Systematic identification of target set-dependent activity cliffs.

Aim: Generating a knowledge base of new activity cliffs (ACs) defined on the basis of compound set-dependent potency distributions, also taking confirmed inactive compounds into account. Methodology: Different AC definitions, representations and search criteria were rationalized and applied. Data: For nearly 100 different target proteins, for which medicinal chemistry and biological screening data were available, target set-dependent ACs were identified. More than 20,000 target set-dependent ACs and associated information are made freely available. Limitations & next steps: As more compound data become available for new targets, the search for target set-dependent ACs, including confirmed inactive compounds will continue. Second-generation ACs will be subjected to systematic structure-activity relationship analysis. Lay abstract: In medicinal chemistry, activity cliffs can be rationalized just like geographical landscape features. Highly potent compounds occupy the top of mountains in activity landscapes and weakly potent compounds the surrounding valley floors. A small chemical change may be sufficient to push a highly potent compound off a steep cliff and transform it into a weakly potent or inactive valley floor compound. Hence, by studying activity cliffs, one can often understand which chemical modifications have large-magnitude effects on biological activity. We report a systematic search for a new type of activity cliffs using compound data from medicinal chemistry and biological screening.