The intricate three-dimensional structures of proteins dictate their diverse biological functions. Accurately representing these structures, particularly the fine-grained orientational relationships between amino acids, is crucial for understanding protein mechanisms and practical computational protein analysis. Current methodologies often fall short in capturing these essential geometric details. We introduce Orientation-Aware Graph Neural Networks (OA-GNNs), a novel deep learning framework that explicitly models local and global geometric characteristics, including inner-residue torsion angles and inter-residue orientations. OA-GNNs achieve this by uniquely extending neural network weights from scalars to 3D directed weights and by implementing an equivariant message passing paradigm that ensures SO(3)-equivariance. This approach allows for a richer, more geometrically meaningful processing of protein structural data. Comprehensive experiments demonstrate that OA-GNNs significantly outperform existing methods in sensing orientational features and achieve state-of-the-art performance across diverse computational biology tasks, including residue identification, computational protein design, model quality assessment, and protein function classification. Our findings highlight the power of orientation-aware learning and establish OA-GNNs as a versatile and robust tool for advancing our understanding of protein structure–function relationships and for developing new therapeutic and biotechnological solutions. The code is available at https://github.com/Ced3-han/OAGNN/tree/main.
AlfordRF, Leaver-FayA, JeliazkovJR, et al.The Rosetta all-atom energy function for macromolecular modeling and design. J Chem Theory Comput, 2017; 13(6):3031–3048.
2.
AlleyEC, KhimulyaG, BiswasS, et al.Unified rational protein engineering with sequence-based deep representation learning. Nat Methods, 2019; 16(12):1315–1322.
3.
AmidiA, AmidiS, VlachakisD, et al.Enzynet: Enzyme classification using 3d convolutional neural networks on spatial representation. PeerJ, 2018; 6:e4750.
BaekM, DiMaioF, AnishchenkoI, et al.Accurate prediction of protein structures and interactions using a three-track neural network. Science, 2021; 373(6557):871–876.
6.
BaldassarreF, Menéndez HurtadoD, ElofssonA, et al.Graphqa: Protein model quality assessment using graph convolutional networks. Bioinformatics, 2021; 37(3):360–366.
7.
BatznerS, SmidtTE, SunL, et al.Se (3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials. arXiv Preprint arXiv:2101.03164 2021.
8.
BeplerT, BergerB. (2018). Learning protein sequence embeddings using information from structure. In International Conference on Learning Representations.
9.
BrandstetterJ, HesselinkR, van der PolE, et al.Geometric and physical quantities improve e (3) equivariant message passing. arXiv Preprint arXiv, 2021.
10.
CaoY, DasP, ChenthamarakshanV, et al. (2021). Fold2seq: A joint sequence (1d)-fold (3d) embedding-based generative model for protein design. In International Conference on Machine Learning, pages 1261–1271. PMLR.
11.
ChengJ, ChoeM-H, ElofssonA, et al.Estimation of model accuracy in casp13. Proteins: Structure, Function, and Bioinformatics, 2019; 87(12):1361–1377.
12.
CohenT, WellingM. (2016). Group equivariant convolutional networks. In International Conference on Machine Learning, pages 2990–2999. PMLR.
13.
DengC, LitanyO, DuanY, et al. (2021). Vector neurons: A general framework for so (3)-equivariant networks. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 12200–12209.
14.
DerevyankoG, GrudininS, BengioY, et al.Deep convolutional networks for quality assessment of protein folds. Bioinformatics, 2018; 34(23):4046–4053.
DuhovnyD, NussinovR, WolfsonHJ. (2002). Efficient unbound docking of rigid molecules. In International Workshop on Algorithms in Bioinformatics, pages 185–200. Springer.
17.
EismannS, TownshendRJ, ThomasN, et al.Hierarchical, rotation-equivariant neural networks to select structural models of protein complexes. Proteins: Structure, Function, and Bioinformatics, 2021; 89(5):493–501.
18.
ElnaggarA, HeinzingerM, DallagoC, et al.Prottrans: Towards cracking the language of life’s code through self-supervised deep learning and high performance computing. arXiv Preprint arXiv, 2020.
19.
