PSI-Lab/cDeepbind

cDeepBind

cDeepBind is a fast and accurate model for learning sequence and structure specificities of RNA binding proteins.

• Download poster at MLCB 2019
• Download pre-print on Biorxiv

Model architecture

Reproducing the training and evaluation workflow

Setup an environment that contains all the necessary packages first.

conda env create -f environment.yml
conda activate cdeepbind
python setup.py develop

To train a new model follow:

python encode_data.py -d 2013
python train_model.py

To run evaluations on a pre-trained model follow:

python evaluate_model.py

Results