GePI

GePI mines biomedical literature using natural language processing to identify genes, proteins, and biomolecular interactions and to construct an up-to-date knowledge graph for large-scale analysis of molecular interaction networks.

Key Features:

• Natural Language Processing (NLP): Applies NLP to detect genes, biomolecular events, and interactions within scientific texts.
• Search and Querying: Supports tailored queries by specific genes or gene lists and employs a search engine for precise retrieval.
• Full-text Filtering: Enables contextual filtering constrained to sentences or paragraphs within full-text documents.
• Knowledge Graph Maintenance: Aggregates extracted interactions into a knowledge graph that is refreshed several times per week.
• Interaction Statistics and Visualizations: Generates computational interaction statistics and visualizations from extracted data.
• Exportable Interaction Tables: Produces downloadable Excel tables containing interaction pairs, involved molecular entities, authors' reported factual certainty, and supporting text snippets.

Scientific Applications:

• Interaction Monitoring: Tracks reported gene and protein interactions across the biomedical literature for temporal and cumulative analysis.
• Hypothesis Generation: Reveals candidate interactions and relationships to support formulation of molecular hypotheses.
• Experimental Validation Support: Provides literature-derived evidence and reported factual certainty to assist validation of experimental results.
• Network Exploration and Mechanism Discovery: Facilitates exploration of complex biological networks and molecular mechanisms relevant to biological processes and disease.

Methodology:

Applies NLP-based text mining to identify genes, biomolecular events, and interactions from full-text biomedical literature; supports full-text filters (sentence/paragraph) and gene-list queries; aggregates extracted interactions into a knowledge graph refreshed several times per week and outputs interaction statistics, visualizations, and Excel tables with entity pairs, text snippets, and authors' expressed factual certainty.