Justas Dauparas
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Justas Dauparas has a total of 225 co-authors. Here are the top co-authors by number of shared publications:
- David Baker
University of Washington, Howard Hughes Medical Institute
25 shared publications - Ivan Anishchenko
Department of Biochemistry, University of Washington, Seattle, WA, USA
10 shared publications - Minkyung Baek
Department of Biological Sciences, Seoul National University, Seoul, Republic of Korea
10 shared publications - Asim K. Bera
University of Washington
9 shared publications - Alex Kang
University of Washington
8 shared publications - Sergey Ovchinnikov
Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, USA
8 shared publications
Latest Publications
- Top Cited PapersOpen Access
- 28 March 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Small-molecule binding and sensing with a designed protein family
The de novo design of small-molecule–binding proteins holds great promise as a potential tool to develop sensors on-demand for arbitrary small molecules. Here we combine deep learning and physics-based methods to generate a family of proteins with diverse and designable pocket geometries, which we employ to computationally design binders for six small-molecule targets. Biophysical characterization of the designed binders reveals nanomolar to low micromolar binding affinities and atomic-level design accuracy. Additionally, we use a cortisol binder to design a chemically induced dimerization (CID) system that enables the construction of a biosensor for cortisol detection. The approach described here demonstrates the potential of the NTF2 fold and deep learning-based protein design in sensor development, paving the way for future platforms to design binders and sensors for small molecules across analytical, environmental, and biomedical applications.
- 28 March 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Top Cited PapersOpen AccessSmall-molecule binding and sensing with a designed protein family
The de novo design of small-molecule–binding proteins holds great promise as a potential tool to develop sensors on-demand for arbitrary small molecules. Here we combine deep learning and physics-based methods to generate a family of proteins with diverse and designable pocket geometries, which we employ to computationally design binders for six small-molecule targets. Biophysical characterization of the designed binders reveals nanomolar to low micromolar binding affinities and atomic-level design accuracy. Additionally, we use a cortisol binder to design a chemically induced dimerization (CID) system that enables the construction of a biosensor for cortisol detection. The approach described here demonstrates the potential of the NTF2 fold and deep learning-based protein design in sensor development, paving the way for future platforms to design binders and sensors for small molecules across analytical, environmental, and biomedical applications.
- Open Access
- 24 December 2025
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Stable de novo protein design via joint conformational landscape and sequence optimization
Generative protein modeling provides advanced tools for designing diverse protein sequences and structures. However, accurately modeling the conformational landscape and designing sequences remain critical challenges: ensuring that the designed sequence reliably folds into the target structure as its most stable conformation, and optimizing the sequence for a given suboptimal fixed input structure. In this study, we present a systematic analysis of jointly optimizing sequence-to-structure and structure-to-sequence mappings. This approach enables us to find optimal solutions for modeling the conformational landscape. We validate our approach with large-scale protein stability measurements, demonstrating that joint optimization is superior for designing stable proteins using a joint model (TrRosetta and TrMRF) and for achieving high accuracy in stability prediction when jointly modeling (half-masked ESMFold pLDDT + ESM2 Pseudo-likelihood). We further investigate features of sequences generated from the joint model and find that they exhibit higher frequencies of hydrophilic interactions, which may help maintain both secondary structure registry and pairing-features not captured by structure-to-sequence modeling alone.
- 24 December 2025
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Open AccessStable de novo protein design via joint conformational landscape and sequence optimization
Generative protein modeling provides advanced tools for designing diverse protein sequences and structures. However, accurately modeling the conformational landscape and designing sequences remain critical challenges: ensuring that the designed sequence reliably folds into the target structure as its most stable conformation, and optimizing the sequence for a given suboptimal fixed input structure. In this study, we present a systematic analysis of jointly optimizing sequence-to-structure and structure-to-sequence mappings. This approach enables us to find optimal solutions for modeling the conformational landscape. We validate our approach with large-scale protein stability measurements, demonstrating that joint optimization is superior for designing stable proteins using a joint model (TrRosetta and TrMRF) and for achieving high accuracy in stability prediction when jointly modeling (half-masked ESMFold pLDDT + ESM2 Pseudo-likelihood). We further investigate features of sequences generated from the joint model and find that they exhibit higher frequencies of hydrophilic interactions, which may help maintain both secondary structure registry and pairing-features not captured by structure-to-sequence modeling alone.
