Source code and data used in the article "Integration of Deep Learning with Ramachandran Plot Molecular Dynamics Simulation for Genetic Variant Classification" (Tam et al. iScience, 2023).

This repository aims to apply a deep learning classification algorithm to determine deleterious variants from VUS based on features extracted from MD simulations using benign and pathogenic variants.

Web app: https://genemutation.fhs.um.edu.mo/DL-RP-MDS/

By: Benjamin Tam, Zixin Qin, Bojin Zhao, San Ming Wang, Chon Lok Lei

The code requires Python (3.6+ and was tested with 3.7.6) and the following dependencies: scikit-learn, tensorflow, keras-tuner, imbalanced-learn. Installing Tensorflow in Windows may require Microsoft Visual C++ Redistributable for Visual Studio 2015, 2017 and 2019. It also needs seaborn for plotting.

To install, navigate to the path where you downloaded this repo and run:

$ pip install --upgrade pip
$ pip install -e .

All takes a required argument -g or --gene to select the gene (TP53, MLH1, or MSH2) for analysis.

• dl-tune.py: Run hyperparameter tuning for DL-RP-MDS models.
• dl-kfold.py: Run k-fold validation for DL-RP-MDS models.
• dl-pred.py: Run DL-RP-MDS predictions.

• data: Contains MD simulation data and labels.
• method: A module containing useful methods, functions, and helper classes for this project; for further details, see here.
• out: Output of the models.

If you publish any work based on the contents of this repository please cite (CITATION file):

Tam et al. (2023). Integration of Deep Learning with Ramachandran Plot Molecular Dynamics Simulation for Genetic Variant Classification. iScience, accepted.