SIRT2i_Predictor: A machine learning-based tool to facilitate the discovery of novel SIRT2 inhibitors
This repository contains code and accompanyed ML models described in this work:
Djokovic, N.; Rahnasto-Rilla, M.; Lougiakis, N.; Lahtela-Kakkonen, M.; Nikolic, K. SIRT2i_Predictor: A Machine Learning-Based Tool to Facilitate the Discovery of Novel SIRT2 Inhibitors. Pharmaceuticals 2023, 16 (1), 127. https://doi.org/10.3390/ph16010127.
Application is deployed on the Streamlit cloud for demo: SIRT2i_Predictor
The code is written in Python. Current installation instructions for SIRT2i_Predictor are tested on Linux/UNIX and Windows.
Desired way to run the application is within conda environment with installed dependences (requires prior installation of Anaconda, or Miniconda).
Installing the dependences:
1. conda create -n sirt2i_predictor python=3.9
2. conda activate sirt2i_predictor
3. conda install -c conda-forge rdkit
4. conda install -c conda-forge scikit-learn
5. python -m pip install tensorflow-cpu or conda install -c conda-forge tensorflow-cpu (Using tensorflow instead tensorflow-cpu works as well, but sometimes requires more memory);
6. conda install -c conda-forge mordred
7. conda install -c conda-forge xgboost
8. python -m pip install streamlit
To run the SIRT2i_Predictor, unzip the SIRT2i_Predictor and navigate to the SIRT2i_Predictor's
directory using the terminal with active sirt2i_predictor environment (step 2 from instructions).
Then run:
streamlit run ./SIRT2i_Predictor.py
SIRT2i_Predictor is then previewed in your default web browser.
Results of predictions are written to the ./results folder.
Video_preview.mp4
Good luck!