analiza aiqbench

This study marks a significant milestone where RFdiffusion2 successfully generated functional metallohydrolases from scratch by integrating quantum chemistry calculations to define active site geometries. The resulting enzymes demonstrated high catalytic efficiency without requiring directed evolution, validating the model's ability to predict complex protein structures that accelerate specific chemical reactions. While practical applications still require further testing for stability and manufacturability, this achievement highlights the growing convergence of generative AI and quantum methods in de novo protein design.