The rapid advancement of cryo-electron microscopy (cryo-EM) has established it as an essential tool in contemporary structural biology and drug discovery, providing high-resolution insights into complex drug-target interactions. To fully leverage cryo-EM’s potential in an industrial setting, focused efforts are required to enhance the efficiency and success rate of structure determination, especially for challenging therapeutic targets.
​ This presentation will detail two key methodological and technological innovations developed at Sanofi that streamline the cryo-EM pipeline.
The first is "crai" (cryo-EM antibody identification), a machine learning-based ChimeraX bundle. This computational tool is designed to automate the robust detection, classification, and precise orientation analysis of antibody fragments (Fabs, Fcs) directly within cryo-EM density maps. By replacing manual fitting and interpretation, "crai" significantly accelerates the structural analysis of therapeutic antibody complexes.
The second innovation addresses sample preparation for difficult-to-resolve targets. We introduce the "Fabuoulous" mutation, a specific modification within an antibody Fab fragment that optimizes its properties for cryo-EM. This modified Fab can be used as a stable, minimally intrusive scaffold to facilitate the structure determination of target molecules without the need for large, conventional fiducial markers or affinity tags. This approach simplifies sample requirements and yields higher-quality maps for small or conformationally dynamic targets.
These innovations highlight a dual strategy—combining advanced machine learning with targeted biochemical engineering—that is critical for maximizing the efficiency and impact of cryo-EM in accelerating the structural-guided design of novel therapeutics at Sanofi. ​