Engineering Novel Enzyme Reactivities with Zero-Shot Predictors

This work is a contribution to the preprint: Substrate-Aware Zero-Shot Predictors for Non-Native Enzyme Activities

Repository Structure

• GCFID_post-processing
  Scripts for the post-processing of variant fitness data from gas chromatogeaphy flame ionization detector.
• ZS
  Contains all zero-shot predictors implemented in this work.
• envs
  Contains .yml files and requirement.txt files to run the scripts in this repo.
• simulated_annealing
  Contains the scripts to run the implementation of parallel-tempering simulated annealing for protein optimization.

Prerequisites

• Python 3.9+ (tested on 3.10)
• Conda for environment management

Installation

Clone the repository

git clone https://github.com/Lukasra00/Master-Thesis.git
cd Master-Thesis

Usage

Simulated Annealing Pipeline

Create and activate the simulated annealing venv by:

python -m venv SA_env
source SA_env/bin/activate
pip install -r envs/simulated_annealing_requirements.txt

The run parameters can be specified as in. simulated_annealing/LUT_example.json

The run is started by:

cd simulated_annealing
python simulated_annealing/reSA.py --LUT_json path/to/LUT_example.json

Zero-Shot Prediction Pipeline

Data and instructions for the individual ZS predictors are given in the thesis methods section. Run individual zs predictors by:

cd ZS.zs
python example_zs_predictor.py

GCFID Post-Processing

Run parameter are to be specified in the run_GCFID_postprocessing.json file.

cd GCFID_post-processing
python GCFID_Post-Processing.py --run_json run_GCFID_postprocessing.json

License

All rights reserved.

Contact

Lukas Radtke – radtkel@ethz.ch
Department of Biosystems Science and Engineering,
ETH Zurich

Division of Chemistry and Chemical Engineering,
California Institute of Technology