Table 6 Information concerning the experimental conditions of the 20 hardest-to-predict compounds from AqSolDBc.

From: Will we ever be able to accurately predict solubility?

ID

CAS

LogS

Remark

Method

A-5961

40530-60-7

−9.22

N.C

Flask

A-2317

1229-55-6

−8.93

Valid

N.S

A-5817

65059-45-2

−8.27

N.C

Flask

A-5546

CID: 83010

−7.74

N.C

N.S

A-2282

520-27-4

−7.51

Valid

Flask

A-5104

131-53-3

−7.27

Valid

Flask

A-5996

72102-84-2

−6.49

Below LOD

Flask

A-2783

10043-11-5

−6.39

Valid

N.S

A-2664

18230-61-0

−6.25

N.C

N.S

A-2162

15305-07-4

−6.19

Valid

Column elution

A-2035

14324-55-1

−5.53

Unstable

Column elution

A-5480

1324-35-2

−4.45

N.C

Flask

A-3034

10010-67-0

−2.75

Self-buffering

N.S

A-2955

26339-90-2

−1.10

Valid

N.S

A-5444

78181-99-4

−0.80

Unstable

N.S

A-5410

70900-27-5

−0.44

Valid

Flask

A-5225

121-54-0

0.07

Valid

Flask

A-1890

15332-99-7

0.65

Unstable

QSAR

A-2918

63500-71-0

2.14

N.C

N.S

  1. The 20 hardest-to-predict compounds display the highest MAE over all models. Remarks accounting for non-valid conditions to our guidelines are specified. The first letter of the ID corresponds to the source of the entry (see Fig. 10). N.C: Non-Conclusive, N.S: Not Specified.