BioEnzyme Designer: A Python CLI Tool for Computational Enzyme Engineering via ML-Based Catalytic Efficiency Prediction and Optimized Mutation Scoring

Abstract

BioEnzyme Designer is a Python-based command-line tool for computational enzyme engineering that combines machine-learning-based catalytic efficiency prediction with optimized mutation scoring. The workflow loads protein structures from local PDB files or retrieves them through an AlphaFold/RCSB fallback, identifies candidate catalytic residues using biochemical heuristics, and ranks point mutations according to physicochemical effects on activity, thermostability, and pH optimum. The pipeline is designed to be lightweight, reproducible, and usable without a graphical interface, making it suitable for researchers working in resource-limited environments. The model was trained on a synthetic dataset of 500 enzymatic variants and achieved an R² of 0.72 in 5-fold cross-validation. The tool generates multiple outputs, including an interactive 3D HTML viewer, a Plotly-based residue property visualization, a Matplotlib mutation score plot, and a structured text report. It also supports the analysis of three representative case studies: lysozyme, trypsin, and a hypothetical enzyme optimized through sequence-based structure retrieval. This release focuses on accessibility and practical use in enzyme design workflows. The project is fully free and available only on GitHub: BioEnzyme Designer v2.0. Planned enhancements include integration of experimental data from BRENDA and SABIO-RK, FoldX/Rosetta validation, multimeric enzyme support, an optional GUI, a REST API, and a cloud-hosted version.