Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 39 of the 39 references it cites were found in a registry or an index.

Protein sequence design by conformational landscape optimization

Open access
Norn, Christoffer, Wicky, Basile I. M., Juergens, David, Liu, Sirui, Kim, David, Tischer, Doug, Koepnick, Brian, Anishchenko, Ivan, Foldit Players, Baker, David, Ovchinnikov, Sergey, Coral, Alan, Bubar, Alex J., Boykov, Alexander, Valle Pérez, Alexander Uriel, MacMillan, Alison, Lubow, Allen, Mussini, Andrea, Cai, Andrew, Ardill, Andrew John, Seal, Aniruddha, Kalantarian, Artak, Failer, Barbara, Lackersteen, Belinda, Chagot, Benjamin, Haight, Beverly R., Taştan, Bora, Uitham, Boris, Roy, Brandon G., de Melo Cruz, Breno Renan, Echols, Brian, Lorenz, Brian Edward, Blair, Bruce, Kestemont, Bruno, Eastlake, C. D., Bragdon, Callen Joseph, Vardeman, Carl, Salerno, Carlo, Comisky, Casey, Hayman, Catherine Louise, Landers, Catherine R, Zimov, Cathy, Coleman, Charles David, Painter, Charles Robert, Ince, Christopher, Lynagh, Conor, Malaniia, Dmitrii, Wheeler, Douglas Craig, Robertson, Douglas, Simon, Vera, Chisari, Emanuele, Kai, Eric Lim Jit, Rezae, Farah, Lengyel, Ferenc, Tabotta, Flavian, Padelletti, Franco, Boström, Frisno, Gross, Gary O., McIlvaine, George, Beecher, Gil, Hansen, Gregory T., de Jong, Guido, Feldmann, Harald, Borman, Jami Lynne, Quinn, Jamie, Norrgard, Jane, Truong, Jason, Diderich, Jasper A., Canfield, Jeffrey Michael, Photakis, Jeffrey, Slone, Jesse David, Madzio, Joanna, Mitchell, Joanne, Stomieroski, John Charles, Mitch, John H., Altenbeck, Johnathan Robert, Schinkler, Jonas, Weinberg, Jonathan Barak, Burbach, Joshua David, Sequeira da Costa, João Carlos, Bada Juarez, Juan Francisco, Gunnarsson, Jón Pétur, Harper, Kathleen Diane, Joo, Keehyoung, Clayton, Keith T., DeFord, Kenneth E., Scully, Kevin F., Gildea, Kevin M., Abbey, Kirk J., Kohli, Kristen Lee, Stenner, Kyle, Takács, Kálmán, Poussaint, LaVerne L., Manalo, Larry C., Withers, Larry C., Carlson, Lilium, Wei, Linda, Fisher, Luke Ryan, Carpenter, Lynn, Ji-hwan, Ma, Ricci, Manuel, Belcastro, Marcus Anthony, Leniec, Marek, Hohmann, Marie, Thompson, Mark, Thayer, Matthew A., Gaebel, Matthias, Cassidy, Michael D., Fagiola, Michael, Lewis, Michael, Pfützenreuter, Michael, Simon, Michael, Elmassry, Moamen M., Benevides, Noah, Kerr, Norah Kathleen, Verma, Nupur, Shannon, Oak, Yin, Owen, Wolfteich, Pascal, Gummersall, Paul, Tłuścik, Paweł, Gajar, Peter, Triggiani, Peter John, Guha, Rajarshi, Mathew Innes, Renton Braden, Buchanan, Ricky, Gamble, Robert, Leduc, Robert, Spearing, Robert, dos Santos Gomes, Rodrigo Luccas Corrêa, Estep, Roger D., DeWitt, Ryan, Moore, Ryan, Shnider, Scott G., Zaccanelli, Scott J., Kuznetsov, Sergey, Burillo-Sanz, Sergio, Mooney, Seán, Vasiliy, Sidoruk, Butkovich, Slava S., Hudson, Spencer Bruce, Pote, Spencer Len, Denne, Stephen Phillip, Schwegmann, Steven A., Ratna, Sumanth, Kleinfelter, Susan C., Bausewein, Thomas, George, Thomas J., de Almeida, Tobias Scherf, Yeginer, Ulas, Barmettler, Walter, Pulley, Warwick Robert, Wright, William Scott, Willyanto, Lansford, Wyatt, Hochart, Xavier, Gaiji, Yoan Anthony Skander, Lagodich, Yuriy, Christian, Vivier · 2021 · Proceedings of the National Academy of Sciences

