Organizations: Stanford University · HKUST (Guangzhou) · Amazon · SJTU · University of Chicago · University of Notre Dame · HKUST · Nanjing University · CISPA · CUHK · Yale University · Universität Hamburg · Lexicodon
Protein-protein interactions underpin nearly all biological processes. Yet, designing proteins with desired interactions remains challenging due to the structural complexity governing molecular recognition. This study views protein design as a search optimization problem over a combinatorial sequence space under expensive structural evaluation. Toward this goal, we introduce a novel tree-based search framework, ProteoCraft, that employs approximate structural gradients from diffusion-based structure predictors as effective proposal policies to specify where the search should explore. Complete no-gradient inference of folding models separately assigns values to the resulting sequences, determining what is good. ProteoCraft allocates this verification budget through a coarse-to-fine paradigm: multi-site edits first explore distinct sequence basins, then verified promising states undergo strict single-site refinement. Across fourteen de novo binder-design settings spanning eight target proteins, five immune-relevant TCR-facing targets, and four localized CDR redesign tasks, ProteoCraft improves design quality and high-confidence yield. These improvements persist under independent AF2-IG and Rosetta-based interface evaluation. Prospective experimental validation further yielded an 82.9% binding hit rate for PD-L1 miniproteins, with a best measured KD of 0.879 nM, and confirmed binding of designed IFNAR2 VHH. Our code is available at https://github.com/smiles724/ProteoCraft_v1.
Figures & tables
Figure 1: Design quality versus computational cost.
Figure 2: Overview. ProteoCraft formulates protein design as proposal–verification search. Given a mutable design region and molecular context, a coarse-to-fine tree search uses truncated diffusion signals to propose candidate sequences, while complete ESMFold2 inference assigns node values.
Method
Mean confidence ↑
High-confidence yield ↑
Pair iPTM
iPTM
pTM
pLDDT
Binder pLDDT
Pair iPTM ≥0.5
Pair iPTM ≥0.7
Pair iPTM ≥0.75 Binder pLDDT ≥0.9
RFdiffusion v3
0.206
0.282
0.593
0.739
0.729
20 (9.5%)
1 (0.5%)
0 (0.0%)
BoltzGen
0.438
0.577
0.748
0.826
0.857
105 (50.0%)
45 (21.4%)
9 (4.3%)
Proteina-Complexa
0.286
0.386
0.653
0.774
0.779
48 (22.9%)
4 (1.9%)
0 (0.0%)
PXDesign-d
0.322
0.433
0.683
0.801
0.832
58 (27.6%)
16 (7.6%)
1 (0.5%)
BoltzDesign1
0.154
0.220
0.562
0.638
0.490
6 (2.9%)
0 (0.0%)
0 (0.0%)
Table 1: ESMFold2 evaluation of de novo binder design on 14 target–hotspot settings across eight target proteins. Each method contributes 15 candidates per hotspot ( N=210 ). The left block reports mean confidence metrics (higher is better); the right block reports high-confidence design yields as count (percentage). The strict criterion requires both Pair iPTM ≥0.75 and Binder pLDDT ≥0.9 . Dashes indicate results not yet reported.
AF2-IG
Rosetta interface evaluation
Method
iPAE ↓
iPTM ↑
pTM ↑
Strict Pass ↑
ΔG↓
ΔSASA↑
SC ↑
ΔG/ΔSASA↓
H-bonds ↑
Packstat ↑
RFdiffusion v3
25.426
0.142
0.550
0 (0.0%)
-49.398
1834.2
0.633
-2.696
4.662
0.612
BoltzGen
26.342
0.200
0.689
0 (0.0%)
-31.138
1902.4
0.558
-1.640
6.463
0.654
Proteina-Complexa
18.927
0.359
0.651
6 (2.9%)
-28.741
1626.0
0.557
-1.671
5.314
0.597
PXDesign-d
18.949
0.349
0.650
20 (9.5%)
-50.477
2029.6
0.603
-2.469
8.595
0.597
BoltzDesign1
19.967
0.369
0.571
1 (0.5%)
-52.903
3251.0
0.607
-1.611
7.743
0.542
Table 2: Independent AF2-IG and Rosetta evaluation of de novo designs. Each was evaluated on N=210 designs. For AF2-IG, lower iPAE and higher iPTM, pTM, and strict pass rate are better. For Rosetta, lower ΔG and ΔG/ΔSASA and higher values for the remaining metrics are better. Bold and underlined values indicate the best and second-best results, respectively.
