Figure 2

Interpretation of compound substructures. A. Predictive method. The average influence on bioactivity of a given substructure is calculated as the difference between the distributions corresponding to: (i) the predicted bioactivity for all compounds containing that substructure, and (ii) the predicted bioactivity using PCM for these compounds, from which that substructure was virtually removed by setting its count to zero. B. Student’s method. In this case, the average substructure influence on bioactivity is evaluated as the difference between the mean values of the pIC₅₀ distributions for those compounds presenting and not presenting a given substructure. The statistical significance of this difference was evaluated with a Student’s t-test.