EUVIMEDEUVIMED entdecken
Crossref

Clarification of Data Set Origin and Attribution in “Machine Learning for Toxicity Prediction Using Chemical Structures: Pillars for Success in the Real World”

Sankalp Jain, Vishal B. Siramshetty, Eugene N. Muratov, Alexander Tropsha, Alexey V. Zakharov

Chemical Research in Toxicology · 10.11.2026

DOI: 10.1021/acs.chemrestox.6c00453 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Chemical Research in Toxicology, online erschienen am 10.11.2026. Verfasst von Sankalp Jain und 4 weiteren Autor:innen. Die Arbeit nennt 5 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

I n their recent Perspective, "Machine Learning for Toxicity Prediction Using Chemical Structures: Pillars for Success in the Real World," Seal et al. 1 provided a thoughtful and timely overview of several key factors that shape the successful application of machine learning in toxicity prediction.We appreciate the authors' effort to highlight considerations around data set selection, molecular representation, model development, validation, and translational use in a manner that will be valuable to the broader toxicology community.We write, however, to clarify an important attribution error regarding a publicly available acute systemic toxicity data set discussed in the article.Seal et al. state "Recently, one of the largest data set with in vivo toxicity data for 80,000 compounds against a total of 59 acute systemic toxicity end points was analyzed by Sosnin et al. and later made publicly available by Jain et al." 1 We wish to emphasize that this attribution does not accurately reflect the origin of the described data set, and, therefore, is incorrect.The data set was not first analyzed by Sosnin et al. 2 and subsequently released by Jain et al. 3 Rather, the data set comprising 80,081 unique compounds, 122,594 measurements, and 59 acute systemic toxicity end points was collected, curated, integrated, modeled, and made publicly available in our study published in the Journal of Chemical Information and Modeling in 2021.3 One of the explicitly stated primary goals of the study was to "collect, curate, and integrate all available data for systemic acute toxicity into the largest publicly available multispecies toxicity data set" 3 and to use this resource for benchmarking single-task and multitask machine learning methods across all 59 end points.Our article also made both the curated data set and prediction models publicly accessible through NCATS Predictor, GitHub, and the NCI Cactus resource.3 Thus, the collection, curation, modeling, and dissemination of the 59 end point data set were integral parts of the Jain et al. 3 study itself, rather than a later release of material previously analyzed elsewhere.The work of Sosnin et al., by contrast, was discussed in our paper as prior work that reported multitask toxicity modeling across 29 end points using data from the commercial RTECS database.Specifically, in the "Comparison to Previous Studies" section, we explicitly stated: "Outside of our own presentation at the CATMOS meeting, only Sosnin et al. reported multitask models for a total of 29 toxicity end points."Our study independently obtained the raw toxicity data from the publicly available ChemIDplus resource and applied a

Zitieren

APA

Jain, S., Siramshetty, V. B., Muratov, E. N., Tropsha, A., & Zakharov, A. V. (2026). Clarification of Data Set Origin and Attribution in “Machine Learning for Toxicity Prediction Using Chemical Structures: Pillars for Success in the Real World”. Chemical Research in Toxicology. https://doi.org/10.1021/acs.chemrestox.6c00453

Vancouver

Jain S, Siramshetty VB, Muratov EN, Tropsha A, Zakharov AV. Clarification of Data Set Origin and Attribution in “Machine Learning for Toxicity Prediction Using Chemical Structures: Pillars for Success in the Real World”. Chemical Research in Toxicology. 2026. doi:10.1021/acs.chemrestox.6c00453

BibTeX
@article{Jain2026,
  author = {Jain, Sankalp and Siramshetty, Vishal B. and Muratov, Eugene N. and Tropsha, Alexander and Zakharov, Alexey V.},
  title = {Clarification of Data Set Origin and Attribution in “Machine Learning for Toxicity Prediction Using Chemical Structures: Pillars for Success in the Real World”},
  journal = {Chemical Research in Toxicology},
  year = {2026},
  doi = {10.1021/acs.chemrestox.6c00453}
}
Kostenlos starten →