Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Article
  • Published:

Generic protein–ligand interaction scoring by integrating physical prior knowledge and data augmentation modelling

Nature Machine Intelligence volume 6, pages 688–700 (2024)Cite this article

Subjects

A preprint version of the article is available at bioRxiv.

Abstract

Developing robust methods for evaluating protein–ligand interactions has been a long-standing problem. Data-driven methods may memorize ligand and protein training data rather than learning protein–ligand interactions. Here we show a scoring approach called EquiScore, which utilizes a heterogeneous graph neural network to integrate physical prior knowledge and characterize protein–ligand interactions in equivariant geometric space. EquiScore is trained based on a new dataset constructed with multiple data augmentation strategies and a stringent redundancy-removal scheme. On two large external test sets, EquiScore consistently achieved top-ranking performance compared to 21 other methods. When EquiScore is used alongside different docking methods, it can effectively enhance the screening ability of these docking methods. EquiScore also showed good performance on the activity-ranking task of a series of structural analogues, indicating its potential to guide lead compound optimization. Finally, we investigated different levels of interpretability of EquiScore, which may provide more insights into structure-based drug design.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on SpringerLink
  • Instant access to the full article PDF.

USD 39.95

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: The pipeline of building the PDBscreen dataset.
Fig. 2: The overall architecture of EquiScore.
Fig. 3: Evaluation of 22 scoring methods on DEKOIS2.0.
Fig. 4: Evaluation of 22 scoring methods on DUD-E in terms of AUROC, BEDROC and EF.
Fig. 5: Performance comparison of EquiScore for rescoring the docking poses generated by different docking methods on DEKOIS2.0.
Fig. 6: Interpretation of EquiScore by visualizing attention distribution.

Similar content being viewed by others

Data availability

The PDBscreen dataset supporting this study’s findings is available via Zenodo at https://doi.org/10.5281/zenodo.8049380 (ref. 67). The test dataset supporting this study’s findings is available via Zenodo at https://doi.org/10.5281/zenodo.8047224 (ref. 68). Original data and supplementary information supporting this study’s findings are available via Zenodo at https://doi.org/10.5281/zenodo.10812637 (ref. 69). Source data are provided with this paper.

Code availability

The code used to generate the results shown in this study is available under an MIT License via GitHub at https://github.com/CAODH/EquiScore (ref. 70) and via Zenodo at https://doi.org/10.5281/zenodo.10812534 (ref. 71).

References

  1. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).

    Google Scholar 

  2. Baek, M. et al. Accurate prediction of protein structures and interactions using a three-track neural network. Science 373, 871–876 (2021).

    Google Scholar 

  3. Muller, S. et al. Target 2035—update on the quest for a probe for every protein. RSC Med. Chem. 13, 13–21 (2022).

    Google Scholar 

  4. Kaplan, A. L. et al. Bespoke library docking for 5-HT(2A) receptor agonists with antidepressant activity. Nature 610, 582–591 (2022).

    Google Scholar 

  5. Lyu, J. et al. Ultra-large library docking for discovering new chemotypes. Nature 566, 224–229 (2019).

    Google Scholar 

  6. Shen, C. et al. Beware of the generic machine learning-based scoring functions in structure-based virtual screening. Brief. Bioinform. 22, bbaa070 (2021).

    Google Scholar 

  7. Guedes, I. A., Pereira, F. S. S. & Dardenne, L. E. Empirical scoring functions for structure-based virtual screening: applications, critical aspects, and challenges. Front. Pharmacol. 9, 411637 (2018).

    Google Scholar 

  8. Shen, C. et al. Accuracy or novelty: what can we gain from target-specific machine-learning-based scoring functions in virtual screening? Brief. Bioinform. 22, bbaa410 (2021).

    Google Scholar 

  9. Zhu, H., Yang, J. & Huang, N. Assessment of the generalization abilities of machine-learning scoring functions for structure-based virtual screening. J. Chem. Inf. Model. 62, 5485–5502 (2022).

    Google Scholar 

  10. Francoeur, P. G. et al. Three-dimensional convolutional neural networks and a cross-docked data set for structure-based drug design. J. Chem. Inf. Model. 60, 4200–4215 (2020).

