DT-L dataset's Contact shell representation training csv

features_pKa_delta_crystal-2019-general-train-prot_Select-mut0-mut2-prot.csv The prepared contact shell feature_pka csv file of training dataset from DT-L dataset (derived from mutating PDBbind v.2019 general set). Specific methods for data filtering: Complexes with Vinardo score deviations from crystal structures more than 1 kcal/mol or ΔScorevina more than 3 kcal/mol were excluded from the DT-L dataset to increase data reliability. combined_pKa_1pseudo_5delta.csv Combining features_pKa_delta_crystal-2019-general-train-prot_Select-mut0-mut2-prot.csv (delta) with re-docking poses (pseudo) from PDBbind v.2019 general set. The re-docking poses The ratio of them is 5 delta to 1 pseudo. The poses generated from re-docking were classified based on their ligand RMSD relative to the native crystallographic pose as either positive docked poses (PDPs; RMSD ≤ 2.0 Å) or negative docked poses (NDPs; RMSD > 2.0 Å).