Abstract

Significance: Almost all proteins fold to their lowest free energy state, which is determined by their amino acid sequence. Computational protein design has primarily focused on finding sequences that have very low energy in the target designed structure. However, what is most relevant during folding is not the absolute energy of the folded state but the energy difference between the folded state and the lowest-lying alternative states. We describe a deep learning approach that captures aspects of the folding landscape, in particular the presence of structures in alternative energy minima, and show that it can enhance current protein design methods.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1073/pnas.2017228118

Cites

39 of 39 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1. Protein sequence design. (A) The goal of fixed backbone protein design is to find a sequence that best specifies the desired structure (P). Traditional energy-based methods have approached the problem heuristically, focusing solely on minimizing the energy of the target conformation in the hope that any stable alternative conformation is unlikely to arise by chance. However, this narrow focus on a single desired structure can produce solutions with low-energy alternative states, as suggested in the energy landscape of sequence α. An ideal method would instead find the sequence that maximizes the probability of the desired structure over all other states. Such a method would select sequence β. (B) Overview of trRosetta fixed backbone sequence design method. Starting with a random matrix of sequence length by number of amino acids (logits), the maximum value at each position is taken to generate a sequence, which is fed into the trRosetta model. The output is the predicted distribution of distances, angles, and dihedrals for every pair of residues (here, we only show distances). The loss is defined as the difference between the target and prediction, and the gradient is computed to minimize the loss. After normalization, the gradient is applied to the logits, and the process is repeated until convergence.
Fig. 1. Protein sequence design. (A) The goal of fixed backbone protein design is to find a sequence that best specifies the desired structure (P). Traditional energy-based methods have approached the problem heuristically, focusing solely on minimizing the energy of the target conformation in the hope that any stable alternative conformation is unlikely to arise by chance. However, this narrow focus on a single desired structure can produce solutions with low-energy alternative states, as suggested in the energy landscape of sequence α. An ideal method would instead find the sequence that maximizes the probability of the desired structure over all other states. Such a method would select sequence β. (B) Overview of trRosetta fixed backbone sequence design method. Starting with a random matrix of sequence length by number of amino acids (logits), the maximum value at each position is taken to generate a sequence, which is fed into the trRosetta model. The output is the predicted distribution of distances, angles, and dihedrals for every pair of residues (here, we only show distances). The loss is defined as the difference between the target and prediction, and the gradient is computed to minimize the loss. After normalization, the gradient is applied to the logits, and the process is repeated until convergence.CC BY-NC-ND · PubMed Central
Fig. 2. trRosetta predicts properties of the folding energy landscape. (A) trRosetta better predicts which designs will have high Boltzmann probabilities than Rosetta-based energy calculations, which only see the target conformation (classifications for Pnear > 0.8, AUCtrRosetta = 0.81 vs. AUCRosetta = 0.65, n = 4,204 designs). (B) trRosetta correctly predicts dilution of probability for designs with multiple low-energy conformations (columns 2 to 4) compared with designs with a single global energy minimum (column 1). Structural decoys were binned, and the mean trRosetta score corrected for background (−logPc(struct.|seq.)) is represented by the color gradient from dark blue (high probability) to red (low probability). (C) Structures of the lowest-energy representatives (indicated by circles on the energy landscape). The designed structures and alternative states are shown on the left and right, respectively, of each column. (D) Selected examples of probability distributions (Cβ–Cβ distance prediction) for specific i,j pairs (numbering indicated on the top) demonstrating bimodality. The actual distances observed in the designed and alternative structures are indicated by vertical lines (blue and red, respectively) and shown as spheres on the corresponding structures. More examples are shown in SI Appendix, Fig. S4, and an analysis of the prediction of bimodality from distributions of individual i,j pairs can be found in SI Appendix, Fig. S5.
