The DOI matches its Crossref record; 68 of the 68 references it cites were found in a registry or an index; 34 of the 40 sources its claims cite were checked for support.
De novo design of protein structure and function with RFdiffusion
Open accessPublished version of a preprint
Abstract
There has been considerable recent progress in designing new proteins using deep-learning methods 1–9 . Despite this progress, a general deep-learning framework for protein design that enables solution of a wide range of design challenges, including de novo binder design and design of higher-order symmetric architectures, has yet to be described. Diffusion models 10,11 have had considerable success in image and language generative modelling but limited success when applied to protein modelling, probably due to the complexity of protein backbone geometry and sequence–structure relationships. Here we show that by fine-tuning the RoseTTAFold structure prediction network on protein structure denoising tasks, we obtain a generative model of protein backbones that achieves outstanding performance on unconditional and topology-constrained protein monomer design, protein binder design, symmetric oligomer design, enzyme active site scaffolding and symmetric motif scaffolding for therapeutic and metal-binding protein design. We demonstrate the power and generality of the method, called RoseTTAFold diffusion (RFdiffusion), by experimentally characterizing the structures and functions of hundreds of designed symmetric assemblies, metal-binding proteins and protein binders. The accuracy of RFdiffusion is confirmed by the cryogenic electron microscopy structure of a designed binder in complex with influenza haemagglutinin that is nearly identical to the design model. In a manner analogous to networks that produce images from user-specified inputs, RFdiffusion enables the design of diverse functional proteins from simple molecular specifications.
Overview
CoChat Summary
TL;DR
RFdiffusion fine-tunes the RoseTTAFold structure-prediction network as a denoising diffusion model to generate protein backbones for unconditional monomers, symmetric oligomers, functional-motif scaffolds, and de novo binders. The method solves 23 of 25 motif-scaffolding benchmark problems and yields experimentally validated binders, metal-binding assemblies, and a cryo-EM structure matching its design model at 0.63 Å.
Key findings
- RFdiffusion solves 23 of 25 motif-scaffolding benchmark problems, compared with 15 for Hallucination and 19 for RFjoint Inpainting; for 19 of 23 solved problems, its success fraction exceeds both baselines.
- It generates unconditional monomers up to 600 residues and outperforms RF Hallucination on in silico success (n = 400 designs per condition, z = 9.5, P = 1.6 × 10⁻²¹); a 100-residue protein can be generated in as little as 11 s versus about 8.5 min for RF Hallucination.
- For binder design against five targets, the overall experimental success rate was 19%, roughly two orders of magnitude above the previous Rosetta-based method; the highest-affinity HA binder, HA_20, bound with KD = 28 nM, and its cryo-EM complex structure matched the design model at 0.63 Å.
- Symmetric and functional designs were experimentally validated: at least 87 of 608 oligomer designs had SEC oligomerization states closely consistent with design models; 18 of 36 tested Ni²⁺-binding designs bound with dissociation constants from low nanomolar to low micromolar; MDM2 binders reached 0.5 and 0.7 nM versus 600 nM for the native p53 peptide.
Methods
- RFdiffusion uses the RoseTTAFold frame representation (Cα coordinate plus N-Cα-C rigid orientation per residue), trains by noising PDB structures for up to 200 steps, and minimizes mean-squared error between frame predictions and true structures; self-conditioning and fine-tuning from pretrained RF weights improved performance.
- For sequence design, ProteinMPNN was used to design sequences encoding RFdiffusion backbones, typically sampling eight sequences per design; in silico success was validated with AlphaFold2 using mean pAE < 5, global backbone r.m.s.d. < 2 Å, and scaffolded-site backbone r.m.s.d. < 1 Å.
- Experimental characterization included circular dichroism, size-exclusion chromatography, negative-stain EM, cryo-EM, biolayer interferometry, and isothermal titration calorimetry across designed monomers, oligomers, metal-binding proteins, and binders.
Limitations
- The paper states that RFdiffusion cannot explicitly model bound small molecules at present, though it can implicitly model substrates with an external potential; it also states that further studies are required to assess success rates on more polar target sites and sites without native binding partners, and that proteome-wide specificity remains to be fully characterized.
- The paper notes that very large proteins are difficult to validate in silico because they are probably beyond single-sequence prediction capabilities of AF2 and ESMFold, and that some binder designs had insufficient material to permit screening at 10 μM.