    Google Scholar 

  11. Ragoza, M., Hochuli, J., Idrobo, E., Sunseri, J. & Koes, D. R. Protein-ligand scoring with convolutional neural networks. J. Chem. Inf. Model. 57, 942–957 (2017).

    Google Scholar 

  12. Li, S. et al. Structure-aware interactive graph neural networks for the prediction of protein-ligand binding affinity. In Proc. 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining (eds Feida, Z. et al.) 975–985 (ACM, 2021); https://doi.org/10.1145/3447548.3467311

  13. Lim, J. et al. Predicting drug-target interaction using a novel graph neural network with 3D structure-embedded graph representation. J. Chem. Inf. Model. 59, 3981–3988 (2019).

    Google Scholar 

  14. Moon, S., Zhung, W., Yang, S., Lim, J. & Kim, W. Y. PIGNet: a physics-informed deep learning model toward generalized drug-target interaction predictions. Chem. Sci. 13, 3661–3673 (2022).

    Google Scholar 

  15. Shen, C. et al. Boosting protein-ligand binding pose prediction and virtual screening based on residue-atom distance likelihood potential and graph transformer. J. Med. Chem. 65, 10691–10706 (2022).

    Google Scholar 

  16. Jiang, D. et al. InteractionGraphNet: a novel and efficient deep graph representation learning framework for accurate protein-ligand interaction predictions. J. Med. Chem. 64, 18209–18232 (2021).

    Google Scholar 

  17. Méndez-Lucio, O., Ahmad, M., del Rio-Chanona, E. A. & Wegner, J. K. A geometric deep learning approach to predict binding conformations of bioactive molecules. Nat. Mach. Intell. 3, 1033–1039 (2021).

    Google Scholar 

  18. Li, Y. & Yang, J. Structural and sequence similarity makes a significant impact on machine-learning-based scoring functions for protein–ligand interactions. J. Chem. Inf. Model. 57, 1007–1012 (2017).

    Google Scholar 

  19. Chen, L. et al. Hidden bias in the DUD-E dataset leads to misleading performance of deep learning in structure-based virtual screening. PLoS ONE 14, e0220113 (2019).

    Google Scholar 

  20. Chatterjee, A. et al. Improving the generalizability of protein-ligand binding predictions with AI-Bind. Nat. Commun. 14, 1989 (2023).

    Google Scholar 

  21. Geirhos, R. et al. Shortcut learning in deep neural networks. Nat. Mach. Intell. 2, 665–673 (2020).

    Google Scholar 

  22. Sastry, G. M., Dixon, S. L. & Sherman, W. Rapid shape-based ligand alignment and virtual screening method based on atom/feature-pair similarities and volume overlap scoring. J. Chem. Inf. Model. 51, 2455–2466 (2011).

    Google Scholar 

  23. Volkov, M. et al. On the frustration to predict binding affinities from protein–ligand structures with deep neural networks. J. Med. Chem. 65, 7946–7958 (2022).

    Google Scholar 

  24. Li, S. et al. MONN: a multi-objective neural network for predicting compound-protein interactions and affinities. Cell Syst. 10, 308–322 (2020).

    Google Scholar 

  25. Cain, S., Risheh, A. & Forouzesh, N. Calculation of protein-ligand binding free energy using a physics-guided neural network. In Proc. IEEE International Conference on Bioinformatics and Biomedicine (BIBM) (eds Chen, Y. et al.) 2487–2493 (IEEE, 2021); https://doi.org/10.1109/bibm52615.2021.9669867

  26. Stärk, H., Ganea, O., Pattanaik, L., Barzilay, R. & Jaakkola, T. Equibind: geometric deep learning for drug binding structure prediction. In Proc. 39th International Conference on Machine Learning (eds Chaudhuri, K. et al.) 20503–20521 (PMLR, 2022); https://doi.org/10.48550/arXiv.2202.05146

  27. Batzner, S. et al. E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials. Nat. Commun. 13, 2453 (2022).

    Google Scholar 

  28. Atz, K., Grisoni, F. & Schneider, G. Geometric deep learning on molecular representations. Nat. Mach. Intell. 3, 1023–1032 (2021).

