pith. sign in

arxiv: 2606.23830 · v1 · pith:ROCBVNK6new · submitted 2026-06-22 · 💻 cs.LG · cs.AI

Deciphering Fingerprints of 3D Molecular Surfaces for Accurate Epitope Prediction

Fang Wu , Weihao Xuan , Jure Leskovec , Yejin Choi , Li Erran Li This is my paper

Pith reviewed 2026-06-26 09:05 UTC · model grok-4.3

classification 💻 cs.LG cs.AI
keywords epitope predictionmolecular surfaceTransformerantibody-antigen interactionprotein surface modelingmachine learningconformational states
0
0 comments X

The pith

SurfBind predicts epitopes by processing 3D molecular surfaces directly with a Transformer that uses patch modeling and binder-aware cross-attention.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper develops SurfBind to predict epitopes, the parts of antigens that antibodies bind to, by using 3D molecular surface data rather than protein sequences or backbone structures. The method employs a Transformer with patch-level modeling of the surface, cross-attention that considers the binder, and a step-by-step coarse to fine prediction process. The goal is to better capture the geometric and chemical features on the surface that enable binding. Tests on datasets such as SAbDab and DB5.5 show it reaches top performance and works well even with antibodies and shapes not seen during training. This matters because epitopes are often discontinuous and surface-driven, which sequence methods miss.

Core claim

SurfBind is a surface-centric learning framework for epitope prediction that operates directly on molecular surface representations. It integrates geometric and physicochemical cues through a Transformer-based architecture with patch-level surface modeling, binder-aware cross-attention, and a hierarchical coarse-to-fine prediction paradigm. Experiments on challenging epitope identification benchmarks, including SAbDab and DB5.5, demonstrate that SurfBind achieves state-of-the-art performance and strong generalization across unseen antibodies and conformational states.

What carries the argument

The binder-aware cross-attention mechanism within a Transformer that processes patch-level molecular surface representations to integrate geometric and physicochemical cues.

If this is right

  • Improved epitope prediction enables more precise design of antibody-based therapies.
  • The framework shows that surface geometry matters more than linear sequence information for identifying binding sites.
  • Strong generalization indicates the method can handle novel protein complexes in drug discovery.
  • Hierarchical prediction allows efficient processing of complex 3D molecular structures.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Surface modeling approaches may extend to predicting other molecular interactions such as enzyme-substrate binding.
  • Integration with experimental data like cryo-EM could be tested to further refine predictions.
  • Similar surface-centric models could advance related tasks such as protein docking.

Load-bearing premise

Molecular surface representations processed through the Transformer with patch-level modeling and binder-aware cross-attention capture the geometric and physicochemical patterns that determine antibody-antigen recognition more effectively than sequence or backbone alternatives.

What would settle it

A new benchmark with diverse conformational states where SurfBind fails to outperform sequence-based methods on epitope identification would show that the surface approach does not capture the determining patterns.

Figures

Figures reproduced from arXiv: 2606.23830 by Fang Wu, Jure Leskovec, Li Erran Li, Weihao Xuan, Yejin Choi.

Figure 1
Figure 1. Figure 1: Schematic overview of our antigen-binding site prediction model. Firstly, the antigen surface is sampled into a point [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: (a, b) BCE vs. non-BCE ratio distributions based on unsupervised SurfBind representations using different codebook [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: (a) Performance comparison for antibody-specific BCE prediction on Sabdab. b. Case study visualization of four protein [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 2
Figure 2. Figure 2: As envisioned in Fig. 2 [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: a. Different approaches to acquiring antibody representations when antibody structures may be inaccessible. b. A [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The overall pipeline of our unsupervised SurfBind method. The input surface point cloud is first preprocessed [PITH_FULL_IMAGE:figures/full_fig_p014_5.png] view at source ↗
read the original abstract

