A Baylor College of Medicine-led team combined high-throughput proteomics and AI-based structure prediction to discover and optimize molecular glue candidates designed to degrade VAV1, an immune signaling protein linked to blood cancers and autoimmune diseases. The research, published in Nature Communications, introduced GluePlex, a computational workflow integrating AI protein folding with physics-based modeling to predict ternary complex assembly for VAV1 degradation without relying on an experimentally determined structure of the complex. In experiments, the team screened a library of molecules using proteomics to identify compounds that reduced VAV1 levels while limiting effects on unrelated proteins, then confirmed degradation dependency on the proteasome and cereblon (CRBN). The work pinpointed a region in VAV1—an SH3-2 domain loop acting as a degron—as essential for degradation. For the field, the study adds a practical path for how AI-enabled modeling can compress the cycle between hit discovery and mechanism-guided optimization in targeted protein degradation programs.