The correlation analysis between the real affinity values and predictive values of S2DTA, DeepDTAF, and MM-GBSA/PBSA. The results show that the real affinity values of the selected samples exhibited a strong linear relationship with the predictive affinity values of both S2DTA (PCC = 0.969, R2 = 0.941) and DeepDTAF (PCC = 0.921, R2 = 0.847). Notably, S2DTA showed slightly superior performance compared to DeepDTAF. In contrast, we observed that the real affinity values of the selected samples did not align well with the free energy predicted by MM-GBSA/PBSA (PCC = 0.597, R2 = 0.356) in terms of a linear relationship, and its predictive values exhibited a weak correlation with the real affinity values.