Molecular surfaces encode the geometric and physicochemical patterns that determine antibody-antigen recognition, central to epitope prediction. However, existing methods rely on sequences or backbone structures and struggle to capture discontinuous, surface-driven epitopes. This study presents SurfBind, a surface-centric learning framework for epitope prediction that operates directly on molecular surface representations. SurfBind integrates geometric and physicochemical cues through a Transformer-based architecture with patch-level surface modeling, binder-aware cross-attention, and a hierarchical coarse-to-fine prediction paradigm. Experiments on challenging epitope identification benchmarks, including SAbDab and DB5.5, demonstrate that SurfBind achieves state-of-the-art performance and strong generalization across unseen antibodies and conformational states, highlighting the value of interaction-aware surface modeling for understanding the crucial mechanisms of protein-protein interactions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 0 minor

Summary. The manuscript introduces SurfBind, a surface-centric framework for epitope prediction that processes 3D molecular surface representations using a Transformer architecture incorporating patch-level modeling, binder-aware cross-attention, and a hierarchical coarse-to-fine prediction paradigm. It claims state-of-the-art performance and strong generalization across unseen antibodies and conformational states on benchmarks including SAbDab and DB5.5, arguing that surface representations better capture discontinuous epitopes than sequence or backbone-based methods.

Significance. If the empirical claims hold after proper validation, the work would provide evidence that direct modeling of molecular surfaces can improve epitope identification by capturing geometric and physicochemical interaction patterns, potentially shifting the field from sequence/backbone-centric approaches toward surface-aware methods for protein-protein interaction analysis.

major comments (2)
  1. [Abstract] Abstract: The central claim that SurfBind 'achieves state-of-the-art performance' on SAbDab and DB5.5 is unsupported because the text supplies no quantitative results, baselines, error bars, statistical tests, or performance numbers, rendering the claim unevaluable.
  2. [Experiments] Experiments section (implied by benchmark references): No details are provided on data splits, training procedures, evaluation protocols, ablation studies, or how generalization to unseen antibodies was ensured, which directly undermines assessment of whether reported improvements are independent of fitting choices or data selection.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment below and are prepared to revise the manuscript accordingly to improve clarity and support for our claims.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that SurfBind 'achieves state-of-the-art performance' on SAbDab and DB5.5 is unsupported because the text supplies no quantitative results, baselines, error bars, statistical tests, or performance numbers, rendering the claim unevaluable.

    Authors: We agree that the abstract does not contain specific quantitative results, baselines, or statistical details. While the full manuscript's Experiments section reports these comparisons (including metrics on SAbDab and DB5.5), we will revise the abstract to incorporate key performance numbers and baseline references to make the state-of-the-art claim directly supported within the abstract itself. revision: yes

  2. Referee: [Experiments] Experiments section (implied by benchmark references): No details are provided on data splits, training procedures, evaluation protocols, ablation studies, or how generalization to unseen antibodies was ensured, which directly undermines assessment of whether reported improvements are independent of fitting choices or data selection.

    Authors: The manuscript references the benchmarks and generalization to unseen antibodies, but we acknowledge that explicit details on data splits, training procedures, evaluation protocols, and ablation studies could be expanded for full transparency. We will revise the Experiments section to include these specifics, such as definitions of unseen antibody splits and ablation results, to allow independent assessment of the improvements. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper describes an empirical machine-learning framework (SurfBind) whose central claims rest on reported performance numbers on external public benchmarks (SAbDab, DB5.5). No derivation chain, equations, or self-citation load-bearing steps are present in the supplied text that would reduce a claimed result to a fitted input or self-definition by construction. The architecture choices are presented as design decisions whose value is asserted via experiment, not via internal tautology. This is the normal case for an applied ML paper whose validity is externally falsifiable on held-out data.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The central claim rests on empirical performance of a neural network whose parameters are fitted to benchmark data; the key domain assumption is that surface geometry and chemistry encode the decisive recognition signals.

free parameters (2)
  • Transformer weights and attention parameters
    Neural network parameters fitted during training on epitope benchmark data.
  • Patch definition and hierarchical prediction thresholds
    Design choices that control surface discretization and coarse-to-fine stages.
axioms (1)
  • domain assumption Molecular surfaces encode the geometric and physicochemical patterns that determine antibody-antigen recognition
    Stated as the central premise motivating the surface-centric approach.

