Applications of Rule‐Based Methods to Data Mining of Polypharmacology Data Sets
Nathalie Jullian, Yannic Tognetti, Mohammad Parsa Afshar · Methods and principles in medicinal chemistry · 2013
Understanding the interaction of drugs with multiple targets has become an important aspect of drug discovery. The availability of “profiling” data is expanding exponentially, and the growing number of statistical analyses of large databases helps improve the activity of molecules, reduce their side effects, and also identify new indications for existing compounds. Here, we focus on the polypharmacology problem from the multiple objective lead optimization perspective and use KEM1, a rule-based data mining method, for the systematic analysis of the data and the generation of novel molecules with a desired multiple target profile. We present an application to the optimization of a series of s-1 receptor binders and discuss how detailed rule-based analysis of polypharmacology data sets will help decipher the complex relations between multiple parameters and endpoints, including infrequent associations that may be difficult to identify by standard numerical approaches.