Table 2 Average performance (squared Pearson correlation coefficient, R2, with respect to ΔGbind0) of 22 standard SFs and ML methods applied to PL-REX geometries of the PL-REX dataset
Target | R2, avg. ± st.dev. | Best R2 (best SF) | SFs with R2 > 0.50 |
|---|---|---|---|
01-CA2 | 0.12 ± 0.11 | 0.34 (Gold ASP) | 0 % |
02-HIV-PR | 0.13 ± 0.14 | 0.58 (ΔvinaRF20) | 5 % |
03-CK2 | 0.14 ± 0.14 | 0.47 (RF-score-VS) | 0 % |
04-AR | 0.53 ± 0.22 | 0.84 (Gold GS) | 73 % |
05-Cath-D | 0.51 ± 0.20 | 0.80 (PLANTS ChemPLP) | 64 % |
06-BACE1 | 0.42 ± 0.23 | 0.76 (X-Score) | 36 % |
07-JAK1 | 0.47 ± 0.22 | 0.77 (Gold GS) | 55 % |
08-Trypsin | 0.49 ± 0.22 | 0.85 (Gold ASP) | 55 % |
09-CDK2 | 0.15 ± 0.06 | 0.25 (GlideSP) | 0 % |
10-MMP12 | 0.16 ± 0.12 | 0.57 (NNScore2) | 5 % |