pith-pipeline@v0.9.1-grok · 5670 in / 1247 out tokens · 25376 ms · 2026-06-26T09:05:05.765447+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

91 extracted references · 4 linked inside Pith

  1. [1]

    Ramisa Alam, Sazan Mahbub, and Md Shamsuzzoha Bayzid. 2023. Pair-EGRET: enhancing the prediction of protein-protein interaction sites through graph attention networks and protein language models.bioRxiv(2023), 2023–12

    2023

  2. [2]

    Hifzur Rahman Ansari and Gajendra PS Raghava. 2010. Identification of confor- mational B-cell Epitopes in an antigen from its primary sequence.Immunome research6, 1 (2010), 1–9

    2010

  3. [3]

    Sankar Basu and Björn Wallner. 2016. DockQ: a quality measure for protein- protein docking models.PloS one11, 8 (2016), e0161879

    2016

  4. [4]

    SYED NISAR HUSSAIN BUKHARI, MUNEER AHMAD DAR, and MUJTABA SHAFI. 2021. USING RANDOM FOREST TO PREDICT T-CELL EPITOPES OF DENGUE VIRUS. (2021)

    2021

  5. [5]

    Syed Nisar Hussain Bukhari, Amit Jain, Ehtishamul Haq, Abolfazl Mehbodniya, and Julian Webber. 2022. Machine learning techniques for the prediction of B-cell and T-cell epitopes as potential vaccine targets with a specific focus on SARS-CoV-2 pathogen: A review.Pathogens11, 2 (2022), 146

    2022

  6. [6]

    Guangyan Chen, Meiling Wang, Yi Yang, Kai Yu, Li Yuan, and Yufeng Yue. 2023. PointGPT: Auto-regressively Generative Pre-training from Point Clouds.arXiv preprint arXiv:2305.11487(2023)

    arXiv 2023

  7. [7]

    Junwei Chen, Bowen Zhao, Shenggeng Lin, Heqi Sun, Xueying Mao, Meng Wang, Yanyi Chu, Liang Hong, Dong-Qing Wei, Min Li, et al. 2024. TEPCAM: Prediction of T-cell receptor–epitope binding specificity via interpretable deep learning. Protein Science33, 1 (2024), e4841

    2024

  8. [8]

    Gabriel Cia, Fabrizio Pucci, and Marianne Rooman. 2023. Critical review of conformational B-cell epitope prediction methods.Briefings in bioinformatics24, 1 (2023), bbac567

    2023

  9. [9]

    Joakim Noddeskov Clifford, Magnus Haraldson Hoie, Sebastian Deleuran, Bjoern Peters, Morten Nielsen, and Paolo Marcatili. 2022. BepiPred-3.0: Improved B-cell epitope prediction using protein language models.Protein Science31, 12 (2022), e4497

    2022

  10. [10]

    Bruna Moreira da Silva, YooChan Myung, David B Ascher, and Douglas EV Pires

  11. [11]

    epitope3D: a machine learning method for conformational B-cell epitope prediction.Briefings in Bioinformatics23, 1 (2022), bbab423

    2022

  12. [12]

    Alice Del Vecchio, Andreea Deac, Pietro Liò, and Petar Veličković. 2021. Neu- ral message passing for joint paratope-epitope prediction.arXiv preprint arXiv:2106.00757(2021)

    arXiv 2021

  13. [13]

    Arthur Deng, Karsten Householder, Fang Wu, Sebastian Thrun, K Christopher Garcia, and Brian Trippe. 2025. Predicting mutational effects on protein binding from folding energy.arXiv preprint arXiv:2507.05502(2025)

    arXiv 2025

  14. [14]

    Bin Deng, Shaolong Zhu, Andrew M Macklin, Jianrong Xu, Cristina Lento, Adnan Sljoka, and Derek J Wilson. 2017. Suppressing allostery in epitope mapping experiments using millisecond hydrogen/deuterium exchange mass spectrometry. InMAbs, Vol. 9. Taylor & Francis, 1327–1336

    2017

  15. [15]