ESMFold2
AF2-IG
Method
Pair iPTM ↑
iPTM ↑
pTM ↑
pLDDT ↑
iPAE ↓
iPTM ↑
pTM ↑
RFdiffusion v3
0.225
0.695
0.780
0.852
27.540
0.161
0.693
BoltzGen
0.384
0.763
0.824
0.843
26.342
0.200
0.689
Proteina-Complexa
0.300
0.756
0.822
0.863
26.439
0.191
0.702
PXDesign-d
0.276
0.717
0.795
0.873
26.115
0.207
0.708
BoltzDesign1
0.228
0.734
0.810
0.817
26.165
0.214
0.713
Table 3: TCR-facing binder design. The left block reports complete ESMFold2 verification and the right block reports independent AF2-IG evaluation.
Method
Case
N
Peptide Pair ↑
Mean Pair ↑
iPTM ↑
pTM ↑
Complex pLDDT ↑
Chain pLDDT ↑
RFdiffusion v3
1QRN α
10
0.591
0.686
0.872
0.896
0.919
0.901
1QRN β
10
0.421
0.560
0.824
0.857
0.905
0.912
9NMW α
10
0.549
0.669
0.884
0.900
0.906
0.896
9NMW β
10
0.579
0.687
0.909
0.920
0.914
0.918
Mean
–
0.535
0.650
0.872
0.893
0.911
0.907
BoltzGen
1QRN α
10
0.715
0.761
0.926
0.934
0.929
0.912
Table 4: ESMFold2 evaluation for localized CDR redesign. Peptide Pair iPTM measures confidence between the redesigned chain and the peptide; Mean Pair iPTM averages pairwise confidence between the redesigned chain and all fixed context chains. All metrics are higher-is-better.
AF2-IG
Rosetta
Method
iPAE ↓
iPTM ↑
pTM ↑
pLDDT ↑
ΔG↓
SC ↑
H-bonds ↑
Packstat ↑
RFdiffusion v3
17.483
0.505
0.665
0.871
-103.9
0.681
14.200
0.488
BindCraft
23.294
0.360
0.602
0.642
-78.3
0.571
14.750
0.618
ProteoCraft-MCTS
14.558
0.753
0.716
0.851
-138.3
0.702
24.575
0.673
Table 5: Independent cross-evaluation of localized CDR redesign.
Figure 3: Wet-lab validation of ProteoCraft-designed binders. Complex structures and biolayer interferometry sensorgrams for the selected de novo binders against PD-L1 (left) and IFNAR2 (right).
Figure 4: Accelerating ProteoCraft through inference optimization.
Method
De novo
TCR
PXDesign-d
18.72
21.01
Proteina-Complexa
14.62
34.61
RFdiffusion v3
56.64
44.26
BoltzDesign1
693.38
2882.20
BindCraft v2
2321.50
—
ProteoCraft
32.34
174.06
Table 6: Runtime comparison. We report mean wall-clock latency (seconds) per binder under each method’s native design and selection procedure. Details are in Appendix C.3 .
Table 8: Hotspot definitions for eight de novo target proteins and five TCR-facing target systems.
PDB / case
TCR target
Antigen / restriction
1QRN
A6 TCR
TAX-A2
3QDJ
DMF5 TCR
MART1-A2
5BS0
A3A TCR
MAGEA3-A1 / Titin-A1
6BJ8
TCR55
HIV-B35
9C3E
1G4 TCR
NYESO-A2
Appendix
Table 9: TCR-facing binder-design systems.
Case
CDR3 sequence
Residue range
9NMW α
CAVNNARLMFGDGTQL
90–105
9NMW β
CASSVAGSPEAFF
89–101
1QRN α
CAVTTDSWGKLQF
88–100
1QRN β
CASRPGLAGGRPEQYF
91–106
Appendix
Table 10: Localized CDR redesign settings.
Target
PXDesign-d
RFdiffusion v3
BoltzDesign1
No One-step
Base / Best-of- N
ProteoCraft-MCTS
BetV1
0.257
0.185
0.087
0.793
0.725
0.832
DerF21-B10
0.462
0.241
0.214
0.698
0.746
0.771
DerF21-B28
0.449
0.170
0.099
0.630
0.729
0.770
DerF21-B63
0.378
0.206
0.140
0.699
0.738
0.787
DerF21-B81
0.392
0.140
0.178
0.600
0.725
0.756
DerF21-B92
0.340
0.194
0.219
0.716
0.695
0.746
Appendix
Table 11: Mean ESMFold2 Pair-chain iPTM across the 15 candidates for each matched de novo target. Higher is better.