Fig. 2. trRosetta predicts properties of the folding energy landscape. (A) trRosetta better predicts which designs will have high Boltzmann probabilities than Rosetta-based energy calculations, which only see the target conformation (classifications for Pnear > 0.8, AUCtrRosetta = 0.81 vs. AUCRosetta = 0.65, n = 4,204 designs). (B) trRosetta correctly predicts dilution of probability for designs with multiple low-energy conformations (columns 2 to 4) compared with designs with a single global energy minimum (column 1). Structural decoys were binned, and the mean trRosetta score corrected for background (−logPc(struct.|seq.)) is represented by the color gradient from dark blue (high probability) to red (low probability). (C) Structures of the lowest-energy representatives (indicated by circles on the energy landscape). The designed structures and alternative states are shown on the left and right, respectively, of each column. (D) Selected examples of probability distributions (Cβ–Cβ distance prediction) for specific i,j pairs (numbering indicated on the top) demonstrating bimodality. The actual distances observed in the designed and alternative structures are indicated by vertical lines (blue and red, respectively) and shown as spheres on the corresponding structures. More examples are shown in SI Appendix, Fig. S4, and an analysis of the prediction of bimodality from distributions of individual i,j pairs can be found in SI Appendix, Fig. S5.CC BY-NC-ND · PubMed Central
Fig. 3. (A and B) trRosetta predicts scaffold designability and experimental success. (A) Across different topologies, trRosetta predictions are better correlated with experimental protease stability—a measure of folding success—than Rosetta energy (R2trRosetta = 0.79 vs. R2Rosetta = 0.00, P value < 0.0001) (SI Appendix, Methods). Data points are the topology-specific mean values, and error bars represent SDs (eight topologies, Ntot = 30,159 designs). Without topological averaging, the correlation decreases (R2trRosetta = 0.20 vs. R2Rosetta = 0.03, P value < 0.0001) because intratopological differences are not well captured by trRosetta (SI Appendix, Fig. S6 has details). (B) trRosetta is significantly better at discriminating experimental success (expression, nonaggregation, and having correct secondary structure content) than Rosetta energy (AUCtrRosetta = 0.81 vs. AUCRosetta = 0.64). Data from 145 Foldit-generated designs (16). (C and D) Designing with a hybrid trRosetta–Rosetta protocol disfavors off-target states. (C) Examples of energy landscapes for two Foldit-generated backbones, each designed with trRosetta, Rosetta, and the trRosetta–Rosetta hybrid protocol. (D) The hybrid protocol improves the quality of the resulting energy landscapes, as determined by the Pnear quantity. trRosetta on its own also improves funnels but only superficially (better performance than Rosetta in the lower Pnear regime). It does not, however, generate a deep minimum in the vicinity of the designed state (poorer performance than Rosetta in the high Pnear regime). (E) The local sequence–structure relationship is idealized in trRosetta designs compared with both native proteins and Rosetta designs. The native structures that were used for redesign were at most 30% sequence identical to any protein in the trRosetta training dataset. Local sequence–structure agreements were measured as the average RMSD between the designed structure and nine-residue fragments from the PDB that were selected based on the sequence of the design. (F) For the same set of native backbones, trRosetta redesigns have a more native-like distribution of hydrophobic residues (F, I, L, V, M, W, Y) on the protein surface than Rosetta redesigns. The degree of burial was assessed with the software DEPTH (33), which computes the distance in ångströms between each residue and bulk solvent. A breakdown by amino acid is in SI Appendix, Fig. S13.
Fig. 3. (A and B) trRosetta predicts scaffold designability and experimental success. (A) Across different topologies, trRosetta predictions are better correlated with experimental protease stability—a measure of folding success—than Rosetta energy (R2trRosetta = 0.79 vs. R2Rosetta = 0.00, P value < 0.0001) (SI Appendix, Methods). Data points are the topology-specific mean values, and error bars represent SDs (eight topologies, Ntot = 30,159 designs). Without topological averaging, the correlation decreases (R2trRosetta = 0.20 vs. R2Rosetta = 0.03, P value < 0.0001) because intratopological differences are not well captured by trRosetta (SI Appendix, Fig. S6 has details). (B) trRosetta is significantly better at discriminating experimental success (expression, nonaggregation, and having correct secondary structure content) than Rosetta energy (AUCtrRosetta = 0.81 vs. AUCRosetta = 0.64). Data from 145 Foldit-generated designs (16). (C and D) Designing with a hybrid trRosetta–Rosetta protocol disfavors off-target states. (C) Examples of energy landscapes for two Foldit-generated backbones, each designed with trRosetta, Rosetta, and the trRosetta–Rosetta hybrid protocol. (D) The hybrid protocol improves the quality of the resulting energy landscapes, as determined by the Pnear quantity. trRosetta on its own also improves funnels but only superficially (better performance than Rosetta in the lower Pnear regime). It does not, however, generate a deep minimum in the vicinity of the designed state (poorer performance than Rosetta in the high Pnear regime). (E) The local sequence–structure relationship is idealized in trRosetta designs compared with both native proteins and Rosetta designs. The native structures that were used for redesign were at most 30% sequence identical to any protein in the trRosetta training dataset. Local sequence–structure agreements were measured as the average RMSD between the designed structure and nine-residue fragments from the PDB that were selected based on the sequence of the design. (F) For the same set of native backbones, trRosetta redesigns have a more native-like distribution of hydrophobic residues (F, I, L, V, M, W, Y) on the protein surface than Rosetta redesigns. The degree of burial was assessed with the software DEPTH (33), which computes the distance in ångströms between each residue and bulk solvent. A breakdown by amino acid is in SI Appendix, Fig. S13.CC BY-NC-ND · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