    Dattatraya V Desai and Urmila Kulkarni-Kale. 2014. T-cell epitope prediction methods: an overview.Immunoinformatics(2014), 333–364

    2014

  16. [16]

    Tom Duff, James Burgess, Per Christensen, Christophe Hery, Andrew Kensler, Max Liani, and Ryusuke Villemin. 2017. Building an orthonormal basis, revisited. JCGT6, 1 (2017)

    2017

  17. [17]

    James Dunbar, Konrad Krawczyk, Jinwoo Leem, Terry Baker, Angelika Fuchs, Guy Georges, Jiye Shi, and Charlotte M Deane. 2014. SAbDab: the structural antibody database.Nucleic acids research42, D1 (2014), D1140–D1146

    2014

  18. [18]

    Yasser EL-Manzalawy, Drena Dobbs, and Vasant Honavar. 2008. Predicting linear B-cell epitopes using string kernels.Journal of Molecular Recognition: An Interdisciplinary Journal21, 4 (2008), 243–255

    2008

  19. [19]

    Reyhaneh Esmaielbeiki, Konrad Krawczyk, Bernhard Knapp, Jean-Christophe Nebel, and Charlotte M Deane. 2016. Progress and challenges in predicting protein interfaces.Briefings in bioinformatics17, 1 (2016), 117–131

    2016

  20. [20]

    Richard Evans, Michael O’Neill, Alexander Pritzel, Natasha Antropova, Andrew Senior, Tim Green, Augustin Zidek, Russ Bates, Sam Blackwell, Jason Yim, et al

  21. [21]

    Protein complex prediction with AlphaFold-Multimer.biorxiv(2021), 2021–10

    2021

  22. [22]

    Haoqiang Fan, Hao Su, and Leonidas J Guibas. 2017. A point set generation network for 3d object reconstruction from a single image. InProceedings of the IEEE conference on computer vision and pattern recognition. 605–613

    2017

  23. [23]

    Pablo Gainza, Freyr Sverrisson, Frederico Monti, Emanuele Rodola, D Boscaini, MM Bronstein, and BE Correia. 2020. Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning.Nature Methods17, 2 (2020), 184–192

    2020

  24. [24]

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollar, and Ross Girshick

  25. [25]

    InProceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Masked autoencoders are scalable vision learners. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 16000–16009

  26. [26]

    Magnus Haraldson Hoie, Frederik Steensgaard Gade, Julie Maria Johansen, Char- lotte Wurtzen, Ole Winther, Morten Nielsen, and Paolo Marcatili. 2024. DiscoTope- 3.0: improved B-cell epitope prediction using inverse folding latent representa- tions.Frontiers in Immunology15 (2024), 1322712

    2024

  27. [27]

    Chloe Hsu, Robert Verkuil, Jason Liu, Zeming Lin, Brian Hie, Tom Sercu, Adam Lerer, and Alexander Rives. 2022. Learning inverse folding from millions of predicted structures.bioRxiv(2022)

    2022

  28. [28]

    Eric Jang, Shixiang Gu, and Ben Poole. 2016. Categorical reparameterization with gumbel-softmax.arXiv preprint arXiv:1611.01144(2016)

    Pith/arXiv arXiv 2016

  29. [29]

    Martin Closter Jespersen, Bjoern Peters, Morten Nielsen, and Paolo Marcatili

  30. [30]

    BepiPred-2.0: improving sequence-based B-cell epitope prediction using conformational epitopes.Nucleic acids research45, W1 (2017), W24–W29

    2017

  31. [31]

    Yize Jiang, Xinze Li, Yuanyuan Zhang, Jin Han, Youjun Xu, Ayush Pandit, Zaixi Zhang, Mengdi Wang, Mengyang Wang, Minjie Shen, et al . 2025. PoseX: AI Defeats Physics Approaches on Protein-Ligand Cross Docking.arXiv preprint arXiv:2505.01700(2025)

    Pith/arXiv arXiv 2025

  32. [32]

    Wengong Jin, Regina Barzilay, and Tommi Jaakkola. 2022. Antibody-Antigen Docking and Design via Hierarchical Equivariant Refinement.arXiv preprint arXiv:2207.06616(2022)

    arXiv 2022

  33. [33]

