Each protein chain in training and testing dataset is represented using 4 lines:
1: Protein ID
2: Amino acid sequence (from Uniprot sequence)
3: The protein-binding annotations where 0 indicates non-protein-binding and 1 stands for protein-binding annotations
4: The other-binding (including DNA-binding, RNA-binding, and small ligand-binding) annotations where 0 indicates non-other-binding and 1 stands for other-binding annotations

Each protein chain in human proteome predictions is represented using 4 lines:
1: Protein ID
2: Amino acid sequence (from Uniprot sequence)
3: Putative protein-binding residues where 1 indicates protein-binding residue and 0 stands for non-protein-binding residue
4: Protein domain annotations where 0 indicates residue not in domain, upper case 'D' stands for domain with putative protein-binding residues, and lower case 'd' represents domain without putative protein-binding residues.