    Google Scholar 

  29. Thurlemann, M., Boselt, L. & Riniker, S. Learning atomic multipoles: prediction of the electrostatic potential with equivariant graph neural networks. J. Chem. Theory Comput. 18, 1701–1710 (2022).

    Google Scholar 

  30. Batool, M., Ahmad, B. & Choi, S. A structure-based drug discovery paradigm. Int. J. Mol. Sci. 20, 2783 (2019).

    Google Scholar 

  31. Imrie, F., Bradley, A. R. & Deane, C. M. Generating property-matched decoy molecules using deep learning. Bioinformatics 37, 2134–2141 (2021).

    Google Scholar 

  32. Mysinger, M. M., Carchia, M., Irwin, J. J. & Shoichet, B. K. Directory of useful decoys, enhanced (DUD-E): better ligands and decoys for better benchmarking. J. Med. Chem. 55, 6582–6594 (2012).

    Google Scholar 

  33. Bauer, M. R., Ibrahim, T. M., Vogel, S. M. & Boeckler, F. M. Evaluation and optimization of virtual screening workflows with DEKOIS 2.0—a public library of challenging docking benchmark sets. J. Chem. Inf. Model. 53, 1447–1462 (2013).

    Google Scholar 

  34. Wang, L. et al. Accurate and reliable prediction of relative ligand binding potency in prospective drug discovery by way of a modern free-energy calculation protocol and force field. J. Am. Chem. Soc. 137, 2695–2703 (2015).

    Google Scholar 

  35. Sieg, J., Flachsenberg, F. & Rarey, M. In need of bias control: evaluating chemical data for machine learning in structure-based virtual screening. J. Chem. Inf. Model. 59, 947–961 (2019).

    Google Scholar 

  36. Berman, H. M. et al. The protein data bank. Nucleic Acids Res. 28, 235–242 (2000).

    Google Scholar 

  37. Adeshina, Y. O., Deeds, E. J. & Karanicolas, J. Machine learning classification can reduce false positives in structure-based virtual screening. Proc. Natl Acad. Sci. USA 117, 18477–18488 (2020).

    Google Scholar 

  38. Bouysset, C. & Fiorucci, S. ProLIF: a library to encode molecular interactions as fingerprints. J. Cheminform. 13, 72 (2021).

    Google Scholar 

  39. Satorras, V. G., Hoogeboom, E. & Welling, M. E(n) equivariant graph neural networks. In Proc. 38th International Conference on Machine Learning (eds Meila, M. & Zhang, T.) 9323–9332 (PMLR, 2021); https://doi.org/10.48550/arXiv.2102.09844

  40. Yun, S., Jeong, M., Kim, R., Kang, J. & Kim, H. J. Graph transformer networks. In Advances in Neural Information Processing Systems 32 (eds Wallach, H. et al.) 11983–11993 (NeurIPS, 2019); https://doi.org/10.48550/arXiv.1911.06455

  41. Friesner, R. A. et al. Glide: a new approach for rapid, accurate docking and scoring. 1. Method and assessment of docking accuracy. J. Med. Chem. 47, 1739–1749 (2004).

    Google Scholar 

  42. Mastropietro, A., Pasculli, G. & Bajorath, J. Learning characteristics of graph neural networks predicting protein–ligand affinities. Nat. Mach. Intell. 5, 1427–1436 (2023).

    Google Scholar 

  43. Yu, Y., Lu, S., Gao, Z., Zheng, H. & Ke, G. Do deep learning models really outperform traditional approaches in molecular docking? Preprint at https://doi.org/10.48550/arXiv.2302.07134 (2023).

  44. Sastry, G. M., Adzhigirey, M., Day, T., Annabhimoju, R. & Sherman, W. Protein and ligand preparation: parameters, protocols, and influence on virtual screening enrichments. J. Comput. Aided Mol. Des. 27, 221–234 (2013).

