[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
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.

  • Matters Arising
  • Published:

Reusability report: Capturing properties of biological objects and their relationships using graph neural networks

Nature Machine Intelligence volume 4, pages 222–226 (2022)Cite this article

Subjects

The Original Article was published on 12 April 2021

Access through your institution
Buy or subscribe

arising from R. Schulte-Sasse et al. Nature Machine Intelligence https://doi.org/10.1038/s42256-021-00325-y (2021).

Graph neural networks (GNNs)1, especially graph convolutional networks (GCNs)2, have been increasingly used to model complex interactions. A basic idea behind GNNs is that some properties of an object, represented by a node in a graph, are reflected by properties of the objects that it directly and indirectly interacts with, where direct interactions are represented by edges in the graph. In biomedicine, GNNs have been used in a variety of applications, such as predicting protein functions and drug–disease associations3. Schulte-Sasse et al. have recently proposed a new use for GCNs in biomedicine: identifying cancer genes4. Their method, EMOGI (explainable multiomics graph integration), integrates multiomic data by aggregating information over a protein–protein interaction (PPI) network. The integrated information was shown to predict cancer genes better than having multiomic data alone or PPI connections alone.

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

Access options

Access through your institution

Buy this article

  • Purchase on SpringerLink
  • Instant access to full article PDF
Buy now

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

Fig. 1: Reproducing the main reported performance results of EMOGI.
Fig. 2: Performance (AUPRC values) of predicting cancer genes using EMOGI but with its PPI network replaced by co-expression network.
Fig. 3: Performance (AUPRC values) of predicting essential genes using EMOGI and baseline methods.
Fig. 4: Comparison of GATs and GCNs in the prediction of cancer genes.

Data availability

The data used in our study are available at https://github.com/kevingroup/emogi-reusability23. All the data used in the original paper by Schulte-Sasse et al. for testing EMOGI are available at http://owww.molgen.mpg.de/~sasse/EMOGI/.

Code availability

The original EMOGI code is available at https://github.com/schulter/EMOGI. Our modified GAT-based version of it is available at https://github.com/kevingroup/emogi-reusability23.

References

  1. Scarselli, F., Gori, M., Tsoi, A. C., Hagenbuchner, M. & Monfardini, G. The graph neural network model. IEEE Trans. Neural Networks 20, 61–80 (2009).

    Article  Google Scholar 

  2. Micheli, A. Neural network for graphs: a contextual constructive approach. IEEE Trans. Neural Networks 20, 498–511 (2009).

    Article  Google Scholar 

  3. Yue, X. et al. Graph embedding on biomedical networks: methods, applications and evaluations. Bioinformatics 36, 1241–1251 (2020).

    Google Scholar 

  4. Schulte-Sasse, R., Budach, S., Hnisz, D. & Marsico, A. Integration of multiomics data with graph convolutional networks to identify new cancer genes and their associated molecular mechanisms. Nat. Mach. Intell. 3, 513–526 (2021).

    Article  Google Scholar 

  5. No node features. GitHub https://github.com/tkipf/gcn/issues/10 (3 March 2017).

  6. Prechelt, L. In Neural Networks: Tricks of the Trade 55–69 (Springer, 1998).

  7. Cui, H., Lu, Z., Li, P. & Yang, C. On positional and structural node features for graph neural networks on non-attributed graphs. In Proc. Workshop of Deep Learning on Graphs: Methods and Applications, The 27th International ACM SIGKDD Conference on Knowledge Discovery and Data Mining (2021).

  8. Obayashi, T., Kagaya, Y., Aoki, Y., Tadaka, S. & Kinoshita, K. COXPRESdb v7: a gene coexpression database for 11 animal species supported by 23 coexpression platforms for technical evaluation and evolutionary inference. Nucleic Acids Res. 47, D55–D62 (2019).

    Article  Google Scholar 

  9. Zhang, X., Xiao, W. & Xiao, W. DeepHE: accurately predicting human essential genes based on deep learning. PLoS Comput. Biol. 16, e1008229 (2020).

    Article  Google Scholar 

  10. Luo, H., Lin, Y., Gao, F., Zhang, C.-T. & Zhang, R. DEG 10, an update of the database of essential genes that includes both protein-coding genes and noncoding genomic elements. Nucleic Acids Res. 42, D574–D580 (2014).

    Article  Google Scholar 

  11. Wu, Z. et al. A comprehensive survey on graph neural networks. IEEE Trans. Neural Networks Learn. Syst. 32, 4–24 (2020).

    Article  MathSciNet  Google Scholar 

  12. Hamilton, W. L., Ying, R. & Leskovec, J. Inductive representation learning on large graphs. In Proc. 31st Int. Conf. Neural Information Processing Systems 1025–1035 (2017).

