Performances of two predictive methods for individual and all P450s (CYP3A, CYP2C9, and CYP2D6 inhibitors)
| Performance | CYP3A | CYP2C9 | CYP2D6 | All CP450s | ||||
|---|---|---|---|---|---|---|---|---|
| Free [I] with HMM Ki, app | Total [I] with Plasma Ki, app | Free [I] with HMM Ki, app | Total [I] with Plasma Ki, app | Free [I] with HMM Ki, app | Total [I] with Plasma Ki, app | Free [I] with HMM Ki, app | Total [I] with Plasma Ki, app | |
| 2-Fold of observed | 26% (5/19)a | 84% (16/19) | 93% (14/15) | 93% (14/15) | 62% (18/29) | 90% (26/29) | 59% (37/63)b | 89% (56/63) |
| No interaction | N.A. | N.A. | 100% (3/3) | 100% (3/3) | 100% (3/3) | 100% (3/3) | 100% (6/6)c | 100% (6/6) |
| Weak interaction | 0% (0/2) | 100% (2/2) | 80% (4/5) | 80% (4/5) | 20% (2/10) | 40% (4/10) | 35% (6/17) | 59% (10/17) |
| Moderate interaction | 22% (2/9)d | 67% (6/9) | 33% (2/6) | 66% (4/6) | 10% (1/10) | 60% (6/10) | 20% (5/25) | 64% (16/25) |
| Strong interaction | 0% (0/8) | 63% (5/8) | 0% (0/1) | 0% (0/1) | 0% (0/6) | 33% (2/6) | 0% (0/15) | 47% (7/15) |
| GMFE | 0.51 | 0.86 | ||||||
| RMSE | 3.94 | 2.61 | ||||||
N.A., there is “no interaction” in 19 CYP3A clinical studies.
↵a 26% (5/19) means that there are total 19 clinical studies related to CYP3A inhibitors. This method was able to predict 5 of 19 within 2-fold of observed data.
↵b 59% (37/63) means that there are total 63 clinical studies related to CYP3A, CYP2C9, and CYP3D6 inhibitors. This method was able to predict 37 of 63 within 2-fold of observed data.
↵c 100% (6/6) means that there are 6 (of total 63) CYP3A clinical studies as “no interaction.” Among these 6 studies, this method was able to predict all clinical studies as no interactions.
↵d 22% (2/9) means that there are 9 (of total 19) CYP3A clinical studies as “moderate interaction.” Among these 9 studies, this method was able to predict 2 clinical studies as moderate interactions.