RiboBridge: protein-conditioned de novo RNA binder structure design
Authors: Zhou Zhang, Hanqun Cao, Minghao Sun, Odin Zhang, Yang Zhang, Pheng Ann Heng, Wujun Li, Tianfan Fu
Organizations: State Key Laboratory for Novel Software Technology at Nanjing University, School of Computer Science, Nanjing University · The Chinese University of Hong Kong · National University of Singapore
Designing RNA molecules that bind to target protein structures requires generating both an RNA conformation and its bound complex geometry, without assuming a predefined binding site. Existing approaches generally model these coupled processes within a unified representation, entangling RNA conformation generation with protein-relative placement. In this work, we introduce RiboBridge, a structure-first framework that separates RNA conformational generation from protein--RNA assembly through cross-frame interaction geometry. RiboBridge first proposes multiple candidate RNA-binding interfaces on the protein surface, generates canonical RNA conformations within a pretrained RNA structural latent space (RiboSphere), and then predicts protein--RNA interaction constraints to bridge the independently defined RNA and protein coordinate frames. The resulting interaction geometry is converted into a geometrically consistent protein-relative pose through rigid placement. We evaluate RiboBridge on a temporally split, sequence-filtered, and redundancy-controlled benchmark of 124 protein--RNA complexes. Under post-hoc interface-correspondence evaluation, where native information is used only after generation to select among already generated predicted-pocket candidates, RiboBridge achieves the highest fNAT and DockQ and the lowest interface and ligand RMSDs among compared methods. Component ablations and stage-wise diagnostics further characterize the contributions of RNA conformational generation, interaction prediction, and rigid placement, identifying the main sources of complex-recovery error.
Table 1: Complex generation performance on the test set under method-specific conditioning and post-hoc interface correspondence. Each method follows the conditioning supported by its released pipeline; baseline-specific settings are detailed in Appendix B.3 . RMSD values are reported in Å.
Method
Seq. source
Rec. ↑
scRMSD (Å) ↓
scLRMSD (Å) ↓
scTM ↑
Des. (%) ↑
ODesign
Direct output
0.4619
10.76
33.91
0.0343
6.45
RFDpoly
NA-MPNN
0.4655
10.90
42.76
0.0638
8.06
RNAFlow
Direct output
0.3781
9.31
48.24
0.0370
6.45
NACraft
Direct output
0.4733
10.69
41.86
0.0682
7.26
RiboBridge
NA-MPNN
0.4683
10.14
40.98
0.0783
8.06
Table 2: Sequence realizability and AlphaFold 3 self-consistency on the 124-target test set. Rec. denotes sequence recovery. Designability is the percentage of all 124 test targets that yield an AF3-valid design with RNA scRMSD below 2 Å.
Variant
Atom11 RMSD (Å) ↓
iRMSD (Å) ↓
fNAT ↑
DockQ ↑
Absolute performance
Full (pred. matched)
7.972
4.134
0.4584
0.4691
No self-correction
8.054
4.163
0.4500
0.4659
Length-only
9.459
4.984
0.3231
0.3622
Shuffled pocket
9.420
4.798
0.3296
0.3745
Target-paired 95% CI of Full advantage
Table 3: Ablation of pocket-conditioned canonical RNA generation. All variants are evaluated on the same 54-target formally matched subset. The lower block reports target-paired bootstrap 95% confidence intervals for the advantage of the full model, with positive values favoring Full for all metrics.
Figure 2: Error decomposition of RiboBridge. Results are computed on the 54-target diagnostic subset satisfying the formal interface-matching criterion. (a) Stage-wise failure-aware DockQ and pose observability. (b) Paired DockQ changes between consecutive stages. Filled markers include unobservable failures; open markers compare targets that are observable at both stages.
Variant
F1@5 Å ↑
F1@8 Å ↑
Dist. MAE (Å) ↓
Pose RMSD (Å) ↓
DockQ ↑
Full
0.272
0.388
3.571
21.92
0.253
− local geometry
0.235
0.331
3.793
22.10
0.241
− triangle reasoning
0.263
0.348
3.683
24.58
0.250
CE-only
0.227
0.353
3.573
24.66
0.246
Table 4: Ablation of cross-frame interaction prediction. Contact F1 and distogram MAE evaluate predictions before rigid placement; pose RMSD and DockQ evaluate the resulting complexes. CE-only removes the explicit 5 Å and 8 Å contact objectives. Pose RMSD and DockQ are computed using the fixed reference RBL solver without nonlinear fallback.
Figure 3: Predicted-pocket-conditioned RNA generation. (A–C) Distinct RNA binding solutions generated from two predicted pockets. (D) Length-conditioned RNA generation at a fixed predicted pocket on a separate target.
