Development of an LLM-augmented bioinformatics workflow for transcriptomics-based chemical grouping in support of EFSA risk assessment

Design, implementation and validation of an end-to-end, LLM-augmented bioinformatics workflow that, given a chemical substance, automatically discovers, downloads and reanalyses transcriptomics datasets from selected public repositories. The system will harmonise key metadata, generate standardised transcriptomic signatures, cluster chemicals based on inferred mode of action, and compare the resulting groups with existing EFSA Cumulative Assessment Groups (CAGs) to support regulatory risk assessment. All code developed will be open source.