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: Designing organic photoelectronic molecules with descriptor conditional recurrent neural networks

Nature Machine Intelligence volume 2, pages 749–752 (2020)Cite this article

Subjects

The Original Article was published on 18 May 2020

arising from P.-C. Kotsias et al. Nature Machine Intelligence https://doi.org/10.1038/s42256-020-0174-5 (2020)

Deep generative models can be trained on unlabelled chemical data to design novel molecules, but it’s challenging to harness the creativity of such models to finding optimal molecules1. In their recent work2, Bjerrum and colleagues present a generative framework based on conditional recurrent neural networks (cRNNs) to translate from a desired property to a string-based chemical representation in the context of drug design. Here we replicate the approach on an unrelated chemical space by designing organic photoelectronic molecules (OPMs) with properties outside the training data. The primary application in the original work was a classification task (identifying active molecules) whereas the task here is to propose molecules with continuous properties close to target values.

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

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

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

Fig. 1: Repurposing cRNN generates novel OPMs.
Fig. 2: Benchmarking cRNN models against a GB-GA baseline.

Data availability

The chemical structures and labels used for training and validation of the supervised and unsupervised models, with the exception of 684 proprietary molecules, are available at https://github.com/learningmatter-mit/Deep-Drug-Coder20.

Code availability

The code used in this paper is available at https://github.com/learningmatter-mit/Deep-Drug-Coder.

References

  1. Schwalbe-Koda, D. & Gómez-Bombarelli, R. in Lecture Notes in Physics Vol. 968 (eds Schütt, K. T. et al.) 445–467 (Springer, 2020).

  2. Kotsias, P.-C. et al. Direct steering of de novo molecular generation using descriptor conditional recurrent neural networks (cRNNs). Nat. Mach. Intell. 2, 254–265 (2020).

    Article  Google Scholar 

  3. Morgan, H. L. The generation of a unique machine description for chemical structures-a technique developed at chemical abstracts service. J. Chem. Doc. 5, 107–113 (1965).

    Article  Google Scholar 

  4. Segler, M. H. S., Kogej, T., Tyrchan, C. & Waller, M. P. Generating focused molecule libraries for drug discovery with recurrent neural networks. ACS Cent. Sci. 4, 120–131 (2018).

    Article  Google Scholar 

  5. Arús-Pous, J. et al. Exploring the GDB-13 chemical space using deep generative models. J. Cheminform. 11, 20 (2019).

    Article  Google Scholar 

  6. Popova, M., Isayev, O. & Tropsha, A. Deep reinforcement learning for de novo drug design. Sci. Adv. 4, eaap7885 (2018).

    Article  Google Scholar 

  7. Gómez-Bombarelli, R. et al. Design of efficient molecular organic light-emitting diodes by a high-throughput virtual screening and experimental approach. Nat. Mater. 15, 1120–1127 (2016).

    Article  Google Scholar 

  8. Hachmann, J. et al. Lead candidates for high-performance organic photovoltaics from high-throughput quantum chemistry – the Harvard Clean Energy Project. Energy Environ. Sci. 7, 698–704 (2014).

    Article  Google Scholar 

  9. Kotsias, P. & Bjerrum, E. J. Deep-Drug-Coder v1.0.0 https://doi.org/10.5281/zenodo.3739063 (accessed 15 May 2020).

  10. Gueymard, C. A. The sun’s total and spectral irradiance for solar energy applications and solar radiation models. Sol. Energy 76, 423–453 (2004).

    Article  Google Scholar 

  11. Jensen, J. H. A graph-based genetic algorithm and generative model/Monte Carlo tree search for the exploration of chemical space. Chem. Sci. 10, 3567–3572 (2019).

    Article  Google Scholar 

  12. Jin, W., Barzilay, R. & Jaakkola, T. Domain extrapolation via regret minimization. Preprint at https://arxiv.org/abs/2006.03908 (2020).

  13. Gómez-Bombarelli, R. et al. Automatic chemical design using a data-driven continuous representation of molecules. ACS Cent. Sci. 4, 268–276 (2018).

    Article  Google Scholar 

  14. Krenn, M., Häse, F., Nigam, A., Friederich, P. & Aspuru-Guzik, A. Self-referencing embedded strings (SELFIES): a 100% robust molecular string representation. Mach. Learn. Sci. Technol. 1, 045024 (2020).

    Article  Google Scholar 

  15. Kusner, M. J., Paige, B. & Hernández-Lobato, J. M. Grammar variational autoencoder. In Proc. 34th International Conference on Machine Learning (eds Precup, D. & Teh, Y. W.) 1945–1954 (2017).

  16. Dai, H., Tian, Y., Dai, B., Skiena, S. & Song, L. Syntax-directed variational autoencoder for molecule generation. In Proc. International Conference on Learning Representations (ICLR, 2018).

  17. Joulin, A. & Mikolov, T. Inferring algorithmic patterns with stack-augmented recurrent nets. In Advances in Neural Information Processing Systems (2015).

  18. Moniz, J. R. A. & Krueger, D. Nested LSTMs. In Proc. Asian Conference on Machine Learning (PMLR, 2017).

  19. Maziarka, Ł. et al. Molecule attention transformer. Preprint at https://arxiv.org/abs/2002.08264 (2020).

  20. Mohapatra, S., Yang, T. & Gomez-Bombarelli, R. OPM-cRNN v0.1-OPM https://doi.org/10.5281/zenodo.4073289 (2020).

  21. Landrum, G. RDKit: Open-source cheminformatics v2018.09.1 https://www.rdkit.org/docs/index.html (2006).

Download references

Acknowledgements

We acknowledge Sumitomo Chemical for providing financial support for this work.

Author information

Authors and Affiliations

  1. Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA

    Somesh Mohapatra, Tzuhsiung Yang & Rafael Gómez-Bombarelli

Authors
  1. Somesh Mohapatra
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Tzuhsiung Yang
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Rafael Gómez-Bombarelli
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

R.G.-B. supervised the research, and planned the project with contributions from S.M. S.M. trained and analysed the machine learning models. T.Y. ran the DFT calculations with contributions from R.G.-B. All authors contributed to the writing of the manuscript.

Corresponding author

Correspondence to Rafael Gómez-Bombarelli.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature Machine Intelligence thanks Olexandr Isayev, Connor Coley and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

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

Supplementary information

Supplementary Information

Supplementary Discussion Sections 1–8, Figs. 1–3 and Tables 1–5.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Mohapatra, S., Yang, T. & Gómez-Bombarelli, R. Reusability report: Designing organic photoelectronic molecules with descriptor conditional recurrent neural networks. Nat Mach Intell 2, 749–752 (2020). https://doi.org/10.1038/s42256-020-00268-w

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s42256-020-00268-w

This article is cited by

Search

Advanced search

Quick links

Nature Briefing AI and Robotics

Sign up for the Nature Briefing: AI and Robotics newsletter — what matters in AI and robotics research, free to your inbox weekly.

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