ToxReason: A Benchmark for Mechanistic Chemical Toxicity Reasoning via Adverse Outcome Pathway
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913013082095616 |
|---|---|
| author | Park, Jueon Jang, Wonjune Kim, Chanhwi Park, Yein Kang, Jaewoo |
| author_facet | Park, Jueon Jang, Wonjune Kim, Chanhwi Park, Yein Kang, Jaewoo |
| contents | Recent advances in large language models (LLMs) have enabled molecular reasoning for property prediction. However, toxicity arises from complex biological mechanisms beyond chemical structure, necessitating mechanistic reasoning for reliable prediction. Despite its importance, current benchmarks fail to systematically evaluate this capability. LLMs can generate fluent but biologically unfaithful explanations, making it difficult to assess whether predicted toxicities are grounded invalid mechanisms. To bridge this gap, we introduce ToxReason, a benchmark grounded in the Adverse Outcome Pathway (AOP) that evaluates organ-level toxicity reasoning across multiple organs. ToxReason integrates experimental drug-target interaction evidence with toxicity labels, requiring models to infer both toxic outcomes and their underlying mechanisms from Molecular Initiating Event (MIE) to Adverse Outcome (AO). Using ToxReason, we evaluate toxicity prediction performance and reasoning quality across diverse LLMs. We find that strong predictive performance does not necessarily imply reliable reasoning. Furthermore, we show that reasoning-aware training improves mechanistic reasoning and, consequently, toxicity prediction performance. Together, these results underscore the necessity of integrating reasoning into both evaluation and training for trustworthy toxicity modeling. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_06264 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | ToxReason: A Benchmark for Mechanistic Chemical Toxicity Reasoning via Adverse Outcome Pathway Park, Jueon Jang, Wonjune Kim, Chanhwi Park, Yein Kang, Jaewoo Quantitative Methods Artificial Intelligence Recent advances in large language models (LLMs) have enabled molecular reasoning for property prediction. However, toxicity arises from complex biological mechanisms beyond chemical structure, necessitating mechanistic reasoning for reliable prediction. Despite its importance, current benchmarks fail to systematically evaluate this capability. LLMs can generate fluent but biologically unfaithful explanations, making it difficult to assess whether predicted toxicities are grounded invalid mechanisms. To bridge this gap, we introduce ToxReason, a benchmark grounded in the Adverse Outcome Pathway (AOP) that evaluates organ-level toxicity reasoning across multiple organs. ToxReason integrates experimental drug-target interaction evidence with toxicity labels, requiring models to infer both toxic outcomes and their underlying mechanisms from Molecular Initiating Event (MIE) to Adverse Outcome (AO). Using ToxReason, we evaluate toxicity prediction performance and reasoning quality across diverse LLMs. We find that strong predictive performance does not necessarily imply reliable reasoning. Furthermore, we show that reasoning-aware training improves mechanistic reasoning and, consequently, toxicity prediction performance. Together, these results underscore the necessity of integrating reasoning into both evaluation and training for trustworthy toxicity modeling. |
| title | ToxReason: A Benchmark for Mechanistic Chemical Toxicity Reasoning via Adverse Outcome Pathway |
| topic | Quantitative Methods Artificial Intelligence |
| url | https://arxiv.org/abs/2604.06264 |