Fig. 9
From: Open-source QSAR models for pKa prediction using multiple machine learning approaches
Scatter plots showing the acidic and basic pKa predictions by OPERA, XGB, and DNN, compared to the averaged ChemAxon and ACD/Labs predictions for the benchmark list of chemicals. a Benchmark acidic pKa predictions vs OPERA acidic pKa predictions. b Benchmark basic pKa predictions vs OPERA basic pKa predictions. c Benchmark acidic pKa predictions vs XGB acidic pKa predictions. d Benchmark basic pKa predictions vs XGB basic pKa predictions. e Benchmark acidic pKa predictions vs DNN acidic pKa predictions. f Benchmark basic pKa predictions vs DNN basic pKa predictions. Values between parenthesis are the 95% confidence intervals based on a 5000-bootstrapping procedure