    Google Scholar 

  45. Harder, E. et al. OPLS3: a force field providing broad coverage of drug-like small molecules and proteins. J. Chem. Theory Comput. 12, 281–296 (2016).

    Google Scholar 

  46. Tuccinardi, T., Poli, G., Romboli, V., Giordano, A. & Martinelli, A. Extensive consensus docking evaluation for ligand pose prediction and virtual screening studies. J. Chem. Inf. Model. 54, 2980–2986 (2014).

    Google Scholar 

  47. Westbrook, J. D. et al. The chemical component dictionary: complete descriptions of constituent molecules in experimentally determined 3D macromolecules in the Protein Data Bank. Bioinformatics 31, 1274–1278 (2015).

    Google Scholar 

  48. UniProt, C. UniProt: the universal protein knowledgebase in 2021. Nucleic Acids Res. 49, D480–D489 (2021).

    Google Scholar 

  49. Wierbowski, S. D., Wingert, B. M., Zheng, J. & Camacho, C. J. Cross‐docking benchmark for automated pose and ranking prediction of ligand binding. Protein Sci. 29, 298–305 (2020).

    Google Scholar 

  50. Shen, C. et al. The impact of cross-docked poses on performance of machine learning classifier for protein–ligand binding pose prediction. J. Cheminform. 13, 1–18 (2021).

    Google Scholar 

  51. Zhang, X. et al. TocoDecoy: a new approach to design unbiased datasets for training and benchmarking machine-learning scoring functions. J. Med. Chem. 65, 7918–7932 (2022).

    Google Scholar 

  52. Su, M., Feng, G., Liu, Z., Li, Y. & Wang, R. Tapping on the black box: how is the scoring power of a machine-learning scoring function dependent on the training set? J. Chem. Inf. Model. 60, 1122–1136 (2020).

    Google Scholar 

  53. Scantlebury, J. et al. A small step toward generalizability: training a machine learning scoring function for structure-based virtual screening. J. Chem. Inf. Model. 63, 2960–2974 (2023).

    Google Scholar 

  54. Ying, C. et al. Do transformers really perform bad for graph representation? In Advances in Neural Information Processing Systems 34 (eds Ranzato, M. et al.) 28877–28888 (NeurIPS, 2021); https://doi.org/10.48550/arXiv.2106.05234

  55. Gilmer, J., Schoenholz, S. S., Riley, P. F., Vinyals, O. & Dahl, G. E. Neural message passing for quantum chemistry. In Proc. 34th International Conference on Machine Learning (eds Precup, D. & Teh, Y. W.) 1263–1272 (PMLR, 2017); https://doi.org/10.5555/3305381.3305512

  56. Jiao, Q. et al. Edge-gated graph neural network for predicting protein-ligand binding affinities. In Proc. IEEE International Conference on Bioinformatics and Biomedicine (BIBM) (eds Huang, Y. et al.) 334–339 (IEEE, 2021); https://doi.org/10.1109/bibm52615.2021.9669846

  57. Shang, C. et al. Edge attention-based multi-relational graph convolutional networks. Preprint at https://doi.org/10.48550/arXiv.1802.04944 (2018).

  58. Gong, L. & Cheng, Q. Exploiting edge features for graph neural networks. In Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition (eds Michael S. B. et al.) 9203–9211 (IEEE, 2019); https://doi.org/10.1109/CVPR.2019.00943

  59. Dwivedi, V. P. & Bresson, X. A generalization of transformer networks to graphs. Preprint at https://doi.org/10.48550/arXiv.2012.09699 (2020).

  60. Bradley, A. P. The use of the area under the roc curve in the evaluation of machine learning algorithms. Pattern Recognit. 30, 1145–1159 (1997).

    Google Scholar 

  61. Xue, Y., Tong, Y. & Neri, F. An ensemble of differential evolution and Adam for training feed-forward neural networks. Inf. Sci. 608, 453–471 (2022).