Appendix figures & tables13 assets
Supplementary material from the paper’s appendix.
Appendix
Setting
Value
Protein encoder width / GVP layers
768/4
Protein graph neighbors
30
Interface queries / query-decoder layers
8/3
Proposal refinement layers
2
Interface probability threshold
0.60
Interface support size
6 – 128
Appendix
Table 5: Architecture and inference settings for interface hypothesis generation and canonical RNA generation.
Setting
Value
Per-chain hidden dimension
128
Per-chain message-passing layers
4
Triangle reasoning stages
20
Protein context anchors
at most 16
Local shape neighborhood size
8
Distogram classes / maximum finite range
37/20 Å
Appendix
Table 6: Architecture and inference settings for cross-frame interaction modeling and rigid placement.
Method
Protein input
Site information
RNA information
Candidate selection
ODesign
Structure
Native-derived hotspot
Requested length
Single output
RFDpoly
Structure
No explicit hotspot
Requested length
Single output
RNAFlow
Structure crop
Native-RNA-guided crop
Requested length
Single output
NACraft
Sequence
No explicit hotspot
Requested length
Highest model confidence
RiboBridge
Structure
Predicted interface
Requested length
Matched hypothesis
Appendix
Table 7: Conditioning information used by each method in the main comparison. For RiboBridge, “Matched hypothesis” means that predicted-interface candidates are generated independently before a correspondence with the native interface is established for evaluation.
Method
RNA sequence source
ODesign
Direct model output
RFDpoly
NA-MPNN
RNAFlow
Direct model output
NACraft
Direct model output
RiboBridge
NA-MPNN
Appendix
Table 8: RNA sequence sources used in the AlphaFold 3 self-consistency evaluation.
Quantity
Value
Raw learned queries / target
8
Test targets
124
Total retained proposals
202
Mean proposals / target
1.629
Median proposals / target
2
Range
1–3
Appendix
Table 9: Retained interface hypotheses in the blind 124-target evaluation.
Interface budget
Matched targets
Recall
Top-1
45/124
36.29%
Top-2
53/124
42.74%
All retained (Top-3)
54/124
43.55%
Appendix
Table 10: Formal native-interface coverage as the retained proposal budget increases.
Selection
Mean Dice
Median Dice
Median center dist. (Å)
Rank-1 proposal
0.503
0.619
10.72
Best-overlap retained proposal
0.573
0.647
8.42
Formal matched proposal
0.835
0.862
2.34
Appendix
Table 11: Correspondence between predicted interfaces and the expanded native interaction region.
Stage
RNA source
Interaction
Placement
Oracle input
Native/native
Native
Native
Native
Full complex
GT-token recon.
GT-token recon.
–
Oracle pose
Tokens + pose
Generated canonical
Generated
–
Oracle pose
Pose
Exact dist. + RBL
Generated
Exact distances
RBL
Distances
Oracle 37-bin + RBL
Generated
Oracle bins
RBL
Distance bins
Predicted map
Generated
Predicted
RBL
None
Appendix
Table 12: Stages used for error decomposition. Consecutive stages replace one source of oracle information with its predicted counterpart.
Stage
Obs.
DockQ
fNAT
iRMSD
lRMSD
Pose RMSD (mean/med.)
Native/native
54/54
1.000
1.000
≈0
0
0/0
GT-token recon.
54/54
0.854
0.834
0.858
1.319
0/0
Generated RNA
54/54
0.464
0.461
4.183
7.728
0/0
Exact dist. + RBL
53/54
0.446
0.441
4.261
7.804
0.011/0.007
Oracle bins + RBL
53/54
0.445
0.453
4.261
8.217
0.888/0.100
Predicted map
44/54
0.253
0.232
4.206
22.869
21.920/21.021
Appendix
Table 13: Stage-wise diagnostics on the 54-target formally matched subset. DockQ is failure-aware; all RMSDs are in Å.
Transition
Isolated source
Δ DockQ loss (95% CI)
Δ observable
Native → GT-token recon.
RiboSphere reconstruction
0.1465 [0.1193,0.1764]
0
GT-token → generated canonical
Canonical generation
0.3893 [0.3290,0.4487]
0
Oracle pose → exact-distance RBL
Rigid placement
0.0181 [0.0000,0.0537]
−1
Exact distance → oracle 37-bin
Distance discretization
0.0016 [−0.0056,0.0101]
0
Oracle 37-bin → predicted map
Interaction prediction
0.1920 [0.1502,0.2335]
−9
Reference RBL → fallback
Nonlinear recovery
−0.0018[−0.0043,0.0000]
+3
Appendix
Table 14: Target-paired changes between consecutive stages. Positive Δ DockQ denotes degradation from the upstream to downstream stage.