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?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 34 references

Cited by

  • Large language models generate functional protein sequences across diverse families

    Madani, A., Krause, B., Greene, E. R., Subramanian, S., Mohr, B. P., Holton, J. M., Olmos, J. L., Xiong, C., Sun, Z. Z., Socher, R., et al. (2023) · Nature Biotechnology · doi:10.1038/s41587-022-01618-2 · Cited by 1,078

  • De novo protein design by deep network hallucination

    Anishchenko, I., Chidyausiku, T. M., Ovchinnikov, S., Pellock, S. J., & Baker, D. (2020) · doi:10.1101/2020.07.22.211482 · Cited by 629

    “On the sampling side, the Monte Carlo approach can be made more efficient by direct gradient-based minimization by tracing the gradients back to the inputs 25 .”
    MethodSelf-citation
  • AlphaFold2 and its applications in the fields of biology and medicine

    Yang, Z., Zeng, X., Zhao, Y., & Chen, R. (2023) · Signal Transduction and Targeted Therapy · doi:10.1038/s41392-023-01381-z · Cited by 577

    “111,112 The integrity of the structures predicted is validated and con fi rmed by standard ab initio folding, protein structure analysis methods, and rigorous all-atom molecular dynamics simulations.”
    Method
  • Scaffolding protein functional sites using deep learning

    Wang, J., Lisanza, S., Juergens, D., Tischer, D., Watson, J. L., Castro, K. M., Ragotte, R., Saragovi, A., Milles, L. F., Baek, M., et al. (2022) · Science · doi:10.1126/science.abn2100 · Cited by 487

    “…fold to the hallucinated structures (12). trRosetta can also be used to design sequences that fold into a target backbone structure by carrying out sequence optimization using a structure recapitulation loss function that rewards similarity of the predicted structure to the target structure (13).”
    BackgroundMethodSelf-citation
  • De novo design of luciferases using deep learning

    Yeh, A. H.-W., Norn, C., Kipnis, Y., Tischer, D., Pellock, S. J., Evans, D., Ma, P., Lee, G. R., Zhang, J. Z., Anishchenko, I., et al. (2023) · Nature · doi:10.1038/s41586-023-05696-3 · Cited by 441

    “…of ideal NTF2-like structures, we developed a deep-learning-based ‘family-wide hallu-cination’ approach that integrates unconstrained de novo design 17,18 and Rosetta sequence-design approaches 19 to enable the generation of an essentially unlimited number of proteins that have a desired fold (Fig.”
    MethodSelf-citation

