Table 1.
Physicochemical properties and Assay Central lysosomotropic machine learning predictions for compounds tested in vitro. Larger prediction scores have a higher probability of activity. An applicability score of 1 indicates that all the fragments are in the model and may indicate the molecule is in the training set (chloroquine is in the training set) (Calculated with ACD/Labs PhysChem Batch program$ (Ploemen et al., 2004)). Predicted pka's (negative log of the acid dissociation constant) were obtained from DrugBank, which were initially calculated using Chemaxon. AlogP (predicted log octanol-water partition coefficient was calculated via Discovery Studio).
| Name | pKa (predicted) | Pka (Experimental) | AlogP | Lysosomotropic Prediction Score | Lysosomotropic Applicability Score |
|---|---|---|---|---|---|
| Chloroquine | 10.32 (Strongest Base) | 4.0, 8.4 and 10.2 (Schroeder and Gerber, 2014) | 4.34 | 1.09 | 1 |
| Artesunate | 3.77 (Strongest Acid), −4.2 (Strongest Base) | 4.6 (Augustijns et al., 1996) | 1.84 | 0.31 | 0.21 |
| Quinacrine | 10.33 (Strongest Base) | N/A | 5.67 | 1.00 | 0.68 |
| Tilorone | ~8.6$ | N/A | 4.56 | 0.75 | 0.69 |
| Pyronaridine | 7.96 (Strongest Acid), 10.08 (Strongest Base) | 7.08, 7.39, 9.88 and 10.30 (Adegoke et al., 2006) | 6.19 | 0.68 | 0.51 |