Table 2.

Summary of datasets discussed.

Name	Description
Dataset0	SAR data points for all 44 targets in Bowes et al. (2012) which are available in ExCAPE-DB.
Dataset1	SAR data points for the 31 targets in Dataset0 for which there were at least 100 actives and 100 non-actives.
Dataset2	SAR data points for targets with least 10,000 non-actives.
Dataset3	SAR data points for targets which had less than 10,000 non-actives, thus the same as Dataset1 with Dataset2 excluded.
Dataset4	SAR points making up the external test, by extracting rows from ExCAPE-DB for a selected set of 1,000 compounds in DrugBank (All withdrawn, and randomly sampled approved, drugs, until reaching 1,000 drugs).

See also Figure 1 for a graphical overview of how each dataset was created.