Our researcher Ana María Fernández Escamilla is currently participating in the International Conference of the Spanish Biophysical Society (SBE), where she is presenting her work titled:
“AI-Driven de novo Design of Selective Focal Adhesion Kinase Inhibitors”
This study, conducted in collaboration with Magnus Bauer (University of Washington) and Daniel Lietha (Centro de Investigaciones Biológicas Margarita Salas, CSIC, Madrid), focuses on recent advancements in deep learning-based tools, particularly AlphaFold (AF) and RoseTTAFold (RF). These tools have revolutionized structural biology by enabling highly accurate protein structure predictions from amino acid sequences. A key innovation is the application of diffusion probabilistic models for denoising (DDPM), exemplified by RFdiffusion.
The researchers used RFdiffusion to design de novo protein inhibitors, referred to as “Deep binders,” targeting the kinase domain of the focal adhesion kinase (FAK) complex. This complex is a critical target in tumor progression and, therefore, in cancer therapy.
Thanks to these innovations, the team has successfully designed proteins capable of binding with high affinity to other proteins—even on surfaces previously considered inaccessible to small molecules. This opens up new possibilities for developing targeted treatments for diseases where traditional therapies are ineffective.
This breakthrough marks a significant step forward in AI-driven rational design of protein-based therapies.