AutoProteinEngine: LLM-Driven AutoML Framework for Protein Engineering

The AutoProteinEngine (AutoPE) is a novel agent framework that utilizes large language models (LLMs) to assist biologists lacking deep learning skills in executing protein engineering tasks via natural language. As outlined in a paper on arXiv (2411.04440), AutoPE combines LLMs with automated machine learning (AutoML) for model selection across protein sequence and graph modalities, automatic hyperparameter tuning, and seamless data retrieval from protein databases. This framework seeks to reduce the barriers to entry in protein engineering, a crucial area for biomedical applications that is typically resource-heavy and inefficient with traditional methods. Evaluated through two practical case studies, AutoPE shows promise in optimizing protein engineering workflows, enhancing access to sophisticated computational tools for the scientific community and potentially speeding up research in drug discovery and enzyme design.