- Evident from the text: many conclusions rely on the AF2-based in silico success metric, experimental validation was performed on subsets of designs rather than all generated designs, and the authors did not extensively explore simultaneous structure-and-sequence design within RFdiffusion.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures











![Extended Data Fig. 6 External potentials for generating pockets around substrate molecules. A–D) Example in silico successful designs for enzyme classes 2–5 (ref. 57, see also Fig. 4). Native enzyme (PDB: 1CWY, 1DE3, 1P1X, 1SNZ); catalytic site (teal); RFdiffusion output (gray: model, colors: AF2 prediction). Metrics (AF2 vs design backbone r.m.s.d., AF2 vs design motif backbone r.m.s.d., AF2 vs design motif full-atom r.m.s.d., AF2 pAE): EC2: 0.93 Å, 0.50 Å, 1.29 Å, 3.51; EC3: 0.92 Å, 0.60 Å, 1.07 Å, 4.59; EC4: 0.93 Å, 0.80 Å, 1.03 Å, 4.41; EC5: 0.78 Å, 0.44 Å, 1.14 Å, 3.32. E–H) Implicit modeling of a substrate while scaffolding a retroaldolase active site triad [TYR1051-LYS1083-TYR1180] from PDB: 5AN7. E) The potential used to implicitly model the substrate, which has both a repulsive and attractive field (see Supplementary Methods 4.4). F) Left: Kernel densities demonstrate that without using the external potential (pink), designs often fall into two failure modes: (1) no pocket, and (2) clashes with the substrate. Right: clashes (substrate < 3 Å of the backbone) & pockets (no clash and > 16 Cα within 3–8 Å of substrate) with and without the potential. Two-proportion z-test: n = 71/51 +/− potential; clashes z = −2.05, p = 0.02, pocket z = −2.27, p = 0.01. Each datapoint represents a design already passing the stringent in silico success metrics (AF2 motif r.m.s.d. < 1 Å, AF2 backbone r.m.s.d. < 2 Å, AF2 pAE < 5). Note that the potential and clash definition pertain only to backbone Cα atoms, and do not currently include sidechain atoms. G) Designs close to the labeled local maxima of the kernel density estimate. Without the potential, the catalytic triad is predominantly (1) exposed on the surface with no residues available to provide substrate stabilization or (2) buried in the protein core, preventing substrate access. With the potential, the catalytic triad is predominantly (3), partially buried in a concave pocket with shape complementary to the substrate. Backbone atoms within 3 Å of the substrate are shown in red. H) A variety of diverse designs with pockets made using the potential, with no clashes between the substrate and the AF2-predicted backbone. The functional form and parameters used for the pocket potential are detailed in Supplementary Methods 4.4. In each case the substrate is superimposed on the AF2 prediction of the catalytic triad.](https://figures.cochat.ai/paper-figures/e5993be7b61c46129fd41a6c74b3b8c9/de-novo-design-of-protein-structure-and-function-with-rfdiffusion-figure-12.jpg)



Verification
Claims the paper makes
17 Valid · 4 Not valid · 3 Not checked · 113 Own results and interpretations
Some cited sources support this claim and some don’t — read each source’s verdict below.
In the paper
“Recent work has adapted DDPMs for protein monomer design by conditioning on small protein ‘motifs’5,9 or on secondary structure and block-adjacency (‘fold’) information8.”
§ De novo design of protein structure and function with RFdiffusion > Main
What the sources say
- Supportedcites [5] Trippe, B. L. et al. Diffusion probabilistic modeling of protein backbones in 3D for the motif-scaffolding problem. in The Eleventh International Conference on Learning Representations (2023). source
The Introduction (p. 1) describes ProtDiff as a diffusion probabilistic model for protein backbones and SMCDiff as its motif-conditioned sampling procedure. The Discussion (p. 9) confirms that the procedure generates scaffolds conditioned on a motif. The secondary-structure and block-adjacency clause is attached to a different citation, so it is not attributed to this source. (LLM, conf=0.93)
- Disputedcites [9] Luo, S. et al. Antigen-specific antibody design and optimization with diffusion-based generative models. in Adv. Neural Information Processing Systems Vol. 35 (eds Koyejo, S. et al) 9754–9767 (Curran Associates, 2022). sourceExperimental — accuracy not yet measured
Section 3.1 (page unavailable) says the paper designs antibody CDRs given an antigen structure and antibody framework, not protein monomers conditioned on small motifs. The Generative Model section likewise describes a diffusion model conditioned on the antigen and antibody framework; it does not describe secondary-structure or block-adjacency conditioning. (LLM, conf=0.87)
Checked by a reasoning model — may be wrong.
