Relational similarity-based graph contrastive learning for DTI prediction.

We believe that this is because when the sequence similarity between the proteins in the test set and those in the training set is low, there may be significant differences in their three-dimensional structures.

also in 13,097 other papers