Fig. 4

Ligand analysis of binding sites and its impact on model performance. a The top 8 small molecules that bind to proteins: ATP, GTP, NAD⁺, NADP⁺, FAD, cAMP, S-Adenosyl methionine (SAM), and FMN. b Performance metrics (Recall, Precision, MCC, and AUROC) of CLAPE-SMB on different datasets including different types of small molecule binding sites, including (c) four main categories of small molecules: nucleotides and their derivatives, coenzymes and cofactors, amino acids, and carbohydrate. Furthermore, proportions of each amino acid at binding and non-binding sites in the ground truth data on different datasets: nucleotide and its derivatives (d), coenzyme and cofactors (e), amino acids (f), and carbohydrate (g)