Skip to content

Latest commit

 

History

History
84 lines (74 loc) · 4.13 KB

README.md

File metadata and controls

84 lines (74 loc) · 4.13 KB
Raw

ECMpy2.0

Automated construction of enzyme-constrained models using ECMpy workflow.

1. Create environment

$ conda create -n ECMpy2 python=3.7  
$ conda activate ECMpy2

2. Install JupyterLab

If you have any questions about installation and usage, please visit https://jupyter.org

$ pip install jupyterlab
$ pip install ipykernel 
$ python -m ipykernel install --user --name ECMpy2 --display-name "ECMpy2"

3. Install the relevant packages

Methond 1: Installation of ECMpy using pip

$ pip install ECMpy

Methond 2: Installation of ECMpy using github

Download ECMpy is via git clone:

$ git clone  https://github.com/tibbdc/ECMpy.git

Install package

$ pip install cobra==0.21.0  
$ pip install openpyxl  
$ pip install requests  
$ pip install pebble  
$ pip install xlsxwriter  
$ pip install Bio   
$ pip install Require  
$ pip install quest   
$ pip install scikit-learn  
$ pip install RDKit  
$ pip install seaborn 
$ pip install pubchempy
$ pip install torch
$ pip install bioservices==1.10.4
$ pip install pyprobar
$ pip install xmltodict
$ pip install plotly
$ pip install -U kaleido
$ pip install nbformat

4. Preprocessing data sources

The "all--radius2--ngram3--dim20--layer_gnn3--window11--layer_cnn3--layer_output3--lr1e-3--lr_decay0.5--decay_interval10--weight_decay1e-6--iteration50","atom_dict.pickle", "bond_dict.pickle", "edge_dict.pickle", 'fingerprint_dict.pickle", and "sequence_dict.pickle" files are derived from the DLKcat method, and you can update it from GitHub(https://github.com/SysBioChalmers/DLKcat.git).
The 'bigg_models_metabolites.txt" file is downloaded from BiGG (http://bigg.ucsd.edu/static/namespace/bigg_models_metabolites.txt).
The "brenda_2023_1.txt" file is downloaded from BRENDA (https://www.brenda-enzymes.org/brenda_download/file_download.php), and "EC_kcat_max.json" is obtained from this file extraction.
The "gene_abundance.csv" file is downloaded and transformed from PaxDB (https://pax-db.org/download).
The "uniprot_data_accession_key.json" is compiled from the UniProt database (only for Swiss-Prot), and we have uploaded to zenodo (https://zenodo.org/record/8119567/files/uniprot_data_accession_key.json?download=1).
The "AutoPACMEN_function.py" file is downloaded and modified from the AutoPACMEN method (https://github.com/klamt-lab/autopacmen.git).

5. Documentation

Full documentation is available at https://ecmpy.readthedocs.io/en/latest/.

Detailed process for constructing enzyme-constrained Models.

  • 00.Model_preview.ipynb
    • Assessment of gene coverage (UniProt ID coverage), reaction coverage (EC number coverage excluding exchange reactions), and metabolite coverage (BiGG ID coverage).
  • 01.get_reactiion_kcat_using_DLKcat.ipynb
    • Using DLKcat for predicting enzyme kinetic parameters directly based on the sequence information of enzymes catalyzing reactions and substrate information.
  • 01.get_reaction_kcat_using_AutoPACMEN.ipynb
    • Employing the AutoPACMEN process for extracting enzyme kinetic parameter information from the BRENDA and SABIO-RK databases.
  • 02.get_ecModel_using_ECMpy.ipynb
    • Using the ECMpy process to construct ecGEM.
  • 03.ecModel_calibration.ipynb
    • An automated parameter calibration process for the ecModel, guided by the principle of enzyme utilization.
  • 04.ecModel_analysis.ipynb
    • Some analysis cases of ecModels.
  • 05.ecModel_ME.ipynbP
    • Predicting metabolic engineering targets using ecModels.
  • 06.One-click_modeling.ipynb
    • Constructing ecGEMs with a one-click approach through the command line.

6. Acknowledgement

Here we are deeply grateful to klamt-lab for releasing the code for AutoPACMEN (https://github.com/klamt-lab/autopacmen) and to SysBioChalmers for sharing the code for DLKcat (https://github.com/SysBioChalmers/DLKcat), which enables ECMpy2.0 to rapidly obtain enzyme kinetics parameter information for the corresponding models. We extend our heartfelt thanks to qLSLab for making the code for GPRuler available (https://github.com/qLSLab/GPRuler), as it has inspired ideas for ECMpy2.0 to automatically acquire the subunit composition of proteins.