    Google Scholar 

  62. Lu, W. et al. Tankbind: Trigonometry-aware neural networks for drug-protein binding structure prediction. In Advances in Neural Information Processing Systems 35 (eds Koyejo, S. et al.) 7236–7249 (NeurIPS, 2022); https://doi.org/10.1101/2022.06.06.495043

  63. Mendez, D. et al. ChEMBL: towards direct deposition of bioassay data. Nucleic Acids Res. 47, D930–D940 (2019).

    Google Scholar 

  64. Liu, T., Lin, Y., Wen, X., Jorissen, R. N. & Gilson, M. K. BindingDB: a web-accessible database of experimentally determined protein-ligand binding affinities. Nucleic Acids Res. 35, D198–D201 (2007).

    Google Scholar 

  65. Irwin, J. J. & Shoichet, B. K. ZINC—a free database of commercially available compounds for virtual screening. J. Chem. Inf. Model. 45, 177–182 (2005).

    Google Scholar 

  66. Truchon, J. F. & Bayly, C. I. Evaluating virtual screening methods: good and bad metrics for the ‘early recognition’ problem. J. Chem. Inf. Model. 47, 488–508 (2007).

    Google Scholar 

  67. Cao, D., Chen, G., Jiang, J. & Zheng, M. PDBscreen with multiple data augmentation strategies suitable for training protein-ligand interaction prediction methods. Zenodo https://doi.org/10.5281/zenodo.8049380 (2023).

  68. Cao, D., Chen, G., Jiang, J., Yu, J. & Zheng, M. TEST dataset pocket for EquiScore. Zenodo https://doi.org/10.5281/zenodo.8047224 (2023).

  69. Cao, D. & Chen, G. Original data and supplementary information for ‘EquiScore is a generic protein–ligand interaction scoring method integrating physical prior knowledge with data-augmentation modeling’. Zenodo https://doi.org/10.5281/zenodo.10812637 (2023).

  70. Cao, D. Code for ‘EquiScore is a generic protein–ligand interaction scoring method integrating physical prior knowledge with data-augmentation modeling’. GitHub https://github.com/CAODH/EquiScore (2023).

  71. Cao, D. Code for ‘EquiScore is a generic protein–ligand interaction scoring method integrating physical prior knowledge with data-augmentation modeling’. Zenodo https://doi.org/10.5281/zenodo.10812534 (2023).

Download references

Acknowledgements

We gratefully acknowledge financial support from the National Natural Science Foundation of China (T2225002 and 82273855 to M.Z., 82204278 to X. Li), the National Key Research and Development Program of China (2023YFC2305904 to M.Z.), the Shanghai Municipal Science and Technology Major Project (to M.Z.), the SIMM-SHUTCM Traditional Chinese Medicine Innovation Joint Research Program (E2G805H to M.Z.) and the Youth Innovation Promotion Association CAS (2023296 to S.Z.). We also acknowledge the Shanghai Supercomputer Center for providing computing resources.

Author information

Author notes

  1. These authors contributed equally: Duanhua Cao, Geng Chen.

Authors and Affiliations

  1. Innovation Institute for Artificial Intelligence in Medicine of Zhejiang University, College of Pharmaceutical Sciences, Zhejiang University, Hangzhou, China

    Duanhua Cao & Mingyue Zheng

  2. Drug Discovery and Design Center, State Key Laboratory of Drug Research, Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai, China

    Duanhua Cao, Geng Chen, Jiaxin Jiang, Jie Yu, Runze Zhang, Mingan Chen, Wei Zhang, Lifan Chen, Feisheng Zhong, Yingying Zhang, Chenghao Lu, Xutong Li, Xiaomin Luo, Sulin Zhang & Mingyue Zheng

  3. University of Chinese Academy of Sciences, Beijing, China

    Geng Chen, Jie Yu, Runze Zhang, Wei Zhang, Lifan Chen, Feisheng Zhong, Xutong Li, Xiaomin Luo, Sulin Zhang & Mingyue Zheng