- Open Access
- 4 November 2025
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 122 (45)
Modeling protein–small molecule conformational ensembles with PLACER
Modeling the conformational heterogeneity of protein–small molecule interactions is important for understanding natural systems and evaluating designed systems but remains an outstanding challenge. We reasoned that while residue-level descriptions of biomolecules are efficient for de novo structure prediction, for probing heterogeneity of interactions with small molecules in the folded state, an entirely atomic-level description could have advantages in speed and generality. We developed a graph neural network called PLACER (protein-ligand atomistic conformational ensemble resolver) trained to recapitulate correct atomic positions from partially corrupted input structures from the Cambridge Structural Database and the Protein Data Bank; the nodes of the graph are the atoms in the system. PLACER accurately generates structures of diverse organic small molecules given knowledge of their atom composition and bonding. When given a description of the larger protein context, it builds up structures of small molecules and protein side chains for protein–small molecule docking. Because PLACER is rapid and stochastic, ensembles of predictions can be readily generated to map conformational heterogeneity. In enzyme design efforts described here and elsewhere, we find that using PLACER to assess the accuracy and preorganization of the designed active sites results in higher success rates and higher activities; we obtain a preorganized retroaldolase with a k cat / K M of 11,000 M −1 min −1 , considerably higher than any pre–deep learning design for this reaction. We anticipate that PLACER will be widely useful for rapidly generating conformational ensembles of small molecule and small molecule–protein systems and for designing higher activity preorganized enzymes.
- 4 November 2025
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 122 (45)
Open AccessModeling protein–small molecule conformational ensembles with PLACER
Modeling the conformational heterogeneity of protein–small molecule interactions is important for understanding natural systems and evaluating designed systems but remains an outstanding challenge. We reasoned that while residue-level descriptions of biomolecules are efficient for de novo structure prediction, for probing heterogeneity of interactions with small molecules in the folded state, an entirely atomic-level description could have advantages in speed and generality. We developed a graph neural network called PLACER (protein-ligand atomistic conformational ensemble resolver) trained to recapitulate correct atomic positions from partially corrupted input structures from the Cambridge Structural Database and the Protein Data Bank; the nodes of the graph are the atoms in the system. PLACER accurately generates structures of diverse organic small molecules given knowledge of their atom composition and bonding. When given a description of the larger protein context, it builds up structures of small molecules and protein side chains for protein–small molecule docking. Because PLACER is rapid and stochastic, ensembles of predictions can be readily generated to map conformational heterogeneity. In enzyme design efforts described here and elsewhere, we find that using PLACER to assess the accuracy and preorganization of the designed active sites results in higher success rates and higher activities; we obtain a preorganized retroaldolase with a k cat / K M of 11,000 M −1 min −1 , considerably higher than any pre–deep learning design for this reaction. We anticipate that PLACER will be widely useful for rapidly generating conformational ensembles of small molecule and small molecule–protein systems and for designing higher activity preorganized enzymes.
- Open Access
RNA sequence design and protein–DNA specificity prediction with NA-MPNN
RNA sequence design and protein–DNA binding specificity prediction can both be framed as nucleic acid inverse-folding problems: finding the most likely nucleic acid sequences given a fixed three-dimensional structure of a nucleic acid or nucleic acid–protein complex. While task-specific tools have been developed, no unified deep learning model for nucleic acid inverse folding has been described; a single model would have larger and more diverse datasets available for training and a considerably greater range of applicability. Here we introduce Nucleic Acid MPNN (NA-MPNN), a message-passing neural network that treats proteins, DNA, and RNA within a unified biopolymer graph representation. NA-MPNN outperforms previous methods on RNA sequence design and fixed-dock protein–DNA specificity prediction, and should be broadly useful for de novo RNA structure design and prediction of DNA-binding specificity.
Open AccessRNA sequence design and protein–DNA specificity prediction with NA-MPNN
RNA sequence design and protein–DNA binding specificity prediction can both be framed as nucleic acid inverse-folding problems: finding the most likely nucleic acid sequences given a fixed three-dimensional structure of a nucleic acid or nucleic acid–protein complex. While task-specific tools have been developed, no unified deep learning model for nucleic acid inverse folding has been described; a single model would have larger and more diverse datasets available for training and a considerably greater range of applicability. Here we introduce Nucleic Acid MPNN (NA-MPNN), a message-passing neural network that treats proteins, DNA, and RNA within a unified biopolymer graph representation. NA-MPNN outperforms previous methods on RNA sequence design and fixed-dock protein–DNA specificity prediction, and should be broadly useful for de novo RNA structure design and prediction of DNA-binding specificity.
