LigUnity training data

Dataset OverviewThis dataset contains protein-ligand binding affinity data organized into training and validation sets, along with supporting files for sequence alignment and temporal splitting experiments. The data is provided in both JSON and LMDB formats for efficient access.File StructureTraining Settrain_label.zip (20.01 MB) - Contains:train_label_blend_seq_full.json: Primary training labels with assay information (structure as described in previous documentation)train_label_pdbbind_seq.json: Additional training labels from PDBBindtrain_prot_all_blend.lmdb (1.65 GB) - Protein/pocket structures for trainingtrain_lig_all_blend.lmdb (3.79 GB) - Ligand structures for trainingValidation Setvalid_label_seq.json (207.89 KB) - Label file for validation setvalid_lig.lmdb (5.96 MB) - Ligand structures for validationvalid_prot.lmdb (5.96 MB) - Protein/pocket structures for validationSupporting Filescache/ - Directory for training cache files (speeds up data loading)align_res.zip (70.44 MB) - Sequence alignment files for training HGNNuniport80.clstr (1.84 MB) - Protein sequence clusters (80% similarity threshold)uniport40.clstr (1.77 MB) - Protein sequence clusters (40% similarity threshold)TIME.zip - Files for split-by-time experiments (move to test_datasets directory in code)