Data for Revisiting Deep Learning DTA Methods from the Dataset Viewpoint
Year: 2026
Type: dataset
Abstract: This archive contains the curated datasets, output files and the trained models stemming from the paper "" curated_dataset raw_datasets: Curated versions of BindingDB, containing protein (super)family info, unique ligand, protein and pair indices, pKd values and others data_for_training: .pth files containing ProtTrans protein embeddings and .txt files containing each dataset input file for the model In what follows: each folder contains the output(s) of the hyperparameter tuning, and two folders: tuned and not_tuned_lr_0.8, with the outputs the models of a 5-fold cross validation training with the hyperparameters tuned and not tuned, respectively. all_1chain_GraphDTA_3000_5_features all_1chain_GraphDTA_3000_10_features unique_entries_GraphDTA_3000_5_features unique_entries_GraphDTA_3000_10_features unique_entries_1chain_GraphDTA_3000_5_features unique_entries_1chain_GraphDTA_3000_10_features unique_entries_1chain_GraphDTA_1000_5_features unique_entries_1chain_GraphDTA_1000_10_features protT5_3000_5_features protT5_3000_10_features protT5_1000_5_features protT5_1000_10_features casterdta_bdb data_for_training: .pth files containing ProtTrans protein embeddings and .txt files containing each dataset input file for the model. In what follows: each folder contains the output(s) of the hyperparameter tuning, and two folders: tuned and not_tuned_lr_0.8, with the outputs the models of a 5-fold cross validation training with the hyperparameters tuned and not tuned, respectively. GraphDTA_5_features GraphDTA_10_features protT5_3000_5_features protT5_3000_10_features
Author: Cardenas, Gustavo
Institution: Ri.MED
Language: English
Subfield: Drug Discovery
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7202128561