- Top Cited PapersOpen Access
- 28 March 2025
- Published by Springer Nature in Nature Methods
- Vol. 22 (4) , 717-723
Atomic context-conditioned protein sequence design using LigandMPNN
Protein sequence design in the context of small molecules, nucleotides and metals is critical to enzyme and small-molecule binder and sensor design, but current state-of-the-art deep-learning-based sequence design methods are unable to model nonprotein atoms and molecules. Here we describe a deep-learning-based protein sequence design method called LigandMPNN that explicitly models all nonprotein components of biomolecular systems. LigandMPNN significantly outperforms Rosetta and ProteinMPNN on native backbone sequence recovery for residues interacting with small molecules (63.3% versus 50.4% and 50.5%), nucleotides (50.5% versus 35.2% and 34.0%) and metals (77.5% versus 36.0% and 40.6%). LigandMPNN generates not only sequences but also sidechain conformations to allow detailed evaluation of binding interactions. LigandMPNN has been used to design over 100 experimentally validated small-molecule and DNA-binding proteins with high affinity and high structural accuracy (as indicated by four X-ray crystal structures), and redesign of Rosetta small-molecule binder designs has increased binding affinity by as much as 100-fold. We anticipate that LigandMPNN will be widely useful for designing new binding proteins, sensors and enzymes.
- 28 March 2025
- Published by Springer Nature in Nature Methods
- Vol. 22 (4) , 717-723
Top Cited PapersOpen AccessAtomic context-conditioned protein sequence design using LigandMPNN
Protein sequence design in the context of small molecules, nucleotides and metals is critical to enzyme and small-molecule binder and sensor design, but current state-of-the-art deep-learning-based sequence design methods are unable to model nonprotein atoms and molecules. Here we describe a deep-learning-based protein sequence design method called LigandMPNN that explicitly models all nonprotein components of biomolecular systems. LigandMPNN significantly outperforms Rosetta and ProteinMPNN on native backbone sequence recovery for residues interacting with small molecules (63.3% versus 50.4% and 50.5%), nucleotides (50.5% versus 35.2% and 34.0%) and metals (77.5% versus 36.0% and 40.6%). LigandMPNN generates not only sequences but also sidechain conformations to allow detailed evaluation of binding interactions. LigandMPNN has been used to design over 100 experimentally validated small-molecule and DNA-binding proteins with high affinity and high structural accuracy (as indicated by four X-ray crystal structures), and redesign of Rosetta small-molecule binder designs has increased binding affinity by as much as 100-fold. We anticipate that LigandMPNN will be widely useful for designing new binding proteins, sensors and enzymes.
- Open Access
Computational design of bifaceted protein nanomaterials
Recent advances in computational methods have led to considerable progress in the design of self-assembling protein nanoparticles. However, nearly all nanoparticles designed to date exhibit strict point group symmetry, with each subunit occupying an identical, symmetrically related environment. This limits the structural diversity that can be achieved and precludes anisotropic functionalization. Here, we describe a general computational strategy for designing multi-component bifaceted protein nanomaterials with two distinctly addressable sides. The method centers on docking pseudosymmetric heterooligomeric building blocks in architectures with dihedral symmetry and designing an asymmetric protein-protein interface between them. We used this approach to obtain an initial 30-subunit assembly with pseudo-D5 symmetry, and then generated an additional 15 variants in which we controllably altered the size and morphology of the bifaceted nanoparticles by designing de novo extensions to one of the subunits. Functionalization of the two distinct faces of the nanoparticles with de novo protein minibinders enabled specific colocalization of two populations of polystyrene microparticles coated with target protein receptors. The ability to accurately design anisotropic protein nanomaterials with precisely tunable structures and functions could be broadly useful in applications that require colocalizing two or more distinct target moieties.
Open AccessComputational design of bifaceted protein nanomaterials
Recent advances in computational methods have led to considerable progress in the design of self-assembling protein nanoparticles. However, nearly all nanoparticles designed to date exhibit strict point group symmetry, with each subunit occupying an identical, symmetrically related environment. This limits the structural diversity that can be achieved and precludes anisotropic functionalization. Here, we describe a general computational strategy for designing multi-component bifaceted protein nanomaterials with two distinctly addressable sides. The method centers on docking pseudosymmetric heterooligomeric building blocks in architectures with dihedral symmetry and designing an asymmetric protein-protein interface between them. We used this approach to obtain an initial 30-subunit assembly with pseudo-D5 symmetry, and then generated an additional 15 variants in which we controllably altered the size and morphology of the bifaceted nanoparticles by designing de novo extensions to one of the subunits. Functionalization of the two distinct faces of the nanoparticles with de novo protein minibinders enabled specific colocalization of two populations of polystyrene microparticles coated with target protein receptors. The ability to accurately design anisotropic protein nanomaterials with precisely tunable structures and functions could be broadly useful in applications that require colocalizing two or more distinct target moieties.