FoutAM. (2017). Protein interface prediction using graph convolutional networks. PhD thesis, Colorado State University.
20.
FuchsF, WorrallD, FischerV, et al.Se (3)-transformers: 3d roto-translation equivariant attention networks. Adv Neural Inf Process Syst, 2020; 33:1970–1981.
21.
GainzaP, SverrissonF, MontiF, et al.Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning. Nat Methods, 2020; 17(2):184–192.
22.
GaoH, JiS. (2019). Graph u-nets. In International Conference on Machine Learning, pages 2083–2092. PMLR.
23.
GasteigerJ, GroßJ, GünnemannS. (2019a). Directional message passing for molecular graphs. In International Conference on Learning Representations.
24.
GasteigerJ, GroßJ, GünnemannS. (2019b). Directional message passing for molecular graphs. In International Conference on Learning Representations.
25.
GligorijevićV, RenfrewPD, KosciolekT, et al.Structure-based protein function prediction using graph convolutional networks. Nat Commun, 2021; 12(1):3168.
26.
Hermosilla CasajusP, SchäferM, LangM, et al. (2021). Intrinsic-extrinsic convolution and pooling for learning on 3d protein structures. In International Conference on Learning Representations, ICLR: Vienna,Austria, pages 1–16. OpenReview. net.
27.
HermosillaP, SchäferM, LangM, et al. (2020). Intrinsic-extrinsic convolution and pooling for learning on 3d protein structures. In International Conference on Learning Representations.
28.
HouJ, AdhikariB, ChengJ. Deepsf: Deep convolutional neural network for mapping protein sequences to folds. Bioinformatics, 2018; 34(8):1295–1303.
29.
IgashovI, OlechnovičK, KadukovaM, et al.VoroCNN: Deep convolutional neural network built on 3d Voronoi tessellation of protein structures. Bioinformatics, 2021; 37(16):2332–2339.
30.
IngrahamJ, GargV, BarzilayR, et al.Generative models for graph-based protein design. Adv Neural Inf Process Syst, 2019:32.
31.
JelinekF, MercerRL, BahlLR, et al.Perplexity—a measure of the difficulty of speech recognition tasks. J Acoust Soc Am, 1977; 62(S1):S63.
32.
JiS, XuW, YangM, et al.3d convolutional neural networks for human action recognition. IEEE Trans Pattern Anal Mach Intell, 2013; 35(1):221–231.
33.
JingB, EismannS, SoniPN, et al.Equivariant graph neural networks for 3d macromolecular structure. ICML 2021 CompBio Workshop, 2021.
34.
JingB, EismannS, SurianaP, et al. (2020). Learning from protein structure with geometric vector perceptrons. In International Conference on Learning Representations.
35.
JumperJ, EvansR, PritzelA, et al.Highly accurate protein structure prediction with alphafold. Nature, 2021; 596(7873):583–589.
36.
KällbergM, WangH, WangS, et al.Template-based protein structure modeling using the RaptorX web server. Nat Protoc, 2012; 7(8):1511–1522.
37.
KarasikovM, PagèsG, GrudininS. Smooth orientation-dependent scoring function for coarse-grained protein quality assessment. Bioinformatics, 2019; 35(16):2801–2808.
KöhlerJ, KleinL, NoéF. (2020). Equivariant flows: Exact likelihood generative learning for symmetric densities. In International Conference on Machine Learning, pages 5361–5370. PMLR.
40.
KwonS, WonJ, KryshtafovychA, et al. (2021). Assessment of protein model structure accuracy estimation in casp14: Old and new challenges. Proteins: Structure, Function, and Bioinformatics.
41.
LiJ, ChenT, LuoS, et al.Hotspot-driven peptide design via multi-fragment autoregressive extension. arXiv Preprint arXiv, 2024a.
42.
LiJ, ChengC, WuZ, et al.Full-atom peptide design based on multi-modal flow matching. arXiv Preprint arXiv, 2024b.
43.
LiS, ZhouJ, XuT, et al. (2021). Structure-aware interactive graph neural networks for the prediction of protein-ligand binding affinity. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, pages 975–985.
44.
