EVEX

EVEX extracts and normalizes biomolecular events from biomedical literature to enable integration with structured biomolecular databases and support literature-based curation and analysis.

Key Features:

• Large-scale literature corpus: Processes a corpus of 21.9 million PubMed abstracts and 460 thousand full-text articles from PubMed Central to identify biomedical knowledge at scale.
• Normalization strategy: Identifies biological concepts and maps them to canonicalized symbols, unique gene/protein identifiers, and broad gene families to facilitate precise data integration.
• Event extraction and normalization: Combines two state-of-the-art text mining components previously validated in community-wide challenges to perform extraction and normalization of biomolecular events.
• Semantic and taxonomic coverage: Captures 40 million biomolecular events involving 76 million gene/protein mentions linked to 122 thousand distinct genes across 5032 species spanning the full taxonomic tree.
• Database linking: Associates extracted entities and events with established resources such as UniProt, KEGG, BioGRID, and NCBI for downstream integration.

Scientific Applications:

• Database and pathway curation: Supplies literature-derived biomolecular events to support manual and automated curation of databases and pathways.
• Knowledge summarization: Enables synthesis of literature evidence for genes, proteins, and interactions across large corpora.
• Genomics and proteomics studies: Provides normalized event and entity data that can be integrated with UniProt, KEGG, BioGRID, and NCBI resources to support analyses in genomics and proteomics.

Methodology:

Automated processing of 21.9 million PubMed abstracts and 460 thousand PubMed Central full-text articles, entity normalization to canonical symbols, gene/protein identifiers and gene families, and event extraction performed by two validated text mining components combined for normalization and event detection.