- Supportedcites [8] Anand, N. & Achim, T. Protein structure and sequence generation with equivariant denoising diffusion probabilistic models. Preprint at https://doi.org/10.48550/arXiv.2205.15019 (2022). source
Citation 8 is attached to the secondary-structure and block-adjacency (‘fold’) conditioning, not to the motif examples cited separately. Section 2.4 (p. 5) defines those constraints, and Section 3 (p. 6) states that the diffusion model conditions on their encoded output to generate protein structures. (LLM, conf=0.94)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Self-conditioning within RFdiffusion notably improved performance on in silico benchmarks encompassing both conditional and unconditional protein design tasks (Fig. 2e and Extended Data Fig. 1e).”
§ De novo design of protein structure and function with RFdiffusion > Main
What the sources say
- Partly supportedcites [25] Chen, T., Zhang, R. & Hinton, G. Analog bits: generating discrete data using diffusion models with self-conditioning. in The Eleventh International Conference on Learning Representations (2023). sourceExperimental — accuracy not yet measured
The Discrete Image Generation section (pp. 6–7) reports that self-conditioning improved results for image diffusion models. It does not report RFdiffusion results or in silico benchmarks for conditional and unconditional protein design, so the claimed improvement in that setting is not supported by this source. (LLM, conf=0.86)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Experimental characterization showed that the method can scaffold a wide range of protein functional motifs with atomic accuracy19”
§ De novo design of protein structure and function with RFdiffusion > Main
What the sources say
- Partly supportedcites [4] Wang, J. et al. Scaffolding protein functional sites using deep learning. Science 377, 387–394 (2022). sourceExperimental — accuracy not yet measured
The 'Generalized functional motif scaffolding' section reports designs spanning a wide range of motifs and says their activities were assessed experimentally. But its sub-angstrom motif accuracy is based on AlphaFold predictions, not experimental structural measurements; the sentence overstates what the experimental characterization established. (LLM, conf=0.88)
- Partly supportedcites [19] Watson, J. L., Bera, A., Juergens, D., Wang, J. & Baker, D. X-ray crystallographic validation of design from this paper. Science 377, 387–394 (2022). sourceExperimental — accuracy not yet measured
The section “Generalized functional motif scaffolding by missing information recovery” (page unavailable) reports designs spanning a wide range of motifs, with motif RMSD typically below 1 Å in AlphaFold predictions. It separately reports experimental tests of the designs’ activities; those tests do not establish atomic structural accuracy. The paper also attributes the broad set of designs to both inpainting and constrained hallucination, rather than RFjoint inpainting alone. (LLM, conf=0.87)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“we established an in silico benchmark test (Supplementary Table 9) comprising 25 motif-scaffolding design problems addressed in six recent publications encompassing several design methodologies4,5,29,38–40”
§ De novo design of protein structure and function with RFdiffusion > Functional-motif scaffolding
What the sources say
- Could not verifycites [4] Wang, J. et al. Scaffolding protein functional sites using deep learning. Science 377, 387–394 (2022). sourceExperimental — accuracy not yet measured
The 'Generalized functional motif scaffolding by missing information recovery' section describes in silico evaluation of the paper’s own designs, but does not describe establishing a benchmark of 25 problems drawn from six publications. Searches for the benchmark, the numbers, and Supplementary Table 9 found no passage in our copy that verifies those details. (LLM, conf=0.74)
- Supportedcites [5] Trippe, B. L. et al. Diffusion probabilistic modeling of protein backbones in 3D for the motif-scaffolding problem. in The Eleventh International Conference on Learning Representations (2023). source