LiZ, YangY, FaraggiE, et al.Direct prediction of profiles of sequences compatible with a protein structure by neural networks with fragment-based local and energy-based nonlocal profiles. Proteins: Structure, Function, and Bioinformatics, 2014; 82(10):2565–2573.
45.
LiaoY-L, SmidtT. Equiformer: Equivariant graph attention transformer for 3d atomistic graphs. arXiv Preprint arXiv, 2022.
46.
LiaoY-L, WoodB, DasA, et al.Equiformerv2: Improved equivariant transformer for scaling to higher-degree representations. arXiv Preprint arXiv, 2023.
47.
LiuY, WangL, LiuM, et al.Spherical message passing for 3d graph networks. 2021.
48.
LundströmJ, RychlewskiL, BujnickiJ, et al.Pcons: A neural-network–based consensus predictor that improves fold recognition. Protein Sci, 2001; 10(11):2354–2362.
49.
LuoS, LiJ, GuanJ, et al. (2022). Equivariant point cloud analysis via learning orientations for message passing. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18932–18941.
50.
MoultJ, FidelisK, KryshtafovychA, et al.Critical assessment of methods of protein structure prediction (CASP)—Round XV. Proteins: Structure, Function, and Bioinformatics, 2014; 82(Suppl 2)(0 2):1–6.
51.
MurzinAG, BrennerSE, HubbardT, et al.Scop: A structural classification of proteins database for the investigation of sequences and structures. J Mol Biol, 1995; 247(4):536–540.
52.
NairV, HintonGE. Rectified linear units improve restricted Boltzmann machines. In Icml, 2010.
53.
NelsonDL, LehningerAL, CoxMM. (2008). Lehninger principles of biochemistry. Macmillan.
54.
O’ConnellJ, LiZ, HansonJ, et al.Spin2: Predicting sequence profiles from protein structures using deep neural networks. Proteins: Structure, Function, and Bioinformatics, 2018; 86(6):629–633.
55.
OlechnovičK, VenclovasČ. Voromqa: Assessment of protein structure quality using interatomic contact areas. Proteins: Structure, Function, and Bioinformatics, 2017; 85(6):1131–1145.
56.
OrengoCA, MichieAD, JonesS, et al.Cath—A hierarchic classification of protein domain structures. Structure, 1997; 5(8):1093–1108.
57.
PagèsG, CharmettantB, GrudininS. Protein model quality assessment using 3D oriented convolutional neural networks. Bioinformatics, 2019; 35(18):3313–3319.
58.
PassaroS, ZitnickCL. (2023). Reducing SO (3) convolutions to SO (2) for efficient equivariant GNNS. In International Conference on Machine Learning, pages 27420–27438. PMLR.
59.
QiY, ZhangJZ. DenseCPD: Improving the accuracy of neural-network-based computational protein sequence design with DenseNet. J Chem Inf Model, 2020; 60(3):1245–1252.
60.
RaoR, BhattacharyaN, ThomasN, et al.Evaluating protein transfer learning with tape. Adv Neural Inf Process Syst, 2019; 32:9689–9701.
61.
RohlCA, StraussCE, MisuraKM, et al.Protein structure prediction using Rosetta. Methods Enzymol, 2004; 383:66–93.
62.
SatorrasVG, HoogeboomE, WellingM. (2021). E (n) equivariant graph neural networks. In International Conference on Machine Learning, pages 9323–9332. PMLR.
63.
SchaapMG, LeijFJ, Van GenuchtenMT. Rosetta: A computer program for estimating soil hydraulic parameters with hierarchical pedotransfer functions. J Hydrol (Amst), 2001; 251(3–4):163–176.
64.
SchüttK, KindermansP-J, Sauceda FelixHE, et al.SchNet: A continuous-filter convolutional neural network for modeling quantum interactions. Adv Neural Inf Process Syst, 2017; 30.
65.
SchüttK, UnkeO, GasteggerM. (2021). Equivariant message passing for the prediction of tensorial properties and molecular spectra. In International Conference on Machine Learning, pages 9377–9388. PMLR.
66.
Shulman-PelegA, NussinovR, WolfsonHJ. Recognition of functional sites in protein structures. J Mol Biol, 2004; 339(3):607–633.
