Semantic knowledge improves molecular machine learning for chemical toxicity prediction
Giuseppe Albi, Arianna Dagliati, Riccardo Bellazzi, Joseph D Romano · 2026
Read Semantic knowledge improves molecular machine learning for chemical toxicity prediction by Giuseppe Albi, Arianna Dagliati, Riccardo Bellazzi, Joseph D Romano (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13427666.txt.
Description
Giuseppe Albi and colleagues show, in a 2026 study, that adding background knowledge helps computers predict whether a chemical is toxic. Classic models read only a molecule's structure and physical traits. The team instead fed graph neural networks extra facts from knowledge graphs, which link chemicals, genes, and known effects in plain relations. Tests used public toxicity datasets, including the Tox21 assay collection.
Models with the added knowledge separated toxic from harmless compounds better than structure-only models. Their confidence estimates stayed just as well calibrated. Explanation tools showed the models focused more on molecular parts already known to drive toxic effects. Safety assessors who screen chemicals before lab testing gain a model that is both sharper and easier to interrogate.
Catalog record
| CREATOR | Giuseppe Albi, Arianna Dagliati, Riccardo Bellazzi, Joseph D Romano |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 87390 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |