Yeast-MetaTwin

Introduction

We designed a systematic workflow for mining underground metabolism, which combines rule-based retrosynthesis approach with deep learning-based enzyme annotation approach. Using this workflow, we constructed Yeast-MetaTwin, the first genome-scale metabolic model that systematically integrates underground networks, Yeast-MetaTwin encompasses 84% of the predicted metabolic enzymes and 92% of the metabolome in yeast.

Usage

• Download the Yeast-MetaTwin package

   git clone https://github.com/LiLabTsinghua/Yeast-MetaTwin.git
  

• Create and activate enviroment

   conda create -n  Yeast_MT python=3.7
   conda activate Yeast_MT
  

• Download required Python package

   conda install ipykernel
   pip install biopython
   pip install fair-esm==2.0.0
   pip install gurobipy  
   pip install matplotlib
   pip install numpy
   pip install pandas
   pip install plotly
   pip install pubchempy
   pip install rdchiral==1.1.0
   pip install rdkit-pypi==2022.9.5
   pip install rxnmapper==0.3.0
   pip install scikit-learn
   pip install seaborn
   pip install cobra        
   pip install torch==1.13.1
  

Reproducible Run

This project consists of four modules, which should be executed in the following order: the retrosynthesis must be run first, while the kcatkm_prediction and ECnumber_prediction can be executed as needed. Within each module, we have indicated the execution order in the filenames of the Jupyter notebooks.

• retrosynthesis: ./Code/retrosynthesis
• ECnumber_prediction: ./Code/ECnumber_prediction
• kcatkm_prediction: ./Code/kcatkm_prediction
• analysis: ./Code/analysis

The data generated from the retrosynthesis will be saved in ./Data_retrosynthesis, and the pre-trained protein model esm-1b required for deep learning will be stored in ./esm. Both of these resources can be found on Zenodo.

Please note that for the different prediction methods in the EC number prediction and kcat/km prediction modules, you need to set up the environment according to their respective GitHub sources. These projects are designed for user-friendly operation.

• ECnumber_prediction:
• CLEAN
• DeepECtransformer
• kcatkm_prediction:
• DLKcat
• UniKP
• TurNuP
• Boost_KM

Model Availability

The Yeast-MetaTwin (non-lipids) and Yeast-MetaTwin (non-lipids and lipids) models are available in ./Data/model.

Citation

Please cite the prpint paper Yeast-MetaTwin for Systematically Exploring Yeast Metabolism through Retrobiosynthesis and Deep Learning

Contact

• Feiran Li (@feiranl), Tsinghua Shenzhen International Graduate School, Tsinghua University, Shenzhen, China
• Ke Wu (@wuke), Tsinghua Shenzhen International Graduate School, Tsinghua University, Shenzhen, China

Last update: 2024-10-07