f-krause / rnadegformer

Fork of RNAdegformer for DAP: Explainable mRNA Degradation Prediction

RNAdegformer

Source code to reproduce results in the paper "RNAdegformer: Accurate Prediction of mRNA Degradation at Nucleotide Resolution with Deep Learning".

How to use the models

I also made a web app to use the models. Check it out at https://github.com/Shujun-He/RNAdegformer-Webapp

Home page

RNA degradation prediction

In this page you can predict RNA degradation at each nucleotide and visualize the attention weights of the RNAdegformer

Requirements

I included a file (environment.yml) to recreate the exact environment I used. Since I also use this environment for computer vision tasks, it includes some other packages as well. This should take around 10 minutes. After installing anaconda:

conda env create -f environment.yml

Then to activate the environment

conda activate torch

Additionally, you will need Nvidai Apex: https://github.com/NVIDIA/apex

git clone https://github.com/NVIDIA/apex
cd apex
pip install .

Also you need to install the Ranger optimizer

git clone https://github.com/lessw2020/Ranger-Deep-Learning-Optimizer
cd Ranger-Deep-Learning-Optimizer
pip install -e .

Repo file structure

The src folder includes all the code needed to reproduce results in the paper and the OpenVaccine competition. Additional instructions are in the folder

src/OpenVaccine includes all the code needed to run a ten-fold model for the openvaccine dataset

Datasets

OpenVaccine dataset

For original dataset, see https://www.kaggle.com/c/stanford-covid-vaccine/data

In addition to the secondary structure features given by Das Lab, I also generated additional secondary structure features at 2 temperatures with 6 biophysical packages (12x), for these features, see https://www.kaggle.com/shujun717/openvaccine-12x-dataset