  13. Veličković, P. et al. Graph attention networks. In 6th Int. Conf. Learning Representations (2018).

  14. Xu, K., Hu, W., Leskovec, J. & Jegelka, S. How powerful are graph neural networks? In 7th Int. Conf. Learning Representations (2019).

  15. Wang, M. et al. Deep graph library: A graph-centric, highly-performant package for graph neural networks. Preprint at https://arxiv.org/abs/1909.01315 (2019).

  16. Fanfani, V., Torne, R. V., Lio’, P. & Stracquadanio, G. Discovering cancer driver genes and pathways using stochastic block model graph neural networks. Preprint at bioRxiv https://doi.org/10.1101/2021.06.29.450342 (2021).

  17. Schulte-Sasse, R.,Budach, S., Hnisz, D. & Marsico, A. Graph convolutional networks improve the prediction of cancer driver genes. In Int. Conf. Artificial Neural Networks 658–668 (2019).

  18. Davis, J. & Goadrich, M. The relationship between precision-recall and roc curves. In Proc. 23rd Int. Conf. Machine Learning 233–240 (2006).

  19. You, J., Ying, R. & Leskovec, J. Position-aware graph neural networks. In Int. Conf. Machine Learning 7134–7143 (2019).

  20. Kipf, T. N. & Welling, M. Semi-supervised classification with graph convolutional networks. In 5th Int. Conf. Learning Representations (2017).

  21. About the data’s feature. GitHub https://github.com/tkipf/gcn/issues/22 (5 February 2018).

  22. Cao, Q. et al. A unified framework for integrative study of heterogeneous gene regulatory mechanisms. Nat. Mach. Intell. 2, 447–456 (2020).

    Article  Google Scholar 

  23. Hong, C., Cao, Q. & Zhang, Z. EMOGI-reusability v1.0 https://doi.org/10.5281/zenodo.5914506 (2022).

Download references

Acknowledgements

We thank Schulte-Sasse et al. for their help in our running of EMOGI and for sharing their processed data. Q.C. is supported by National Natural Science Foundation of China Youth Program 32100515. K.Y.Y. was supported by Hong Kong Research Grants Council Collaborative Research Funds C4015-20E, C4045-18W, C4057-18E and C7044-19G and General Research Funds 14107420 and 14203119, the Hong Kong Epigenomics Project (EpiHK), and the Chinese University of Hong Kong Young Researcher Award, Outstanding Fellowship and Project Impact Enhancement Fund.

Author information

Author notes

  1. Kevin Y. Yip

    Present address: Sanford Burnham Prebys Medical Discovery Institute, La Jolla, CA, USA

  2. These authors contributed equally: Chenyang Hong, Qin Cao.

Authors and Affiliations

  1. Department of Computer Science and Engineering, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong

    Chenyang Hong, Qin Cao, Zhenghao Zhang & Kevin Y. Yip

  2. School of Biomedical Sciences, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong

    Qin Cao & Stephen Kwok-Wing Tsui

  3. Hong Kong Bioinformatics Centre, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong

    Stephen Kwok-Wing Tsui & Kevin Y. Yip

  4. Hong Kong Institute of Diabetes and Obesity, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong

    Kevin Y. Yip

Authors
  1. Chenyang Hong
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Qin Cao
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Zhenghao Zhang
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Stephen Kwok-Wing Tsui
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Kevin Y. Yip
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

Q.C. and K.Y.Y. conceived and supervised the project. C.H., Q.C., Z.Z., S.K.T. and K.Y.Y. designed the computational experiments and data analyses. C.H. and Z.Z. prepared the data. C.H. and Q.C. implemented the methods, conducted the experiments and performed data analyses. C.H., Q.C. and K.Y.Y. wrote the manuscript.

Corresponding authors

Correspondence to Qin Cao or Kevin Y. Yip.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Machine Intelligence thanks Marinka Zitnik for her 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.

Supplementary information

Supplementary Information

Supplementary Fig. 1 and Tables 1–3.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Hong, C., Cao, Q., Zhang, Z. et al. Reusability report: Capturing properties of biological objects and their relationships using graph neural networks. Nat Mach Intell 4, 222–226 (2022). https://doi.org/10.1038/s42256-022-00454-y

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s42256-022-00454-y

Search

Advanced search

Quick links

Nature Briefing: Cancer

Sign up for the Nature Briefing: Cancer newsletter — what matters in cancer research, free to your inbox weekly.

Get what matters in cancer research, free to your inbox weekly. Sign up for Nature Briefing: Cancer