git clone git@github.com:Sultans0fSwing/MFE.git
cd MFE
conda env create -f environment.yaml
We use the PDBbind dataset (version 2016) as the dataset for the protein-ligand binding affinity prediction task. The official original dataset can be downloaded from http://pdbbind.org.cn/. Then, place the downloaded original dataset in the data directory.
Before starting the training, please use process.py to transform the raw data into the format required by the model. Remember, you need to specify the directories for the training set, validation set, and test set yourself.
After processing the dataset, please use train_lba.py for training. Set the number of training epochs to 50, the learning rate to 0.0001, and fix the regularization coefficient at 5e-4. The output of the model can be found in the output directory.