Fig 1.
Schematic procedure of the database integration.
Fig 2.
Histogram showing the number of assay records for each compound.
The vertical axis representing the frequency is shown in logarithmic scale.
Fig 3.
Comparison between mean pIC50 values of 209 compounds measured by binding assays and electrostatic assays.
Fig 4.
Box plot showing the distribution of IC50 values of 263 compounds with more than three reported IC50 values.
The compounds were sorted by mean values.
Fig 5.
Step-wise procedure for compound classification.
(a) Classification by unanimous results using all assay values. Positive, unclassified, and negative compounds are shown in red, black, and blue, respectively. The dashed line indicates the threshold value between positive and negative compounds (IC50 = 10μM) (b) Distribution of assay values after removal of outlier values differing more than 10-fold from the mean IC50 values from undefined compounds. (c) Classification by a two-thirds majority of filtered IC50 values. The remaining unclassified compounds with nearly equal numbers of positive and negative assay results were considered as inconclusive compounds, and omitted from the subsequent analysis of physicochemical properties.
Table 1.
The number of compounds and their Murcko frameworks in each database.
Fig 6.
Transition in the number of unique compounds and coverage of Murcko frameworks for hERG inhibitors, inactive compounds, and all reported compounds.
Coverage of Murcko frameworks was calculated as the ratio to those of all ChEMBL22 compounds (438,551).
Fig 7.
Histogram of the number of compounds in each Murcko frameworks.
Fig 8.
Box plot of IC50 distribution about each Murcko framework.
(a) IC50 values for 82 Murcko frameworks containing more than 10 compounds sorted by the IC50 values of the most potent inhibitors. (b) Corresponding IC50 values reported before 2009. The horizontal axes of both plots represent each of the 82 Murcko frameworks.
Fig 9.
Six Murcko frameworks showing more than 10,000-fold potency differences.
Fig 10.
Distribution of 6 physicochemical properties showing significant differences.
The distributions of hERG inhibitors and inactive compounds are shown as red and blue lines, respectively.
Table 2.
Mean values (standard deviations) of 12 physicochemical properties for hERG inhibitors, inactive compounds, and all compounds in the integrated database.