VAE analysis using DNA methylation - Paper

This repository is adapted from the Tybalt repository developed by Way et al. This repository is a fork of the previously mentioned repo, but due to a technical error, it was uploaded to GitHub independently. Some of the description below is addapted from the original Way et al. repo.

NEED TO UPDATE ANALYSIS TO USE UMAP INSTEAD OF t-SNE

UMAP Repo, UMAP Paper

Need to create a "data" folder with data for analysis before beginning.

The current reference to "data" folder can be replaced with the "test_data" folder for a quick proof of principle to get the repository running.

How to use this repository

In order to use this repository, you need to complete a few steps to ensure we are keeping the code clean.

1. Create a fork of this repository for your own GitHub account. DO NOT edit the Christensen-Lab-Dartmouth repository directly.
2. Once you have created your own fork, edit the code and documentation that you would like to change.
3. When you want to commit your changes back to the main repository, create a pull request and then assign the review to AlexanderTitus initially.

Analysis Steps

1. Select top N most variable CpGs by Median Absolute Deviation (MAD) - 1-data_preparation.ipynb
2. Train the VAE model - 2-Model_training.ipynb
3. Conducte t-SNE dimensionality reduction - 3-tSNE.ipynb
4. Visualize the t-SNE results - 4-data_visualization.ipynb
5. More to come...