Skip to main content

Table 4 Performance of different logP methods against the Martel dataset

From: JPlogP: an improved logP predictor trained using predicted data

Predictor Performance % Binned absolute errors
RMSE 0–0.5 0.5–1 1–1.5 1.5–2 ≥ 2
JPlogP-Coeff 1.04 39.7 31.5 14.7 7.3 6.6
JPlogP—library 1.05 38.6 32.4 14.9 7.4 6.7
LogP4Average 1.12 33.9 29.4 19.2 10.8 6.6
XlogP3-AA 1.16 33.0 27.9 20.7 10.5 8.1
AAM 1.18 34.2 27.6 17.5 11.7 8.9
Molinspiration 1.21 32.7 26.3 20.1 11.7 9.2
SlogP 1.24 32.4 27.6 19.8 10.7 9.5
ALogP (Vega) 1.24 34.4 27.4 17.5 9.8 10.9
Biobyte CLogP 1.24 32.0 31.3 16.8 9.1 10.9
XLogP2 1.28 31.8 24.5 19.9 11.6 12.2
AlogPS logP 1.29 25.3 28.3 20.1 16.3 10.0
ACD 1.39 33.1 25.2 17.5 10.6 13.6
KowWIN 1.42 32.0 23.2 18.2 11.9 14.7
Meylan (Vega) 1.66 25.7 23.2 17.1 14.3 19.7
Mannhold LogP 1.71 14.1 16.3 23.2 20.9 25.5
MLogP (Vega) 1.95 12.3 14.7 17.7 19.5 35.8
AlogP (CDK) 3.72 1.6 2.5 5.0 9.1 81.9