The discovery of bioactive small molecules is an expensive and time consuming yet central task in drug discovery. A potentially superior way to identify new drug candidates is the selective optimization of side activities (SOSA), which employs known drugs for their side-activities as lead compounds, while diminishing their original activity. Virtually every small molecule drug interacts with more than one molecular target and, thus, has side-activities.

We have observed such side-activity for the cysteinyl leukotriene receptor 1 (CysLT₁R) antagonist cinalukast on the nuclear peroxisome proliferator-activated receptor α (PPARα). We chose this synthetically challenging experimental drug to study whether the application of well-established computational optimization and ranking methods can help identify the most promising variations for SOSA and reduce the synthetic efforts needed in this lead optimization concept. We employed SeeSAR to visualize the critical parts for supportive or undesirable interactions with the nuclear receptor. In a proof-of-concept study, we confirmed the suitability of the HYDE ranking for the task of compound prioritization concerning potency on PPARα and screened an automatically generated virtual library of approximately 8000 close cinalukast analogues using a self-designed KNIME^® workflow with FlexX and HYDE. The top-ranking molecules from this first aspect of SOSA were then computationally studied for CysLT₁R antagonism using a random forest model trained on fingerprint representations of known CysLT₁R antagonists. A computationally favoured cinalukast analogue was synthesized and its in vitro profiling confirmed the predicted activity shift towards higher activation efficacy on PPARα and markedly improved selectivity over CysLT₁R compared to the lead compound.