NeuRank: Learning to Ranking with Neural Networks for Drug-Target Interaction Prediction

主要库版本： tensorflow 1.14.0

• Run NeuRank:

$ python NeuRank.py --path datasets/ --data_name Enzyme --epoches 40 --batch_size 64 --num_factors 64 --layers [32,16] --reg 0.00001 --num_neg 4 --lr 0.001 --min_loss 0.01 --cv 10 --mode 0

• Run NeuRanks:

$ python NeuRanks.py --path datasets/ --data_name Enzyme --epoches 100 --batch_size 64 --num_factors 64 --layers [32,16] --reg [0.00001,0.000001,0.000001] --num_neg 4 --lr 0.001 --min_loss 0.01 --cv 10 --mode 0

• Run pNeuRank:

$ python pNeuRank.py --path datasets/ --data_name Enzyme --epoches 40 --batch_size 64 --num_factors 64 --layers [32] --reg 0.00001 --num_neg 4 --lr 0.001 --min_loss 0.1 --cv 10 --mode 0

• path：Input data path.
• data_name：Name of dataset
• epoches：Number of epoches.
• batch_size：Batch size.
• num_factors：Embedding size.
• layers：Size of each layer. Note that the first hidden layer is the interaction layer.
• reg: Regularization for user and item embeddings.
• num_neg: Number of negative instances to pair with a positive instance.
• lr: Learning rate.
• min_loss: The minimum value for stopping loss function.
• cv: K-fold Cross Validation.
• mode: the mode for training: 0 -> train for drug-target pairs; 1 -> train for new drugs; 2 -> train for new target

Homepage: http://zhouxiuze.com

个人博客： http://snailwish.com/