mrALDE: Multi-Reaction Active Learning-Assisted Directed Evolution

Code to reproduce the findings from our study titled "Machine Learning-Assisted Evolution of Broadly Functional Enzyme Libraries."

Code Organization:

• processing: analysis of epistasis and normalization of raw data
• data: datasets from each round and downstream processing/visualizations
• MALDE: source code for multi-objective optimization
• configs: config files for running multi-objective optimization
• output: output from multi-objective optimization, pairwise epistasis analysis

Multi-Objective Optimization

Create conda environment based on environment.yml

Command to evaluate the acquisition function and query new candidates (for example to reproduce the first round of our study):

• python execute.py --config_file "refpt0_initial.json"