Table 9.
Target | Algorithm | Split 1 Percent Error of Failure | Split 2 Percent Error of Failure | Split 3 Percent Error of Failure | Mean Percent Error of Failure |
---|---|---|---|---|---|
ALK | NBN | 45 | 45 | 50 | 47 |
RF | 30 | 40 | 40 | 37 | |
PNN | 20 | 30 | 50 | 33 | |
Aurora B | NBN | 30 | 30 | 35 | 32 |
RF | 20 | 20 | 25 | 22 | |
PNN | - | - | - | - | |
β-2 | NBN | 50 | 50 | 50 | 50 |
RF | 35 | 25 | 45 | 35 | |
PNN | 10 | 20 | 5 | 12 | |
c-Abl | NBN | 45 | 50 | 45 | 47 |
RF | 35 | 25 | 35 | 32 | |
PNN | 30 | 25 | 15 | 23 | |
Factor Xa | NBN | 35 | 45 | 45 | 42 |
RF | 30 | 30 | 30 | 30 | |
PNN | 20 | 10 | 30 | 20 | |
HIV Protease | NBN | 50 | 45 | 50 | 48 |
RF | 35 | 20 | 15 | 23 | |
PNN | 25 | 15 | 25 | 22 | |
JAK2 | NBN | 40 | 30 | 40 * | 35 |
RF | 40 | 30 | 35 | 35 | |
PNN | - | - | - | - | |
MEK1 | NBN | 40 | 30 | 40 | 37 |
RF | 30 | 35 | 25 | 30 | |
PNN | 5 | 5 | 25 ** | 12 | |
PARP1 | NBN | 5 | 45 | 50 | 33 |
RF | 40 | 25 | 40 | 35 | |
PNN | 25 ** | 5 | 25 ** | 18 | |
TYRO3 | NBN | 30 | 10 | 5 | 15 |
RF | 20 | 5 | 5 * | 10 | |
PNN | - | - | - | - |
* failed in the control step, therefore another random split was used. ** reparameterization shifted the point of failure.