Table 1 The iPPIs dataset.

From: New machine learning and physics-based scoring functions for drug discovery

Protein (short name)

Totala

Affinities (kcal mol−1)

Trainingb

Testc

Bcl2-like/BAX

10

−12.636d, −5.244e

7

3

Bromodomain2/Histone

2

−9.968, −8.561

2

0

Bromodomain4/Histone

11

−9.931, −6.145

9

2

K-Ras/SOS1

1

−4.712

1

0

MDM2-like/P53

20

−12.768, −6.737

14

6

Menin

1

−10.404

0

1

Xiap/Smac

7

−11.278, −5.378

6

1

E1/E2

1

−10.051

1

0

IL2/IL2R

1

−6.910

1

0

LEDGF/Integrase

4

−10.490, −6.676

2

2

ZipA/ftsZ

2

−6.685, −5.544

2

0

Total

60

 

45

15

  1. aTotal number of protein–ligand complexes in the dataset.
  2. bNumber of complexes in the training set.
  3. cNumber of complexes in the random test set.
  4. dBinding affinity of the strongest protein–ligand interactions.
  5. eBinding affinity of the weakest protein–ligand complex.