See all 168

Metadata

Authors
Norn, C., Wicky, B. I. M., Juergens, D., Liu, S., Kim, D., Tischer, D., Koepnick, B., Anishchenko, I., Players, F., Baker, D., Ovchinnikov, S., Coral, A., Bubar, A. J., Boykov, A., Valle Pérez, A. U., MacMillan, A., Lubow, A., Mussini, A., Cai, A., Ardill, A. J., Seal, A., Kalantarian, A., Failer, B., Lackersteen, B., Chagot, B., Haight, B. R., Taştan, B., Uitham, B., Roy, B. G., de Melo Cruz, B. R., Echols, B., Lorenz, B. E., Blair, B., Kestemont, B., Eastlake, C. D., Bragdon, C. J., Vardeman, C., Salerno, C., Comisky, C., Hayman, C. L., Landers, C. R., Zimov, C., Coleman, C. D., Painter, C. R., Ince, C., Lynagh, C., Malaniia, D., Wheeler, D. C., Robertson, D., Simon, V., Chisari, E., Kai, E. L. J., Rezae, F., Lengyel, F., Tabotta, F., Padelletti, F., Boström, F., Gross, G. O., McIlvaine, G., Beecher, G., Hansen, G. T., de Jong, G., Feldmann, H., Borman, J. L., Quinn, J., Norrgard, J., Truong, J., Diderich, J. A., Canfield, J. M., Photakis, J., Slone, J. D., Madzio, J., Mitchell, J., Stomieroski, J. C., Mitch, J. H., Altenbeck, J. R., Schinkler, J., Weinberg, J. B., Burbach, J. D., Sequeira da Costa, J. C., Bada Juarez, J. F., Gunnarsson, J. P., Harper, K. D., Joo, K., Clayton, K. T., DeFord, K. E., Scully, K. F., Gildea, K. M., Abbey, K. J., Kohli, K. L., Stenner, K., Takács, K., Poussaint, L. L., Manalo, L. C., Withers, L. C., Carlson, L., Wei, L., Fisher, L. R., Carpenter, L., Ji-hwan, M., Ricci, M., Belcastro, M. A., Leniec, M., Hohmann, M., Thompson, M., Thayer, M. A., Gaebel, M., Cassidy, M. D., Fagiola, M., Lewis, M., Pfützenreuter, M., Simon, M., Elmassry, M. M., Benevides, N., Kerr, N. K., Verma, N., Shannon, O., Yin, O., Wolfteich, P., Gummersall, P., Tłuścik, P., Gajar, P., Triggiani, P. J., Guha, R., Mathew Innes, R. B., Buchanan, R., Gamble, R., Leduc, R., Spearing, R., dos Santos Gomes, R. L. C., Estep, R. D., DeWitt, R., Moore, R., Shnider, S. G., Zaccanelli, S. J., Kuznetsov, S., Burillo-Sanz, S., Mooney, S., Vasiliy, S., Butkovich, S. S., Hudson, S. B., Pote, S. L., Denne, S. P., Schwegmann, S. A., Ratna, S., Kleinfelter, S. C., Bausewein, T., George, T. J., de Almeida, T. S., Yeginer, U., Barmettler, W., Pulley, W. R., Wright, W. S., Willyanto, Lansford, W., Hochart, X., Gaiji, Y. A. S., Lagodich, Y., & Christian, V.
Affiliations
  • University of Washington (US)
  • Harvard University (US)
  • Howard Hughes Medical Institute (US)
Corresponding author
  • Baker, David
  • Ovchinnikov, Sergey
Venue
Proceedings of the National Academy of Sciences
Published
2021-03-12
Volume
118
Issue
11
ISSN
  • 0027-8424 (ISSN-L)
  • 1091-6490
OpenAlex
W3134186384
Cited by
144
References
34
Citations by year
2026: 19 · 2025: 15 · 2024: 26 · 2023: 33 · 2022: 30 · 2021: 21
FWCI
8.62
Type
journal article
Language
en
Open access
Locations
  • Proceedings of the National Academy of Sciences (publishedVersion, cc-by-nc-nd)
  • PubMed Central (publishedVersion, cc-by-nc-nd)
  • PubMed (publishedVersion)
Funders
  • Novo Nordisk Foundation (NNF17OC0030446)
  • European Molecular Biology Organization (ALTF 139-2018)
  • HHS | NIH | NIH Office of the Director (DP5OD026389)
  • Howard Hughes Medical Institute (N/A)
  • Audacious Project at the Institute for Protein Design (N/A)
  • HHS | NIH | National Institute of Allergy and Infectious Diseases (HHSN272201700059C)
  • Schmidt Futures program (N/A)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Structure and Dynamics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › RNA and protein synthesis mechanisms
  • Physical Sciences › Chemistry › Organic Chemistry › Chemical Synthesis and Analysis
Keywords
protein design, backpropagation, de novo protein design, energy landscape, folding stability
MeSH terms
Models, Molecular, Protein Conformation, Proteins, Thermodynamics, Neural Networks, Computer, Protein Folding

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.

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: Protein sequence design by conformational landscape optimization. https://app.cochat.ai/papers/10.1073/pnas.2017228118
The paper
Norn, C., Wicky, B. I. M., Juergens, D., Liu, S., Kim, D., Tischer, D., Koepnick, B., Anishchenko, I., Players, F., Baker, D., Ovchinnikov, S., Coral, A., Bubar, A. J., Boykov, A., Valle Pérez, A. U., MacMillan, A., Lubow, A., Mussini, A., Cai, A., . . . Christian, V. (2021). Protein sequence design by conformational landscape optimization. Proceedings of the National Academy of Sciences. https://doi.org/10.1073/pnas.2017228118

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.