    Robbie P Joosten, Fei Long, Garib N Murshudov, and Anastassis Perrakis. 2014. The PDB_REDO server for macromolecular structure model optimization.IUCrJ 1, 4 (2014), 213–220

    2014

  34. [34]

    Diederik P Kingma and Jimmy Ba. 2014. Adam: A method for stochastic opti- mization.arXiv preprint arXiv:1412.6980(2014)

    Pith/arXiv arXiv 2014

  35. [35]

    Lucien F Krapp, Luciano A Abriata, Fabio Cortes Rodriguez, and Matteo Dal Per- aro. 2023. PeSTo: parameter-free geometric deep learning for accurate prediction of protein binding interfaces.Nature Communications14, 1 (2023), 2175

    2023

  36. [36]

    Jinwoo Leem, Laura S Mitchell, James HR Farmery, Justin Barton, and Jacob D Galson. 2022. Deciphering the language of antibodies using self-supervised learning.Patterns(2022), 100513

    2022

  37. [37]

    Guanlue Li, Xufeng Zhao, Fang Wu, and Sören Laue. 2026. Joint design of protein surface and backbone using a diffusion bridge model.Advances in Neural Information Processing Systems38 (2026), 169682–169708

    2026

  38. [38]

    Pengpai Li and Zhi-Ping Liu. 2023. GeoBind: segmentation of nucleic acid binding interface on protein surface with geometric deep learning.Nucleic Acids Research 51, 10 (2023), e60–e60

    2023

  39. [39]

    Siyuan Li, Luyuan Zhang, Zedong Wang, Di Wu, Lirong Wu, Zicheng Liu, Jun Xia, Cheng Tan, Yang Liu, Baigui Sun, et al. 2023. Masked Modeling for Self-supervised Representation Learning on Vision and Beyond.arXiv preprint arXiv:2401.00897 (2023)

    arXiv 2023

  40. [40]

    Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Dollar. 2017. Focal loss for dense object detection. InProceedings of the IEEE international conference on computer vision. 2980–2988

    2017

  41. [41]

    Zeming Lin et al. 2022. Language models of protein sequences at the scale of evolution enable accurate structure prediction.bioRxiv(2022)

    2022

  42. [42]

    ChuNan Liu, Lilian Denzler, Yihong Chen, Andrew Martin, and Brooks Paige

  43. [43]

    AsEP: Benchmarking deep learning methods for antibody-specific epitope prediction.Advances in Neural Information Processing Systems37 (2024), 11700– 11734

    2024

  44. [44]

    Zhihai Liu, Yan Li, Li Han, Jie Li, Jie Liu, Zhixiong Zhao, Wei Nie, Yuchen Liu, and Renxiao Wang. 2015. PDB-wide collection of binding data: current status of the PDBbind database.Bioinformatics31, 3 (2015), 405–412

    2015

  45. [45]

    Shitong Luo, Yufeng Su, Zuofan Wu, Chenpeng Su, Jian Peng, and Jianzhu Ma

  46. [46]

    Rotamer Density Estimator is an Unsupervised Learner of the Effect of Mutations on Protein-Protein Interaction.bioRxiv(2023), 2023–02

    2023

  47. [47]

    Matt McPartlon and Jinbo Xu. 2023. Deep learning for flexible and site-specific protein docking and design.BioRxiv(2023), 2023–04

    2023

  48. [48]

    Niloy J Mitra and An Nguyen. 2003. Estimating surface normals in noisy point cloud data. InProceedings of the nineteenth annual symposium on Computational geometry. 322–328

    2003

  49. [49]

    Guy M Morton. 1966. A computer oriented geodetic data base and a new technique in file sequencing. (1966)

    1966

  50. [50]

    Stelios K Mylonas, Apostolos Axenopoulos, and Petros Daras. 2021. DeepSurf: a surface-based deep learning approach for the prediction of ligand binding sites on proteins.Bioinformatics37, 12 (2021), 1681–1690

    2021

  51. [51]

