Fig. 2: Machine learning methods performance evaluation.

A Summary of the test performance of the ΔΔG prediction across machine learning methods under three training scenarios in terms of RMSE, Pearson correlation, and AUPRC. Means with error bars (standard deviation) for all competing methods. B Scatter plots of the experimental versus calculated ΔΔG values in Scenario 3. X-axis denotes the experimental ΔΔG values (kcal mol⁻¹). Y-axis denotes the calculated ΔΔG value (kcal mol⁻¹). Each ΔΔG estimate is color-coded according to its absolute error w. r. t. the experimental ΔΔG value; at 300 K, the 1.4 kcal mol⁻¹ error corresponds to a 10-fold error in the K_d change and 2.8 kcal mol⁻¹ error corresponds to a 100-fold error in the K_d change.