code-for-TurboPETase

code for TurboPETase paper

Cui YL et al. Deep learning-aided redesign of a hydrolase for near 100% PET depolymerization under industrially relevant conditions. 2023. In submission.

Install

conda create -n ml python=3.9
conda activate ml
conda install pytorch pytorch-cuda=11.8 -c pytorch -c nvidia
pip install x-transformers
pip install mlm-pytorch
pip install numpy pandas scipy seaborn matplotlib tqdm

Inference

see mlm_data_analysis.ipynb.

Train

python mlm.py

We also provide a colab notebook to train on your own sequence. It utilized the colabfold's MSA.

Note

The *_new.pt files have been updated to fit the newer version of x-transformers, BUT THEY HAVE NOT BEEN TESTED!

For scientific problems, please refer to wub_at_im.ac.cn or cuiyl_at_im.ac.cn (replace the at with @ before sending the email).
For code problems, please open a new issue.