    Bjoern Peters, Morten Nielsen, and Alessandro Sette. 2020. T cell epitope predic- tions.Annual Review of Immunology38, 1 (2020), 123–145

    2020

  52. [52]

    Ksenia Polonsky, Tal Pupko, and Natalia T Freund. 2023. Evaluation of the Ability of AlphaFold to Predict the Three-Dimensional Structures of Antibodies and KDD ’26, August 09–13, 2026, Jeju Island, Republic of Korea Fang Wu, Weihao Xuan, Jure Leskovec, Yejin Choi, and Li Li Epitopes.The Journal of Immunology211, 10 (2023), 1578–1588

    2023

  53. [53]

    Lenka Potocnakova, Mangesh Bhide, and Lucia Borszekova Pulzova. 2016. An introduction to B-cell epitope mapping and in silico epitope prediction.Journal of immunology research2016, 1 (2016), 6760830

    2016

  54. [54]

    Charles Ruizhongtai Qi, Li Yi, Hao Su, and Leonidas J Guibas. 2017. Pointnet++: Deep hierarchical feature learning on point sets in a metric space.Advances in neural information processing systems30 (2017)

    2017

  55. [55]

    Huan Qi, Mingliang Ma, Chuansheng Hu, Zhao-wei Xu, Fan-lin Wu, Nan Wang, Dan-yun Lai, Yang Li, Hainan Zhang, He-wei Jiang, et al. 2021. Antibody binding epitope mapping (AbMap) of hundred antibodies in a single run.Molecular & Cellular Proteomics20 (2021)

    2021

  56. [56]

    Tianyi Qiu, Lu Zhang, Zikun Chen, Yuan Wang, Tiantian Mao, Caicui Wang, Yewei Cun, Genhui Zheng, Deyu Yan, Mengdi Zhou, et al. 2023. SEPPA-mAb: spatial epitope prediction of protein antigens for mAbs.Nucleic Acids Research (2023), gkad427

    2023

  57. [57]

    Aditya Ramesh, Mikhail Pavlov, Gabriel Goh, Scott Gray, Chelsea Voss, Alec Radford, Mark Chen, and Ilya Sutskever. 2021. Zero-shot text-to-image generation. InInternational Conference on Machine Learning. PMLR, 8821–8831

    2021

  58. [58]

    Saleh Riahi, Jae Hyeon Lee, Taylor Sorenson, Shuai Wei, Sven Jager, Reza Olfati- Saber, Yanfeng Zhou, Anna Park, Maria Wendt, Herve Minoux, et al. 2023. Surface ID: a geometry-aware system for protein molecular surface comparison.Bioin- formatics39, 4 (2023), btad196

    2023

  59. [59]

    Alexander Rives et al . 2021. Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences.Proceedings of the National Academy of Sciences118, 15 (2021), e2016239118

    2021

  60. [60]

    Jose L Sanchez-Trincado, Marta Gomez-Perosanz, and Pedro A Reche. 2017. Fundamentals and methods for T-and B-cell epitope prediction.Journal of im- munology research2017, 1 (2017), 2680160

    2017

  61. [61]

    Kristof T Schutt, Huziel E Sauceda, P-J Kindermans, Alexandre Tkatchenko, and K-R Muller. 2018. Schnet–a deep learning architecture for molecules and materials.The Journal of Chemical Physics148, 24 (2018)

    2018

  62. [62]

    Tatiana I Shashkova et al. 2022. SEMA: Antigen B-cell conformational epitope prediction using deep transfer learning.Frontiers in immunology(2022), 5272

    2022

  63. [63]

    Ruth E Soria-Guerra, Ricardo Nieto-Gomez, Dania O Govea-Alonso, and Sergio Rosales-Mendoza. 2015. An overview of bioinformatics tools for epitope predic- tion: implications on vaccine development.Journal of biomedical informatics53 (2015), 405–414

    2015

  64. [64]

    Vitalii Stebliankin, Azam Shirali, Prabin Baral, Jimeng Shi, Prem Chapagain, Kalai Mathee, and Giri Narasimhan. 2023. Evaluating protein binding interfaces with transformer networks.Nature Machine Intelligence5, 9 (2023), 1042–1053