Case
Total Length
PXDesign-d
RFdiffusion3
Complexa BoN+AF2
BoltzDesign1
BindCraft v2
ProteoCraft
De novo binder design
IFNAR2
199
13.53
49.21
14.28
458
512
11.62
PD-L1
208
14.13
49.40
14.10
504
303
13.54
PD1
220
16.40
63.64
12.05
505
1128
14.96
DerF21
257
11.49
56.46
15.05
522
253
17.17
BetV1
261
14.40
48.17
16.05
686
11523
19.76
Appendix
Table 12: Runtime comparison across different targets, reported in seconds per binder. Averages are computed separately over eight de novo and five TCR cases. Besides, — indicates no filter-passing designs within the TCR sampling budget for BindCraft v2.
Figure 5: Design quality with accelerated ESMFold2 inference.
τ
Pair iPTM ↑
iPTM ↑
pTM ↑
pLDDT ↑
1
0.682
0.871
0.896
0.886
2
0.777
0.930
0.939
0.921
4
0.730
0.914
0.931
0.909
Appendix
Table 13: Sensitivity to the one-step diffusion noise level. All settings use the same three targets and matched search configuration.
A central challenge in de novo protein design is generating proteins with desired structural and functional properties. Existing design models typically require a dedicated input representation for each property, limiting the design objectives they support. We develop Prompteina, the first all-atom structure generation model that supports new and broader design objectives specified in natural language, analogous to text-to-image generation. To train Prompteina, we construct Prompteina-Data, a paired text--protein structure dataset of 1.8M structures and 50M+ captions describing global and residue-specific structural and functional properties. We adapt Proteina-Complexa, a pretrained structure generation model, through text-conditioned pretraining and finetuning that aligns residues with their corresponding text. Prompteina outperforms baselines on 15 of 18 structural properties and generates proteins that more closely match descriptions of their intended functions. In binder design, it leads baselines on at least 13 of 20 targets and improves interface confidence by 14% over the strongest baseline. With iterative in silico feedback, Prompteina's gain over non-iterative generation is 57% larger than the next-best baseline on the hardest targets. Finally, we experimentally validate Prompteina designs against GM2A, a target with no reported de novo binders. We identify nine binders, with the strongest at an apparent KD of 107 nM. These results establish natural language as an effective and powerful conditioning interface for protein structure generation.
Joseph Lee, Yanjing Li, Bo Qiang +7
University of Washington · National University of Singapore · Allen Institute for AI (Ai2) +1
All-atom structure predictors model diverse molecular interactions, but using their learned structural priors for binder design remains challenging. Here we present TorchCraft, a unified binder-design framework that optimizes sequence logits through a frozen all-atom predictor. Implemented in TorchFold, TorchCraft combines confidence, contact, geometric, and sequence-prior objectives within a shared optimization procedure for minibinders, framework-conditioned VHHs, cyclic peptides, and ligand-binding proteins. Using pretrained AlphaFold 3 weights, TorchCraft generated representative minibinders and VHHs with experimentally measured binding across four targets in each format, without post hoc sequence redesign. Computational benchmarks further demonstrated the framework's applicability to cyclic peptides and ligand-conditioned pocket design. TorchCraft extends predictor inversion to multiple binder formats and molecular contexts, providing a common framework for reusing all-atom structural priors in design.
Protein interaction modeling is central to protein design, which has been transformed by machine learning with applications in drug discovery and beyond. In this landscape, structure-based de novo binder design is cast as either conditional generative modeling or sequence optimization via structure predictors ("hallucination"). We argue that this is a false dichotomy and propose Proteina-Complexa, a novel fully atomistic binder generation method unifying both paradigms. We extend recent flow-based latent protein generation architectures and leverage the domain-domain interactions of monomeric computationally predicted protein structures to construct Teddymer, a new large-scale dataset of synthetic binder-target pairs for pretraining. Combined with high-quality experimental multimers, this enables training a strong base model. We then perform inference-time optimization with this generative prior, unifying the strengths of previously distinct generative and hallucination methods. Proteina-Complexa sets a new state of the art in computational binder design benchmarks: it delivers markedly higher in-silico success rates than existing generative approaches, and our novel test-time optimization strategies greatly outperform previous hallucination methods under normalized compute budgets. We also demonstrate interface hydrogen bond optimization, fold class-guided binder generation, and extensions to small molecule targets and enzyme design tasks, again surpassing prior methods. Code, models and new data will be publicly released.
Kieran Didi, Zuobai Zhang, Guoqing Zhou +11
NVIDIA · University of Oxford · Mila - Québec AI Institute +8