interpretation and translational research.

Methodology:

Applies advanced natural language processing (NLP) to process and interpret over 33 million PubMed abstracts, automating extraction of semantic relations among genetic variants, genes, diseases, and chemicals and extracting supporting sentences/snippets, yielding more than 2 million relations.