SysBioChalmers / DLKcat

Deep learning and Bayesian approach applied to enzyme turnover number for the improvement of enzyme-constrained genome-scale metabolic models (ecGEMs) reconstruction

DLKcat

Introduction

The DLKcat toolbox is a Matlab/Python package for prediction of kcats and generation of the ecGEMs. The repo is divided into two parts: DeeplearningApproach and BayesianApproach. DeeplearningApproach supplies a deep-learning based prediction tool for kcat prediction, while BayesianApproach supplies an automatic Bayesian based pipeline to construct ecModels using the predicted kcats.

Usage

• Please check the instruction README file under these two section Bayesianapproach and DeeplearningApproach for reporducing all figures in the paper.

• For people who are interested in using the trained deep-learning model for their own kcat prediction, we supplied an example. please check usage for detailed information in the file DeeplearningApproach/README under the DeeplearningApproach.

  • input for the prediction is the Protein sequence and Substrate SMILES structure/Substrate name, please check the file in DeeplearningApproach/Code/example/input.tsv
  • output is the correponding kcat value

Citation

• Please cite the paper Deep learning-based kcat prediction enables improved enzyme-constrained model reconstruction""

Notes

We noticed there is a mismatch of reference list in Supplementary Table 2 of the publication, therefore we made an update for that. New supplementary Tables can be found here

Contact

• Feiran Li (@feiranl), Chalmers University of Technology, Gothenburg, Sweden
• Le Yuan (@le-yuan), Chalmers University of Technology, Gothenburg, Sweden

Last update: 2022-04-09