Ibrahim Imam

PETzyme: An AI Research Agent for PET-Degrading Enzyme Discovery Enzymes offer a promising route to recycling PET plastics, but identifying useful candidates remains challenging. Experimental results are scattered across publications, measured under different conditions, and often difficult to compare. Researchers must spend considerable time determining which findings are reliable and what experiments to perform next. PETzyme is an AI-powered scientific research assistant that transforms this fragmented evidence into actionable research insights. Powered by Meta Muse Spark 1.3, it orchestrates scientific tools to investigate experimental data from NREL and PAZy, explore protein sequences using ESM-2, and support experimental planning. Unlike a conventional chatbot that simply retrieves information, PETzyme executes scientific workflows. Its Python tools evaluate assay conditions, track measurement provenance, identify overlapping publications, qualify unit conversions, and flag scientifically unsupported comparisons. Muse coordinates the investigation while deterministic validators ground the results in actual experimental evidence. Researchers can ask follow-up questions, specify laboratory constraints, and receive evidence-backed suggestions for their next experiments, with uncertainties clearly identified. PETzyme's goal is to accelerate enzyme research by helping scientists move from scattered experimental data to more informed, testable decisions.