Variational Autoencoders for Anti-cancer Drug Response Prediction

Welcome to my first cancer machine learning project. It's my second years' undergraduate research. Here's my teammates: Yuan and Varus and Dexin, who are all undergradaute at their third, second and first year. We are supervised by Prof. Manolis Kellis from MIT CSAIL LAB. Our paper link:(https://arxiv.org/pdf/2008.09763.pdf), accepted by 2021 ICLR AI4PH workshop (paper 33)

Check your environment

You should have a pytorch version of 1.5.0 or above. Also you should have a cuda version of 9 or above.

The following order in the command line can download the newest version of pytorch and torchvision with cuda version of 10.2.

conda install pytorch torchvision cudatoolkit=10.2 -c pytorch

In /Code/CancerML/mlps_drug_exp env, run train.py to test our model on breast cancer dataset(dont need to change anything)

python train.py

run train_pan.py to test our model on pancancer datasets

python train_pan.py

Citation

If you use to cite our results in your research paper, please consider citing:

@misc{dong2021variational,
      title={Variational Autoencoder for Anti-Cancer Drug Response Prediction}, 
      author={Hongyuan Dong and Jiaqing Xie and Zhi Jing and Dexin Ren},
      year={2021},
      journals={AI4PH workshop, ICLR 2021},
}