67.
StrodthoffN, WagnerP, WenzelM, et al.UDSMProt: Universal deep sequence models for protein classification. Bioinformatics, 2020; 36(8):2401–2409.
68.
StrokachA, BecerraD, Corbi-VergeC, et al.Fast and flexible protein design using deep graph neural networks. Cell Syst, 2020; 11(4):402–411.e4.
69.
SverrissonF, FeydyJ, CorreiaBE, et al. (2021). Fast end-to-end learning on protein surfaces. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 15272–15281.
70.
ThomasN, SmidtT, KearnesS, et al.Tensor field networks: Rotation-and translation-equivariant neural networks for 3d point clouds. arXiv Preprint arXiv, 2018.
71.
TorngW, AltmanRB. 3D deep convolutional neural networks for amino acid environment similarity analysis. BMC Bioinformatics, 2017; 18(1):302–323.
72.
TownshendRJL, VögeleM, SurianaPA, et al. (2021). ATOM3D: Tasks on molecules in three dimensions. In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track.
73.
TownshendR, BediR, SurianaP, et al.End-to-end learning on 3d protein structure for interface prediction. Adv Neural Inf Process Syst, 2019; 32:15642–15651.
74.
UzielaK, Menendez HurtadoD, ShuN, et al.ProQ3D: Improved model quality assessments using deep learning. Bioinformatics, 2017; 33(10):1578–1580.
75.
VaswaniA, ShazeerN, ParmarN, et al.Attention is all you need. In Advances in Neural Information Processing Systems, 2017, pages 5998–6008.
76.
VecchioA, DeacA, LiòP, et al. (2021). Neural message passing for joint paratope-epitope prediction.
77.
VeličkovićP, CucurullG, CasanovaA, et al. (2018). Graph attention networks. In International Conference on Learning Representations.
78.
VoetD, VoetJG. (2010). Biochemistry. John Wiley & Sons.
79.
WangJ, CaoH, ZhangJZ, et al.Computational protein design with deep learning neural networks. Sci Rep, 2018; 8(1):6349.
80.
WangX, FlanneryST, KiharaD. Protein docking model evaluation by graph neural networks. Front Mol Biosci, 2021; 8:647915.
81.
WatsonJL, JuergensD, BennettNR, et al.De novo design of protein structure and function with RFdiffusion. Nature, 2023; 620(7976):1089–1100.
82.
WebbEC, et al. (1992). Enzyme nomenclature 1992. Recommendations of the Nomenclature Committee of the International Union of Biochemistry and Molecular Biology on the Nomenclature and Classification of Enzymes. Academic Press.
83.
WeilerM, GeigerM, WellingM, et al.3D steerable CNNS: Learning rotationally equivariant features in volumetric data. Adv Neural Inf Process Syst, 2018; 31.
84.
XuJ, BergerB. Fast and accurate algorithms for protein side-chain packing. J ACM, 2006; 53(4):533–557.
85.
XuK, HuW, LeskovecJ, et al. (2018). How powerful are graph neural networks? In International Conference on Learning Representations.
86.
YimJ, TrippeBL, De BortoliV, et al.SE(3) diffusion model with application to protein backbone generation. arXiv Preprint arXiv:2302.02277 2023.
87.
YingZ, YouJ, MorrisC, et al.Hierarchical graph representation learning with differentiable pooling. Adv Neural Inf Process Syst, 2018; 31.
88.
ZemlaA. LGA: A method for finding 3D similarities in protein structures. Nucleic Acids Res, 2003; 31(13):3370–3374.
89.
ZhangJ, ZhangY. A novel side-chain orientation dependent potential derived from random-walk reference state for protein fold selection and structure prediction. PLoS One, 2010; 5(10):e15386.
90.
ZhangY, ChenY, WangC, et al.ProDCoNN: Protein design using a convolutional neural network. Proteins: Structure, Function, and Bioinformatics, 2020; 88(7):819–829.
91.
ZhemchuzhnikovD, IgashovI, GrudininS. (2022). 6DCNN with roto-translational convolution filters for volumetric data processing. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pages 4707–4715.
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