Section 5.2 (p. 7) reports motif-scaffolding tests using motifs from proteins 6exz and 5trv; Appendix H.1 (p. 24) describes another target, 5tpn. This supports citing the paper as one of the publications that addressed motif-scaffolding design problems. The citing authors describe the 25-problem benchmark as their own, not as something established by this source. (LLM, conf=0.88)
- Supportedcites [29] Lee, J. S. & Kim, P. M. ProteinSGM: Score-based generative modeling for de novo protein design. Preprint at bioRxiv https://doi.org/10.1101/2022.07.13.499967 (2022). source
The citation identifies one of the publications whose design methodology informed the citing authors’ benchmark; it does not attribute the establishment of the 25-problem benchmark to ProteinSGM. In “Conditional generation” (page unavailable), ProteinSGM presents scaffold-inpainting design cases, including generation of scaffolds around the barstar functional site. (LLM, conf=0.88)
- Supportedcites [38] Sesterhenn, F. et al. De novo protein design enables the precise induction of RSV-neutralizing antibodies. Science 368, eaay5051 (2020). source
The citation identifies this paper as one of the motif-scaffolding publications used to assemble the citing authors’ benchmark; the benchmark’s size and establishment are the citing authors’ own statements. The Results section (page unavailable) describes TopoBuilder as a protocol for designing protein topologies around functional motifs, so the paper fits that role. (LLM, conf=0.88)
- Not checked — no full text of the sourcecites [39] Yang, C. et al. Bottom-up de novo design of functional proteins with complex structural features. Nat. Chem. Biol. 17, 492–500 (2021). source
- Partly supportedcites [40] Glasgow, A. et al. Engineered ACE2 receptor traps potently neutralize SARS-CoV-2. Proc. Natl Acad. Sci. USA 117, 28046–28055 (2020). sourceExperimental — accuracy not yet measured
The Results and Materials and Methods describe computational redesign of the ACE2–spike binding interface, so this is a publication addressing a protein-design problem. Our copy does not describe that work as motif scaffolding. The benchmark of 25 problems is the citing authors’ own work, not a finding attributed to this source. (LLM, conf=0.78)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
Every source the paper cites for this claim was checked and supports it.
In the paper
“Vázquez Torres et al. demonstrate the ability of RFdiffusion to design picomolar affinity binders to flexible helical peptides54”
§ De novo design of protein structure and function with RFdiffusion > Discussion
What the sources say
- Supportedcites [54] Torres, S. V. et al. De novo design of high-affinity protein binders to bioactive helical peptides. Preprint at bioRxiv https://doi.org/10.1101/2022.12.10.519862 (2022). source
The Peptide binder design with RFdiffusion section reports de novo RFdiffusion binders to the helical peptides Bim and PTH, with an estimated affinity of ~100 pM for Bim and an upper bound of 500 pM for PTH. The Main section explains that peptide targets are often partially or entirely unstructured in isolation, supporting the description of them as flexible. (LLM, conf=0.94)
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
82 more claims · 7 with their cited sources checked
Create a free account to see all 137 claims and every source verdict.
- It may state two separate claims as one. 48
- Probably not a substantial claim. 3
- The quote only partly states this claim. 1
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“We also use Evo to generate genome-scale sequences with plausible high-level architecture more than1 megabase (Mb) in length, a scale that is orders of magnitude greater than previous methods (10, 13, 14).”
From Sequence modeling and design from molecular to genome scale with Evo (2024) · cites this paper as [10]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ De novo design of protein structure and function with RFdiffusion > Functional-motif scaffolding
“e, Out of 95 designs, 55 showed binding to MDM2 (more than 50% of maximum response).”
§ De novo design of protein structure and function with RFdiffusion > Functional-motif scaffolding
“The overall success rate was quite high: out of the 96 designs, 55 showed some detectable binding at 10 μM (Fig. 4e and Supplementary Fig. 10h).”