- Open Access
Modeling protein-small molecule conformational ensembles with PLACER
Modeling the conformational heterogeneity of protein-small molecule interactions is important for understanding natural systems and evaluating designed systems, but remains an outstanding challenge. We reasoned that while residue level descriptions of biomolecules are efficient for de novo structure prediction, for probing heterogeneity of interactions with small molecules in the folded state an entirely atomic level description could have advantages in speed and generality. We developed a graph neural network called PLACER (Protein-Ligand Atomistic Conformational Ensemble Resolver) trained to recapitulate correct atomic positions from partially corrupted input structures from the Cambridge Structural Database and the Protein Data Bank; the nodes of the graph are the atoms in the system. PLACER accurately generates structures of diverse organic small molecules given knowledge of their atom composition and bonding, and given a description of the larger protein context, builds up structures of small molecules and protein side chains for protein-small molecule docking. Because PLACER is rapid and stochastic, ensembles of predictions can be readily generated to map conformational heterogeneity. In enzyme design efforts described here and elsewhere, we find that using PLACER to assess the accuracy and pre-organization of the designed active sites results in higher success rates and higher activities; we obtain a preorganized retroaldolase with a k cat / K M of 11000 M -1 min -1 , considerably higher than any pre-deep learning design for this reaction. We anticipate that PLACER will be widely useful for rapidly generating conformational ensembles of small molecule and small molecule-protein systems, and for designing higher activity preorganized enzymes.
Open AccessModeling protein-small molecule conformational ensembles with PLACER
Modeling the conformational heterogeneity of protein-small molecule interactions is important for understanding natural systems and evaluating designed systems, but remains an outstanding challenge. We reasoned that while residue level descriptions of biomolecules are efficient for de novo structure prediction, for probing heterogeneity of interactions with small molecules in the folded state an entirely atomic level description could have advantages in speed and generality. We developed a graph neural network called PLACER (Protein-Ligand Atomistic Conformational Ensemble Resolver) trained to recapitulate correct atomic positions from partially corrupted input structures from the Cambridge Structural Database and the Protein Data Bank; the nodes of the graph are the atoms in the system. PLACER accurately generates structures of diverse organic small molecules given knowledge of their atom composition and bonding, and given a description of the larger protein context, builds up structures of small molecules and protein side chains for protein-small molecule docking. Because PLACER is rapid and stochastic, ensembles of predictions can be readily generated to map conformational heterogeneity. In enzyme design efforts described here and elsewhere, we find that using PLACER to assess the accuracy and pre-organization of the designed active sites results in higher success rates and higher activities; we obtain a preorganized retroaldolase with a k cat / K M of 11000 M -1 min -1 , considerably higher than any pre-deep learning design for this reaction. We anticipate that PLACER will be widely useful for rapidly generating conformational ensembles of small molecule and small molecule-protein systems, and for designing higher activity preorganized enzymes.
- Top Cited PapersOpen Access
- 19 July 2024
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 385 (6706) , 276-282
Binding and sensing diverse small molecules using shape-complementary pseudocycles
We describe an approach for designing high-affinity small molecule–binding proteins poised for downstream sensing. We use deep learning–generated pseudocycles with repeating structural units surrounding central binding pockets with widely varying shapes that depend on the geometry and number of the repeat units. We dock small molecules of interest into the most shape complementary of these pseudocycles, design the interaction surfaces for high binding affinity, and experimentally screen to identify designs with the highest affinity. We obtain binders to four diverse molecules, including the polar and flexible methotrexate and thyroxine. Taking advantage of the modular repeat structure and central binding pockets, we construct chemically induced dimerization systems and low-noise nanopore sensors by splitting designs into domains that reassemble upon ligand addition.
- 19 July 2024
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 385 (6706) , 276-282
Top Cited PapersOpen AccessBinding and sensing diverse small molecules using shape-complementary pseudocycles
We describe an approach for designing high-affinity small molecule–binding proteins poised for downstream sensing. We use deep learning–generated pseudocycles with repeating structural units surrounding central binding pockets with widely varying shapes that depend on the geometry and number of the repeat units. We dock small molecules of interest into the most shape complementary of these pseudocycles, design the interaction surfaces for high binding affinity, and experimentally screen to identify designs with the highest affinity. We obtain binders to four diverse molecules, including the polar and flexible methotrexate and thyroxine. Taking advantage of the modular repeat structure and central binding pockets, we construct chemically induced dimerization systems and low-noise nanopore sensors by splitting designs into domains that reassemble upon ligand addition.
