raw/README.md

Description of the structural data

Unzip the *.tar.gz dataset with command tar -xzvf hiqbind.tar.gz will get two directories corresponding to to the "small molecule" and "polymer" subset of HiQBind.

-- raw_data_hiq_sm/
-- raw_data_hiq_poly/

In each of the directory, you will see a file structure like this:

-- 1a69/
   |-- 1a69_FMB_A_240/
       |-- 1a69_FMB_A_240_ligand.pdb
       |-- 1a69_FMB_A_240_protein.pdb
       |-- 1a69_FMB_A_240_protein_hetatm.pdb
       |-- 1a69_FMB_A_240_hetatm.pdb
       |-- 1a69_FMB_A_240_ligand_refined.sdf
       |-- 1a69_FMB_A_240_protein_refined.pdb
   |-- 1a4m_FMB_B_240/
   |-- 1a4m_FMB_C_240/
-- 1a85/

Description of the naming conventions:

• 1a69: 4-letter PDB ID
• FMB: Name of the ligand. If the ligand is a polymer, it will be format like "ACE-DIP", where "ACE" is the name of the first residue and "DIP" is the name of the last residue.
• A: Ligand chain ID.
• 240: Ligand residue number. If the ligand is a polymer, it will be format like "1-3", where "1" is the residue number of the first residue and "3" is the number of the last residue. Note the residue number may contain insertion code, or be a negative integer or zero.

Description of the files:

• *_ligand.pdb: ligand structure extracted from the original PDB (not processed)
• *_protein.pdb: protein structure extracted from the original PDB (not processed). A protein is defined as chains within 10 angstrom of the ligand structure.
• *_protein_hetatm.pdb: protein structure with additives (solvents, ions) extracted from the original PDB (not processed). Additives are specified with "HETATM" atoms that are within 4 angstroms of the protein chains.
• *_hetatm.pdb: additives' structure extracted from the original PDB (not processed)
• *_ligand_refined.sdf: refined ligand structures (hydrogen added, correct bond order, better tautomer states/protonataion states) with PDBBind-Opt workflow.
• *_protein_refined.pdb: refined protein structures (hydrogen added, missing atoms/residues added) with PDBBind-Opt workflow.

Description of columns in the metadata csv file:

• PDBID: string, 4-letter PDB code
• Resolution: string or float, resolution of the crystal structure or "NMR" if the structure is resolved by NMR
• Year: int, initial deposit year in PDB database
• Ligand Name: string, name of the ligand. If the ligand is a polymer, it will be format like "ACE-DIP", where "ACE" is the name of the first residue and "DIP" is the name of the last residue.
• Ligand Chain: string, chain of the ligand.
• Ligand Residue Number: string, residue number of the ligand. If the ligand is a polymer, it will be format like "1-3", where "1" is the residue number of the first residue and "3" is the number of the last residue. Note the residue number may contain insertion code, or be a negative integer or zero.
• Binding Affinity Measurement: string, "kd", "ki" or "ic50". Note in some sources, binding data is labeled to be "Ka" or "Kb", they are converted to Kd using Ka = 1/Kd.
• Binding Affinity Sign: string, could be "=", ">=", "<=" or "~".
• Binding Affinity Value: float, value of the binding affinity
• Binding Affinity Unit: string, coule be "fM", "pM", "nM", "uM", "mM", "M"
• Log Binding Affinity: float, binding affinity in log unit
• Binding Affinity Source: string, could be "PDBBind", "BindingMOAD", "BindingDB" or "BioLiP"
• Binding Affinity Annotation: string, the annotation in the original source.
• Protein UniProtID: string, UniProtID of the proteins, seperated by a comma if the ligand bound to more than one chain.
• Protein UniProtName: string, Name of the proteins, separated by a comma if the ligand bound to more than one chain.
• Ligand SMILES: string, SMILES of the ligand.
• Ligand MW: float, Molecular weight of the ligand.
• Ligand LogP: float, LogP value of the ligand computed by RDKit.
• Ligand TPSA: float, TPSA value of the ligand computed by RDKit.
• Ligand NumRotBond: int, Number of rotatable bonds in the ligand.
• Ligand NumHeavyAtoms: int, Number of heavy atoms in the ligand.
• Ligand NumHDon: int, Number of hydrogen bond donors in the ligand.
• Ligand NumHAcc: int, Number of hydrogen bond acceptors in the ligand.
• Ligand QED: float, QED value of the ligand computed by RDKit.