Publisher Correction: Extending machine learning beyond interatomic potentials for predicting molecular properties

Nikita Fedik, R.i. Zubatyuk, Maksim Kulichenko, Nicholas Lubbers, Justin S. Smith, Benjamin Nebgen, Richard A. Messerly, Ying Wai Li, Alexander I. Boldyrev, Kipton Barros, Olexandr Isayev, Sergei TretiakNature Reviews Chemistry 2022 · DOI

TL;DR

This publisher correction addresses a typographical error and a data presentation issue in a previous article. The correction clarifies that the machine learning model for predicting molecular properties was initially mislabeled in a figure and that a different dataset was inadvertently presented. This ensures the accuracy of the published research on extending machine learning beyond traditional interatomic potentials.

What Problem Does This Paper Address?

This document is a publisher correction, not a research paper itself. It addresses errors found in a previously published article titled "Extending machine learning beyond interatomic potentials for predicting molecular properties." The original article likely aimed to improve the prediction of molecular properties using machine learning, potentially by moving beyond the limitations of traditional interatomic potentials. The correction specifically points out two issues: a typo in a chemical formula presented in Figure 5c and the incorrect inclusion of data from a different source article in the same figure. These errors could lead to misinterpretations of the model's performance and the underlying scientific findings.

Methods

This document does not describe methods as it is a correction notice. The original paper, which this notice corrects, likely employed machine learning techniques to develop models for predicting molecular properties. The correction implies that the original paper used a specific machine learning model (referred to as "4G-HDNNP" and "3G-HDNNP") and presented results in a figure. The correction itself does not introduce new methods but clarifies the presentation of existing ones.

Datasets

This document does not explicitly list datasets used for training or evaluation. However, it references two sources of data:

  • The original article's data, which includes results from a "4G-HDNNP" model.
  • A different source article (Ko, T. W. et al. Nat. Commun. https://doi.org/10.1038/s41467-020-20427-2 (2021)), from which data was inadvertently included in the original publication.

The correction indicates that the figure has been updated in the HTML and PDF versions of the article to reflect the correct data.

Key Insights and Ideas

  • Accuracy in Scientific Reporting is Paramount: This correction highlights the critical importance of meticulous accuracy in scientific publications. Even seemingly small errors, like a typo in a chemical formula or the inclusion of incorrect data, can mislead readers and undermine the credibility of the research.
  • Reproducibility and Data Integrity: The correction underscores the need for researchers and publishers to ensure that the data presented accurately reflects the experiments and analyses performed. The inadvertent inclusion of data from another source is a serious issue for reproducibility.
  • Evolving Machine Learning Models: The mention of "4G-HDNNP" and "3G-HDNNP" suggests a progression in machine learning models for molecular property prediction, with newer versions (like 4G) building upon or replacing older ones (like 3G).

Key Results

This document does not present quantitative results. It only details the corrections made to the original article:

  • Typographical Error Correction: The chemical formula "C10H12" in the top-left of Figure 5c has been corrected to "C10H2".
  • Data Presentation Correction: The "4G-HDNNP" result in the bottom section of Figure 5c was incorrectly preceded by the "3G-HDNNP" result from a different source article. This has been rectified.

Limitations and Future Work

This document is a correction notice and does not discuss limitations or future work of the original research. It solely focuses on rectifying errors in the published article.

Summary updated

BibTeX
@article{fedik2022publisher,
  author = {Nikita Fedik and R.I. Zubatyuk and Maksim Kulichenko and Nicholas Lubbers and Justin S. Smith and Benjamin Nebgen and Richard A. Messerly and Ying Wai Li and Alexander I. Boldyrev and Kipton Barros and Olexandr Isayev and Sergei Tretiak},
  title = {{Publisher Correction: Extending machine learning beyond interatomic potentials for predicting molecular properties}},
  journal = {Nature Reviews Chemistry},
  year = {2022},
  doi = {10.1038/s41570-022-00446-x},
}

Loading related content...

Figure