BELL: Biomodel Evidence and LLM-based Logic

Building accurate and predictive mechanistic models requires careful biological interaction curation and verification against existing knowledge. When done manually, these tasks become impractical, especially with massive extraction of interactions facilitated by advanced natural language processing methods and large language models (LLMs). We present BELL (Biomodel Evidence and LLM-based Logic), a biocuration support framework that automates evidence retrieval, scoring, and explanation for interaction-level verification. BELL processes each interaction through a five-step pipeline: entity grounding, database ranking, evidence retrieval from seven biological databases, a heuristic four-dimension programmatic scoring, and chain-of-thought explanation with a recommended curator action generated by LLMs. We applied BELL on 210 protein-protein interactions from a curated Glioblastoma Multiforme (GBM) model. Results show that 49.5% of interactions achieved HIGH confidence and 72.4% received a positive curator recommendation, while qualitative flags precisely directed curator attention to evidence gaps. BELL is integrated into the KALIMBA curation platform and available at www.boheme.pitt.edu/Kalimba.