Integrating computational and mixture-based screening of combinatorial libraries

Mixture-based synthetic combinatorial library (MB-SCL) screening is a well-established experimental approach for rapidly retrieving structure-activity relationships (SAR) and identifying hits. Virtual screening is also a powerful approach that is increasingly being used in drug discovery programs and has a growing number of successful applications. However, limited efforts have been made to integrate both techniques. To this end, we combined experimental data from a MB-SCL of bicyclic guanidines screened against the κ-opioid receptor and molecular similarity methods. The activity data and similarity analyses were integrated in a biometric analysis-similarity map. Such a map allows the molecules to be categorized as actives, activity cliffs, low similarity to the reference compounds, or missed hits. A compound with IC(50) = 309 nM was found in the "missed hits" region, showing that active compounds can be retrieved from a MS-SCL via computational approaches. The strategy presented in this work is general and is envisioned as a general-purpose approach that can be applied to other MB-SCLs.