Table 1.
Method names, category, and submission type for all the log P calculation submissions
| Method name | Category | Submission type |
|---|---|---|
| ClassicalGSG DB2 [84–86] | Empirical | Blind |
| TFE MLR [87] | Empirical | Blind |
| ClassicalGSG DB4 [84–86] | Empirical | Blind |
| Chemprop [88] | Empirical | Blind |
| TFE-SM8-vacuum-opt | Physical (QM) | Blind |
| GROVER | Empirical | Blind |
| ClassicalGSG DB1 [84–86] | Empirical | Blind |
| ffsampled_deeplearning_cl1 | Empirical | Blind |
| ClassicalGSG DB3 [84–86] | Empirical | Blind |
| COSMO-RS [89] | Physical (QM) | Blind |
| TFE_Attentive_FP | Empirical | Blind |
| ffsampled_deeplearning_cl2 | Empirical | Blind |
| TFE-SM12-vacuum-opt | Physical (QM) | Blind |
| TFE-SM8-solvent-opt | Physical (QM) | Blind |
| REF1 ChemAxon [80] | Empirical | Reference |
| TFE IEFPCM MST [90] | Physical (QM) | Blind |
| TFE MD neat oct (GAFF/TIP4P) | Physical (MM) | Blind |
| NULL0 mean clogP FDA [38] | Empirical | Reference |
| NES-1 (GAFF2/OPC3) G | Physical (MM) | Blind |
| NES-1 (GAFF2/OPC3) J | Physical (MM) | Blind |
| NES-1 (GAFF2/OPC3) B | Physical (MM) | Blind |
| MD (GAFF/TIP3P) [91] | Physical (MM) | Blind |
| TFE wet oct (GAFF/TIP4P) | Physical (MM) | Blind |
| MD (CGenFF/TIP3P) [91] | Physical (MM) | Blind |
| EC_RISM_wet [92] | Physical (QM) | Blind |
| TFE-SMD-vacuum-opt | Physical (QM) | Blind |
| MD-EE-MCC (GAFF-TIP4P-Ew) [93] | Physical (MM) | Blind |
| TFE b3lypd3 [94] | Physical (QM) | Blind |
| MD (OPLS-AA/TIP4P) [91] | Physical (MM) | Blind |
| MD LigParGen (OPLS-AA/TIP4P) [91] | Physical (MM) | Blind |
| TFE-SMD-solvent-opt | Physical (QM) | Blind |
| TFE-NHLBI-TZVP-QM | Physical (QM) | Blind |
| Ensemble EPI physprop | Empirical | Blind |
| Ensemble Martel | Empirical | Blind |
| QSPR_Mordred2D_TPOT_AutoML | Empirical | Blind |
| TFE-NHLBI-NN-IN | Empirical | Blind |
The “submission type” column indicates if submission was a blind submission (denoted by “Blind”) or a post-deadline reference or null calculation (denoted by “Reference”). The table is ordered from lowest to highest RMSE, although many consecutively listed methods are statistically indistinguishable. All calculated error statistics are available in Table S1