Citations
Cites
Highly accurate protein structure prediction with AlphaFold
Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Žídek, A., Potapenko, A., et al. (2021) · Nature · doi:10.1038/s41586-021-03819-2 · Cited by 43,764
Berman, H. M. (2000) · Nucleic Acids Research · doi:10.1093/nar/28.1.235 · Cited by 40,196
Denoising Diffusion Probabilistic Models
Ho, J., Jain, A., & Abbeel, P. (2020) · arXiv · doi:10.48550/arxiv.2006.11239 · Cited by 8,712
Evolutionary-scale prediction of atomic level protein structure with a language model
Lin, Z., Akin, H., Rao, R., Hie, B., Zhu, Z., Lu, W., Smetanin, N., Verkuil, R., Kabeli, O., Shmueli, Y., et al. (2022) · doi:10.1101/2022.07.20.500902 · Cited by 5,990
Accurate prediction of protein structures and interactions using a 3-track network
Baek, M., DiMaio, F., Anishchenko, I., Dauparas, J., Ovchinnikov, S., Lee, G. R., Wang, J., Cong, Q., Kinch, L. N., Schaeffer, R. D., et al. (2021) · doi:10.1101/2021.06.14.448402 · Cited by 5,549
Self-citationDeep Unsupervised Learning using Nonequilibrium Thermodynamics
Sohl-Dickstein, J., Weiss, E. A., Maheswaranathan, N., & Ganguli, S. (2015) · arXiv · doi:10.48550/arxiv.1503.03585 · Cited by 2,637
Zero-shot text-to-image generation
Ramesh, A., Pavlov, M., Goh, G., Gray, S., Voss, C., Radford, A., Chen, M., & Sutskever, I. (2021) · arXiv · doi:10.48550/arxiv.2102.12092 · Cited by 2,417
THEORY OF PROTEIN FOLDING: The Energy Landscape Perspective
Onuchic, J. N., Luthey-Schulten, Z., & Wolynes, P. G. (1997) · Annual Review of Physical Chemistry · doi:10.1146/annurev.physchem.48.1.545 · Cited by 2,210
Robust deep learning–based protein sequence design using ProteinMPNN
Dauparas, J., Anishchenko, I., Bennett, N., Bai, H., Ragotte, R. J., Milles, L. F., Wicky, B. I. M., Courbet, A., de Haas, R. J., Bethel, N., et al. (2022) · Science · doi:10.1126/science.add2187 · Cited by 2,210 · cited 6×
“We explored two different strategies for training RFdiffusion: (1) in a manner akin to ‘canonical’ diffusion models, with predictions at each timestep independent of predictions at previous timesteps (as in previous work(5,8,9,16)), and (2) with self-conditioning(25), in which the model can condition on previous predi…”
InfluentialSelf-citationStructure of the MDM2 Oncoprotein Bound to the p53 Tumor Suppressor Transactivation Domain
Kussie, P. H., Gorina, S., Marechal, V., Elenbaas, B., Moreau, J., Levine, A. J., & Pavletich, N. P. (1996) · Science · doi:10.1126/science.274.5289.948 · Cited by 2,206
Research works printed but not linked: 3
- Watson, J. L., Bera, A., Juergens, D., Wang, J. & Baker, D. X-ray crystallographic validation of design from this paper. Science 377, 387–394 (2022). · DOI · Paper page
- Leach, A., Schmon, S. M., Degiacomi, M. T. & Willcocks, C. G. Denoising diffusion probabilistic models on SO(3) for rotational alignment. In Proc. ICLR 2022 Workshop on Geometrical and Topological Representation Learning (2022).
- Anand, N. & Huang, P. Generative modeling for protein structures. in Adv. Neural Information Processing Systems Vol. 31 (eds Bengio, S. et al) (Curran Associates, 2018).
Cited by
Foundation models for generalist medical artificial intelligence
Moor, M., Banerjee, O., Abad, Z. S. H., Krumholz, H. M., Leskovec, J., Topol, E. J., & Rajpurkar, P. (2023) · Nature · doi:10.1038/s41586-023-05881-4 · Cited by 2,028
Generalized biomolecular modeling and design with RoseTTAFold All-Atom
Krishna, R., Wang, J., Ahern, W., Sturmfels, P., Venkatesh, P., Kalvet, I., Lee, G. R., Morey-Burrows, F. S., Anishchenko, I., Humphreys, I. R., et al. (2024) · Science · doi:10.1126/science.adl2528 · Cited by 978
“Diffusion-based methods can generate proteins in the context of a protein target that bind with considerable affinity and specificity ( 45 ) and can be trained to explicitly condition on structural features( 46 ).”
Self-citationSimulating 500 million years of evolution with a language model
Hayes, T., Rao, R., Akin, H., Sofroniew, N. J., Oktay, D., Lin, Z., Verkuil, R., Tran, V. Q., Deaton, J., Wiggert, M., et al. (2025) · Science · doi:10.1126/science.ads0018 · Cited by 937
From nature to industry: Harnessing enzymes for biocatalysis
Buller, R., Lutz, S., Kazlauskas, R. J., Snajdrova, R., Moore, J. C., & Bornscheuer, U. T. (2023) · Science · doi:10.1126/science.adh8615 · Cited by 706
Self-Driving Laboratories for Chemistry and Materials Science
Tom, G., Schmid, S. P., Baird, S. G., Cao, Y., Darvish, K., Hao, H., Lo, S., Pablo-García, S., Rajaonson, E. M., Skreta, M., et al. (2024) · Chemical Reviews · doi:10.1021/acs.chemrev.4c00055 · Cited by 707
Metadata
- Authors
- Watson, J. L., Juergens, D., Bennett, N. R., Trippe, B. L., Yim, J., Eisenach, H. E., Ahern, W., Borst, A. J., Ragotte, R. J., Milles, L. F., Wicky, B. I. M., Hanikel, N., Pellock, S. J., Courbet, A., Sheffler, W., Wang, J., Venkatesh, P., Sappington, I., Torres, S. V., Lauko, A., De Bortoli, V., Mathieu, E., Ovchinnikov, S., Barzilay, R., Jaakkola, T. S., DiMaio, F., Baek, M., & Baker, D.
