MTOR-NLP

MTOR-NLP produced far more species-like strings than the human-curated pathway retained. MTOR-NLP recovered substantially more of the curated graph under the most relaxed matching strategy than under strict matching. MTOR-NLP's exact species-name precision was below 1%, yet those exact matches still covered 45.88% of curated species. MTOR-NLP used TEES to process full-text papers and convert extracted entities and events into SBML pathway representations. MTOR-NLP was derived from the papers cited by the curated mTOR map, with text successfully extracted from 501 papers.