    2023

  65. [65]

    Martin Steinegger and Johannes Soding. 2017. MMseqs2 enables sensitive protein sequence searching for the analysis of massive data sets.Nature biotechnology 35, 11 (2017), 1026–1028

    2017

  66. [66]

    Freyr Sverrisson, Jean Feydy, Bruno E Correia, and Michael M Bronstein. 2021. Fast end-to-end learning on protein surfaces. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 15272–15281

    2021

  67. [67]

    Xiaoyu Tian, Haoxi Ran, Yue Wang, and Hang Zhao. 2023. GeoMAE: Masked Geometric Target Prediction for Self-supervised Point Cloud Pre-Training. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 13570–13580

    2023

  68. [68]

    Jerome Tubiana, Dina Schneidman-Duhovny, and Haim J Wolfson. 2022. Scan- Net: an interpretable geometric deep learning model for structure-based protein binding site prediction.Nature Methods19, 6 (2022), 730–739

    2022

  69. [69]

    Aaron Van Den Oord, Oriol Vinyals, et al. 2017. Neural discrete representation learning.Advances in neural information processing systems30 (2017)

    2017

  70. [70]

    Mihaly Varadi, Stephen Anyango, Mandar Deshpande, Sreenath Nair, Cindy Natassia, Galabina Yordanova, David Yuan, Oana Stroe, Gemma Wood, Agata Laydon, et al. 2022. AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic acids research50, D1 (2022), D439–D444

    2022

  71. [71]

    Mihaly Varadi, Damian Bertoni, Paulyna Magana, Urmila Paramval, Ivanna Pidruchna, Malarvizhi Radhakrishnan, Maxim Tsenkov, Sreenath Nair, Milot Mirdita, Jingi Yeo, et al. 2024. AlphaFold Protein Structure Database in 2024: providing structure coverage for over 214 million protein sequences.Nucleic Acids Research52, D1 (2024), D368–D375

    2024

  72. [72]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Lukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need.Advances in neural information processing systems30 (2017)

    2017

  73. [73]

    Thom Vreven et al. 2015. Updates to the integrated protein–protein interaction benchmarks: docking benchmark version 5 and affinity benchmark version 2. Journal of molecular biology427, 19 (2015), 3031–3041

    2015

  74. [74]

    Xiao Wang, Genki Terashi, Charles W Christoffer, Mengmeng Zhu, and Daisuke Kihara. 2020. Protein docking model evaluation by 3D deep convolutional neural networks.Bioinformatics36, 7 (2020), 2113–2118

    2020

  75. [75]

    Fang Wu. 2024. A semi-supervised molecular learning framework for activity cliff estimation. InProceedings of the Thirty-Third International Joint Conference on Artificial Intelligence. 6080–6088

    2024

  76. [76]

    Fang Wu. 2025. DiffAntiSeq: A Controllable Diffusion Model for Efficient Anti- body Library Design. InLLM for Scientific Discovery: Reasoning, Assistance, and Collaboration

    2025

  77. [77]

    Fang Wu. 2026. A semi-supervised molecular learning framework for activity cliff estimation.arXiv preprint arXiv:2601.04507(2026)

    arXiv 2026

  78. [78]

    Fang Wu, Shuting Jin, Yinghui Jiang, Xurui Jin, Bowen Tang, Zhangming Niu, Xiangrong Liu, Qiang Zhang, Xiangxiang Zeng, and Stan Z Li. 2022. Pre-Training of Equivariant Graph Matching Networks with Conformation Flexibility for Drug Binding.Advanced Science9, 33 (2022), 2203796

    2022

  79. [79]

    Fang Wu, Shuting Jin, Xiangru Tang, Junlin Xu, Mark Gerstein, Li Erran Li, and James Zou. 2026. D-flow: Multi-modality flow matching for d-peptide design. IEEE Journal of Biomedical and Health Informatics(2026)

    2026

  80. [80]

    Fang Wu, Siyuan Li, Lirong Wu, Stan Z Li, Dragomir Radev, and Qiang Zhang

Showing first 80 references.