  4. School of Pharmaceutical Science and Technology, Hangzhou Institute for Advanced Study, UCAS, Hangzhou, China

    Geng Chen & Mingyue Zheng

  5. School of Physical Science and Technology, Shanghai Tech University, Shanghai, China

    Mingan Chen

  6. Lingang Laboratory, Shanghai, China

    Mingan Chen

  7. Division of Life Science and Medicine, University of Science and Technology of China, Hefei, China

    Yingying Zhang

  8. School of Chinese Materia Medica, Nanjing University of Chinese Medicine, Jiangsu, China

    Chenghao Lu & Mingyue Zheng

Authors
  1. Duanhua Cao
    View author publications

    Search author on:PubMed Google Scholar

  2. Geng Chen
    View author publications

    Search author on:PubMed Google Scholar

  3. Jiaxin Jiang
    View author publications

    Search author on:PubMed Google Scholar

  4. Jie Yu
    View author publications

    Search author on:PubMed Google Scholar

  5. Runze Zhang
    View author publications

    Search author on:PubMed Google Scholar

  6. Mingan Chen
    View author publications

    Search author on:PubMed Google Scholar

  7. Wei Zhang
    View author publications

    Search author on:PubMed Google Scholar

  8. Lifan Chen
    View author publications

    Search author on:PubMed Google Scholar

  9. Feisheng Zhong
    View author publications

    Search author on:PubMed Google Scholar

  10. Yingying Zhang
    View author publications

    Search author on:PubMed Google Scholar

  11. Chenghao Lu
    View author publications

    Search author on:PubMed Google Scholar

  12. Xutong Li
    View author publications

    Search author on:PubMed Google Scholar

  13. Xiaomin Luo
    View author publications

    Search author on:PubMed Google Scholar

  14. Sulin Zhang
    View author publications

    Search author on:PubMed Google Scholar

  15. Mingyue Zheng
    View author publications

    Search author on:PubMed Google Scholar

Contributions

M.Z. designed the study. D.C. developed the method and implemented the code. G.C. and D.C. collected and processed training data. D.C., G.C., J.J. and J.Y. benchmarked the methods. All authors contributed to the analysis of the results. D.C., G.C. and M.Z. wrote the paper. All authors read and approved the paper.

Corresponding author

Correspondence to Mingyue Zheng.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Machine Intelligence thanks the anonymous reviewers for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Ablation results for VS and analogs ranking tasks.

a: VS performance is measured by 1.0% EF on DEKOIS2.0 (number of data points n = 81). The white points in the violin plots represent the means for each bin. b: Analogs ranking is measured by Spearman’s coefficient on LeadOpt (number of data points n = 8). The white points represent the average of coefficient values weighted by the number of ligands in each group (details on sample size for each group in Table 1).

Source data

Extended Data Table 1 Statistics of PDBscreen
Full size table
Extended Data Table 2 Spearman Correlation Coefficients on LeadOpt
Full size table
Extended Data Table 3 Statistics of PDBbind2020, CASF-2016, DUD-E, and DEKOIS2.0
Full size table
Extended Data Table 4 List of Node and Edge Features
Full size table

Supplementary information

Supplementary Information (download PDF )

Supplementary Tables 1–4, Figs. 1–5 and other supplementary content.

Reporting Summary (download PDF )

Supplementary Data 1 (download XLSX )

Raw data for Supplementary Fig. 3.

Supplementary Data 2 (download XLSX )

Raw data for Supplementary Fig. 4.

Source data

Source Data Fig. 3 (download XLSX )

Raw data for Fig. 3 and Supplementary Fig. 1.

Source Data Fig. 4 (download XLSX )

Raw data for Fig. 4 and Supplementary Fig. 2.

Source Data Fig. 5 (download XLSX )

Raw data for Fig. 5.

Source Data Extended Data Fig. 1 (download XLSX )

Raw data for Extended Data Fig. 1.

Source Data Extended Data Table 2 (download XLSX )

Raw data for Extended Data Table 2.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Cao, D., Chen, G., Jiang, J. et al. Generic protein–ligand interaction scoring by integrating physical prior knowledge and data augmentation modelling. Nat Mach Intell 6, 688–700 (2024). https://doi.org/10.1038/s42256-024-00849-z

Download citation

  • Received:

  • Accepted:

  • Published:

  • Version of record:

  • Issue date:

  • DOI: https://doi.org/10.1038/s42256-024-00849-z

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing