This is an official implementation of "RAMP: Response-Aware Multi-task Learning with Contrastive Regularization for Cancer Drug Response Prediction" by Kanggeun Lee, Dongbin Cho, Jinho Jang, Hyoung-oh Jeong, Jiwon Seo, Won-Ki Jeong and Semin Lee.
-
The environment settings for RA-NS are described in here.
-
Building docker images to test 10-fold nested cross validation using extracted embedding vectors.
-
Choose your environment either tensorflow or pytorch.
cd src/response_prediction/docker/tensorflow
./build.sh
or
cd src/response_prediction/docker/pytorch
./build.sh
-
You should ensure that a global "PATH" variable have to be modified to your environment path in demo_tf.py and demo_torch.py
-
All experiments in the manuscript were done in Tensorflow.
cd src/response_prediction/scripts
./demo_tf.sh or ./demo_torch.sh