- Affiliations
- University of Washington (US)
- Centre National de la Recherche Scientifique (FR)
- Harvard University (US)
- Massachusetts Institute of Technology (US)
- Seoul National University (KR)
- Howard Hughes Medical Institute (US)
- Corresponding author
- Baker, David
- ORCID
- Watson, Joseph L.
- Juergens, David
- Bennett, Nathaniel R.
- Trippe, Brian L.
- Yim, Jason
- Eisenach, Helen E.
- Ahern, Woody
- Borst, Andrew J.
- Ragotte, Robert J.
- Milles, Lukas F.
- Wicky, Basile I. M.
- Hanikel, Nikita
- Pellock, Samuel J.
- Courbet, Alexis
- Sheffler, William
- Wang, Jue
- Venkatesh, Preetham
- Sappington, Isaac
- Torres, Susana Vázquez
- Lauko, Anna
- Ovchinnikov, Sergey
- Barzilay, Regina
- Jaakkola, Tommi S.
- DiMaio, Frank
- Baek, Minkyung
- Baker, David
- Venue
- Nature
- Published
- 2023-07-11
- Volume
- 620
- Issue
- 7976
- Pages
- 1089-1100
- ISSN
- 0028-0836 (ISSN-L)
- 1476-4687
- PMC
- PMC10468394
- OpenAlex
- W4383957026
- Semantic Scholar
- eb8c0df75993c1c19c51cde9345e45fc260f661c
- Cited by
- 2,325
- References
- 57
- Citations by year
- 2026: 899 · 2025: 795 · 2024: 473 · 2023: 153
- FWCI
- 172.50
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- Nature (publishedVersion, cc-by)
- bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by-nd)
- PubMed (publishedVersion)
- PubMed Central (submittedVersion, cc-by)
- RePEc: Research Papers in Economics (submittedVersion)
- Apollo (University of Cambridge) (submittedVersion, cc-by)
- Funders
- National Science Foundation (HDTRA1-19-1-0003, 1918839)
- Howard Hughes Medical Institute
- U.S. Department of Energy (DESC0018940)
- Bill and Melinda Gates Foundation (INV-010680)
- Washington Research Foundation
- Amgen
- Sanofi
- Washington State University
- Open Philanthropy Project
- Defense Advanced Research Projects Agency
- Defense Threat Reduction Agency (HDTRA1-19-1-0003, HDTRA1)
- Office of Science (DE-SC0018940 MOD03, DESC0018940)
- Engineering and Physical Sciences Research Council (EP/T005386/1)
- National Institute on Aging (U19AG065156)
- National Institute of General Medical Sciences (T32GM007250, T32GM008268)
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Structure and Dynamics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Machine Learning in Bioinformatics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Structural Biology › Enzyme Structure and Function
- Keywords
- protein design, de novo protein design, diffusion models, RoseTTAFold, protein binder design, metal-binding proteins, protein backbone, influenza haemagglutinin, therapeutic proteins
- MeSH terms
- Deep Learning, Protein Binding, Proteins, Hemagglutinin Glycoproteins, Influenza Virus, Catalytic Domain, Cryoelectron Microscopy
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-05 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: De novo design of protein structure and function with RFdiffusion. https://app.cochat.ai/papers/10.1038/s41586-023-06415-8
- The paper
- Watson, J. L., Juergens, D., Bennett, N. R., Trippe, B. L., Yim, J., Eisenach, H. E., Ahern, W., Borst, A. J., Ragotte, R. J., Milles, L. F., Wicky, B. I. M., Hanikel, N., Pellock, S. J., Courbet, A., Sheffler, W., Wang, J., Venkatesh, P., Sappington, I., Torres, S. V., . . . Baker, D. (2023). De novo design of protein structure and function with RFdiffusion. Nature. https://doi.org/10.1038/s41586-023-06415-8
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.