Figure 4: Qualitative examples of complex recovery and conformation–placement errors. (a–c) High-recovery complexes. (d–f) Examples with different degrees of RNA conformational and protein-relative pose deviation. Generated RNA is purple, native RNA is gray, and the local protein environment is pale green. RMSD measures RNA conformational deviation after alignment; placement error measures deviation of the recovered pose without an additional RNA superposition. DockQ evaluates recovery of the native complex.
Component
Trainable params.
Hardware
Training budget
Training time
Interface predictor
102.66M
8 GPUs
20 epochs
∼ 4 h
Canonical generator
38.5M
8 GPUs
40 epochs
∼ 4 h
Interaction predictor
13.3M
8 GPUs
50 epochs
∼ 13 h
Appendix
Table 15: Training cost of the three trainable components in RiboBridge.
Quantity
Runtime
Full 124-target run
1872.8 s
Mean / target
15.10 s
Median / target
11.01 s
10th percentile
6.06 s
90th percentile
28.33 s
Median, 1 retained pocket
6.47 s
Appendix
Table 16: End-to-end inference cost on the blind 124-target evaluation. Runtime includes pocket prediction and downstream generation for all retained interface hypotheses.
RNA function arises from the coupling of nucleotide sequence and three-dimensional structure, motivating their joint design. Coordinating global folding with nucleotide-level detail remains challenging under limited structural supervision. We introduce La-Ribo, a generative framework for RNA sequence-structure co-design via geometry-latent flow matching. La-Ribo retains a sparse phosphate-sugar--base scaffold and encodes nucleotide identity and local conformation in residue-wise latents. A shared flow network generates both jointly, and an RNA-specific decoder then reconstructs all heavy atoms. To expand supervision, we construct a quality-controlled corpus of 168,561 RNA structures, integrating experimental data with predictions from three folding models, including 10,631 MSA-supported structures generated in this work. La-Ribo improves designability and codesignability over the evaluated baselines across sampling budgets and two refolding models, and the same prior supports scaffold-conditioned inverse folding without additional training.
Runze Ma, Will Hua, Shuangjia Zheng
Shanghai Jiao Tong University · Independent Researcher
RNA molecules often function through multiple experimentally observed conformational states. A central challenge is to capture ensemble diversity without sacrificing structural fidelity: samples may collapse toward averaged representative structures or drift into unsupported conformational regions. We introduce REFLEX (RNA Ensemble generation calibrated by FLEXibility), a sequence-conditioned generator for RNA backbone-frame ensembles based on a heteroscedastic stochastic bridge on SE(3). During training, experimental B-factors calibrate residue-wise bridge widths, exposing each observed conformer through local perturbation neighborhoods with residue-specific scales while keeping constrained regions sharper. A learned flexibility condition guides the bridge-field network to recover target conformer-specific frames from these calibrated noisy states. At inference, REFLEX uses only the input sequence, requiring no MSA-subsampling heuristics. On a multi-conformer RNA benchmark, REFLEX achieves the best precision-recall trade-off for ensemble coverage across evaluation thresholds, while recovery-stress ablations indicate gains in target-frame recovery under stronger perturbations.
Accurate RNA structure modeling remains difficult because RNA backbones are highly flexible, non-canonical interactions are prevalent, and experimentally determined 3D structures are comparatively scarce. We introduce RiboSphere, a framework that learns discrete geometric representations of RNA by combining vector quantization with flow matching. Our design is motivated by the modular organization of RNA architecture: complex folds are composed from recurring structural motifs. RiboSphere uses a geometric transformer encoder trained using mean-centered coordinates and random rotation augmentation to produce geometry-aware features, which are discretized with finite scalar quantization (FSQ) into a finite vocabulary of latent codes. Conditioned on these discrete codes, a flow-matching decoder reconstructs atomic coordinates, enabling high-fidelity structure generation. We find that the learned code indices are enriched for specific RNA motifs, suggesting that the model captures motif-level compositional structure rather than acting as a purely compressive bottleneck. Across benchmarks, RiboSphere achieves strong performance in structure reconstruction (RMSD 1.25,Å, TM-score 0.84), and its pretrained discrete representations transfer effectively to inverse folding and RNA--ligand binding prediction, with robust generalization in data-scarce regimes. Code is available at https://github.com/Zhangz312/RiboSphere.
Zhou Zhang, Hanqun Cao, Cheng Tan +3
State Key Laboratory for Novel Software Technology at Nanjing University, School of Computer Science, Nanjing University · The Chinese University of Hong Kong · Stanford University