- Open Access
- 19 March 2024
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 121 (13)
Rapid and automated design of two-component protein nanomaterials using ProteinMPNN
The design of protein–protein interfaces using physics-based design methods such as Rosetta requires substantial computational resources and manual refinement by expert structural biologists. Deep learning methods promise to simplify protein–protein interface design and enable its application to a wide variety of problems by researchers from various scientific disciplines. Here, we test the ability of a deep learning method for protein sequence design, ProteinMPNN, to design two-component tetrahedral protein nanomaterials and benchmark its performance against Rosetta. ProteinMPNN had a similar success rate to Rosetta, yielding 13 new experimentally confirmed assemblies, but required orders of magnitude less computation and no manual refinement. The interfaces designed by ProteinMPNN were substantially more polar than those designed by Rosetta, which facilitated in vitro assembly of the designed nanomaterials from independently purified components. Crystal structures of several of the assemblies confirmed the accuracy of the design method at high resolution. Our results showcase the potential of deep learning–based methods to unlock the widespread application of designed protein–protein interfaces and self-assembling protein nanomaterials in biotechnology.
- 19 March 2024
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 121 (13)
Open AccessRapid and automated design of two-component protein nanomaterials using ProteinMPNN
The design of protein–protein interfaces using physics-based design methods such as Rosetta requires substantial computational resources and manual refinement by expert structural biologists. Deep learning methods promise to simplify protein–protein interface design and enable its application to a wide variety of problems by researchers from various scientific disciplines. Here, we test the ability of a deep learning method for protein sequence design, ProteinMPNN, to design two-component tetrahedral protein nanomaterials and benchmark its performance against Rosetta. ProteinMPNN had a similar success rate to Rosetta, yielding 13 new experimentally confirmed assemblies, but required orders of magnitude less computation and no manual refinement. The interfaces designed by ProteinMPNN were substantially more polar than those designed by Rosetta, which facilitated in vitro assembly of the designed nanomaterials from independently purified components. Crystal structures of several of the assemblies confirmed the accuracy of the design method at high resolution. Our results showcase the potential of deep learning–based methods to unlock the widespread application of designed protein–protein interfaces and self-assembling protein nanomaterials in biotechnology.
- Open Access
Atomic context-conditioned protein sequence design using LigandMPNN
Protein sequence design in the context of small molecules, nucleotides, and metals is critical to enzyme and small molecule binder and sensor design, but current state-of-the-art deep learning-based sequence design methods are unable to model non-protein atoms and molecules. Here, we describe a deep learning-based protein sequence design method called LigandMPNN that explicitly models all non-protein components of biomolecular systems. LigandMPNN significantly outperforms Rosetta and ProteinMPNN on native backbone sequence recovery for residues interacting with small molecules (63.3% vs. 50.4% & 50.5%), nucleotides (50.5% vs. 35.2% & 34.0%), and metals (77.5% vs. 36.0% & 40.6%). LigandMPNN generates not only sequences but also sidechain conformations to allow detailed evaluation of binding interactions. Experimental characterization demonstrates that LigandMPNN can generate small molecule and DNA-binding proteins with high affinity and specificity. One-sentence summary: We present a deep learning-based protein sequence design method that allows explicit modeling of small molecule, nucleotide, metal, and other atomic contexts.
Open AccessAtomic context-conditioned protein sequence design using LigandMPNN
Protein sequence design in the context of small molecules, nucleotides, and metals is critical to enzyme and small molecule binder and sensor design, but current state-of-the-art deep learning-based sequence design methods are unable to model non-protein atoms and molecules. Here, we describe a deep learning-based protein sequence design method called LigandMPNN that explicitly models all non-protein components of biomolecular systems. LigandMPNN significantly outperforms Rosetta and ProteinMPNN on native backbone sequence recovery for residues interacting with small molecules (63.3% vs. 50.4% & 50.5%), nucleotides (50.5% vs. 35.2% & 34.0%), and metals (77.5% vs. 36.0% & 40.6%). LigandMPNN generates not only sequences but also sidechain conformations to allow detailed evaluation of binding interactions. Experimental characterization demonstrates that LigandMPNN can generate small molecule and DNA-binding proteins with high affinity and specificity. One-sentence summary: We present a deep learning-based protein sequence design method that allows explicit modeling of small molecule, nucleotide, metal, and other atomic contexts.