pySiRC is a simple web application for predicting rate constant using machine learning models. The models used are: XGBoost, Random Forest and MultiLayer Perceptron (Neural network). It is possible to make predictions for oxidation reactions with the radicals OH and SO4 with just a few clicks. To access this application remotely use the link pysirc.com.br. Another way to use this application is to install and run it locally. PySiRC was developed using the python language and has some dependencies.
- RDkit: Draw molecules and convert smiles to fingerprints.
- Numpy: Create matrices and mathematical operations.
- Cirpy: Convert cas number to smiles.
- Streamlit: Python framework to creating dashboards.
- Pandas: Data manipulation.
- Seaborn: Plots based in matplotlib.
- Scikit-learn: Framework to perform ML models.
- XGBoost: Perform a XGBoost model.
The rdkit via pip (rdkit-pypi) is only available for linux (need glibc>=2.17) and macOS systems.
$ pip3 install streamlit numpy cirpy pandas seaborn scikit-learn XGBoost tqdm rdkit-pypi
Conda version recommended here
$ chmod +x Anaconda3-4.4.0-Linux-x86_64.sh
$ ./Anaconda3-4.4.0-Linux-x86_64.sh
Installing rdkit from conda-forge:
$ conda install -c conda-forge rdkit
Install all dependencies packages:
$ pip install streamlit numpy cirpy pandas seaborn scikit-learn XGBoost tqdm
$ sudo python3 apt install python3-rdkit
Install pip to set up all python dependencies
$ sudo python3 apt install python3-pip
Install all dependencies packages:
$ sudo pip3 install streamlit numpy cirpy pandas seaborn scikit-learn XGBoost tqdm
Download this repository, manually or via git:
$ git clone https://github.com/jeffrichardchemistry/pySiRC
Enter the pySiRC folder and run the command:
$ cd .../pySiRC
$ streamlit run pySiRC.py
That done, the application will open in the browser and the terminal will have the local ip for access to all devices connected on the same network