Abstract: Targeting pathological protein for degradation via the cellular ubiquitin-proteasome machinery offers an alternative to regular occupancy-based protein inhibition and opens the available target space to include targets previously considered ‘undruggable’. Induced degradation can be facilitated by stabilization of a complex consisting of a protein of interest (POI) and one of the >600 E3 ubiquitin ligases encoded by the human genome. Successful degradation further requires that the Ubiquitin ligase complex retain functional potential to modify targeted POI with ubiquitin chains hereby labelling it for degradation. Small molecule protein-protein interaction stabilizers such as molecular glues and proteolysis targeting chimeras (PROTACs) have shown great promise as hijackers of cellular E3 ubiquitin ligases to label POI for degradation. However, for various reasons the discovery of novel molecules with desirable properties has proven highly challenging. PROTACs rely on the identification of binders to an E3 ligase and a protein of interest (POI) and creating proximity by linking the two molecules. While these heterobifunctional molecules are relatively easier to identify from a drug discovery perspective, they have proven challenging to progress into drugs due to their large size and challenges in defining the optimal dose e.g. due to competition between the binary and ternary complexes, the so-called hook effect. In contrast, molecular glues are more difficult to discover but are easier to progress into drugs e.g. due to lower molecular weight and no hook effect. Correspondingly, of the relatively few molecular glues, originally discovered by serendipity, many derivatives have successfully been subjected to further development. However currently, no method for de novo discovery of molecular glue has been reported. We here present the first method for direct identification of molecular glue stabilizing the interaction between two proteins, an E3 ligase and POI, straight out of a 500-million-member DNA-encoded small molecule library upon screening inside a living cell. We envision the future success rate of molecular glue identification will increase as more of the >600 different E3 ligases are brought into play and as biological understanding and technology evolves.