I made this web application (inspired by https://github.com/fatihozturkh2o/explorna_wave ) using h2o's wave so that models in my paper (https://www.biorxiv.org/content/10.1101/2021.01.28.428629v1) can be used easily. In addition, if your computer is too slow to run the app, you can also use Kaggle notebooks (you get 40 free GPU hours per week) I created to use the models:
https://www.kaggle.com/shujun717/nucleic-transformer-promoter-inference
https://www.kaggle.com/shujun717/nucleic-transformer-virus-inference
Here you can classify DNA promoters and visualize the top kmers
Here you can classify DNA virus and visualize the top kmers
In this page you can predict RNA degradation at each nucleotide and visualize the attention weights of the Nucleic Transformer
Clone the repo to your local. Preferably, This should be run on a Linux Machine (I recommend Ubuntu).
Before everything make sure that wave is ready. For instructions on installation: https://wave.h2o.ai/docs/installation Use this version of wave: https://github.com/h2oai/wave/releases/tag/v0.10.0
Start the server with
cd $HOME/wave
./wavedThis needs to be running for the app to work. So leave the terminal with waved running and open a new one to do the rest of the setup
1. Open terminal in the cloned folder and run: make setup
2. Then run the bash file: for Ubuntu: bash Ubuntu_setup.sh , for Mac: bash MacOS_setup.sh
3. Now you are ready to run the app: ./venv/bin/python run.py
3. Go to http://localhost:10101/nucleictransformer in your browser.