Deep-learning methods have revolutionized protein structure prediction and design but are presently limited to protein-only systems. We describe RoseTTAFold All-Atom (RFAA), which combines a residue-based representation of amino acids and DNA bases with an atomic representation of all other groups to model assemblies that contain proteins, nucleic acids, small molecules, metals, and covalent modifications, given their sequences and chemical structures. By fine-tuning on denoising tasks, we developed RFdiffusion All-Atom (RFdiffusionAA), which builds protein structures around small molecules. Starting from random distributions of amino acid residues surrounding target small molecules, we designed and experimentally validated, through crystallography and binding measurements, proteins that bind the cardiac disease therapeutic digoxigenin, the enzymatic cofactor heme, and the light-harvesting molecule bilin.

The design of enzymes with complex active sites that mediate multistep reactions remains an outstanding challenge. With serine hydrolases as a model system, we combined the generative capabilities of RFdiffusion with an ensemble generation method for assessing active site preorganization at each step in the reaction to design enzymes starting from minimal active site descriptions. Experimental characterization revealed catalytic efficiencies ( k _cat / K _m ) up to 2.2 × 10 ⁵ M ⁻¹ s ⁻¹ and crystal structures that closely match the design models (Cα root mean square deviations <1 angstrom). Selection for structural compatibility across the reaction coordinate enabled identification of new catalysts remove with five different folds distinct from those of natural serine hydrolases. Our de novo approach provides insight into the geometric basis of catalysis and a roadmap for designing enzymes that catalyze multistep transformations.

Designing new enzymes typically begins with idealized arrangements of catalytic functional groups around a reaction transition state, then attempts to generate protein structures that precisely position these groups. Current AI-based methods can create active enzymes but require predefined residue positions and rely on reverse-building residue backbones from side-chain placements, which limits design flexibility. Here we show that a new deep generative model, RoseTTAFold diffusion 2 (RFdiffusion2), overcomes these constraints by designing enzymes directly from functional group geometries without specifying residue order or performing inverse rotamer generation. RFdiffusion2 successfully generates scaffolds for all 41 active sites in a diverse benchmark, compared to 16 using previous methods. We further design enzymes for three distinct catalytic mechanisms and identify active candidates after experimentally testing fewer than 96 sequences in each case. These results highlight the potential of atomic-level generative modeling to create de novo enzymes directly from reaction mechanisms.

Deep learning has accelerated protein design, but most existing methods are restricted to generating protein backbone coordinates and often neglect interactions with other biomolecules. We present RFdiffusion3 (RFD3), a diffusion model that generates protein structures in the context of ligands, nucleic acids and other non-protein constellations of atoms. Because all polymer atoms are modeled explicitly, conditioning the model on complex sets of atom-level constraints for enzyme design and other challenges is both simpler and more effective than previous approaches. RFD3 achieves improved performance compared to prior approaches on a range of in silico benchmarks with one tenth the computational cost. Finally, we demonstrate the broad applicability of RFD3 by designing and experimentally characterizing DNA binding proteins and cysteine hydrolases. The ability to rapidly generate protein structures guided by complex sets of atom-level constraints in the context of arbitrary non-protein atoms should further expand the range of functions attainable through protein design.

De novo enzyme design seeks to build proteins containing ideal active sites with catalytic residues surrounding and stabilizing the transition state(s) of the target chemical reaction^{1,2,3,4,5,6,7}. The generative artificial intelligence method RFdiffusion^8,9 solves this problem, but requires specifying both the sequence position and backbone coordinates for each catalytic residue, limiting sampling. Here we introduce RFdiffusion2, which eliminates these requirements, and use it to design zinc metallohydrolases starting from quantum chemistry-derived active site geometries. From an initial set of 96 designs tested experimentally, the most active has a catalytic efficiency (k_cat/K_M) of 16,000 M⁻¹ s⁻¹, orders of magnitude higher than previously designed metallohydrolases^6,7,10,11. A second round of 96 designs yielded 3 additional highly active enzymes, with k_cat/K_M values of up to 53,000 M⁻¹ s⁻¹ and a catalytic rate constant (k_cat) of up to 1.5 s⁻¹. The design models of the four most active designs differ from known structures and from each other, and the crystal structure of the most active design is very close to the design model, demonstrating the accuracy of the design method. The most active enzymes are predicted by PLACER¹² and Chai-1 (ref. ¹³) to have preorganized active sites that effectively position the substrate for nucleophilic attack by a water molecule activated by the bound metal. The ability to generate highly active enzymes directly from the computer, without experimental optimization, should enable a new generation of potent designer catalysts^14,15.

De novo enzyme design starts from ideal active site descriptions consisting of constellations of catalytic residue functional groups around reaction transition state(s), and seeks to generate protein structures that can accurately hold the site in place. Highly active enzymes have been designed starting from such descriptions using the generative AI method RFdiffusion [1–3], but there are two current methodological limitations. First, the geometry of the active site can only be specified at the residue level, so for each catalytic residue functional group placed around the reaction transition state, the possible locations of the residue backbone must be enumerated by building side chain rotamers back from the functional group. Second, the location of the catalytic residues along the sequence must be specified in advance, which considerably limits the space of solutions which can be sampled. Here we describe a new deep generative method, Rosetta Fold diffusion 2 (RFdiffusion2), that solves both problems, enabling enzymes to be designed from sequence agnostic descriptions of functional group locations without inverse rotamer generation. We first evaluate RFdiffusion2 on an in silico enzyme design benchmark of 41 diverse active sites and find that it is able to successfully build proteins scaffolding all 41 sites, compared to 16/41 with prior state-of-the-art deep learning methods. Next, we design enzymes around three diverse catalytic sites and characterize the designs experimentally; in each case we identify active catalysts in testing less than 96 sequences. RFdiffusion2 demonstrates the potential of atomic resolution generative models for the design of de novo enzymes directly from their reaction mechanisms.