S4: Computational approaches to acute mammalian systemic toxicity
Computational approaches to acute mammalian systemic toxicity
There is great interest in assessing the in vivo toxicity of chemicals by using non-animal alternative approaches. Since acute mammalian toxicity is not adequately predicted by current computational or in vitro models, our goal is to develop a tiered framework wherein salient safety information can be quickly derived early in product development. Towards this end, we (1) compiled large inventories of acute toxicity data from public and in-house data, (2) systematically catalogued potential molecular initiating events (MIEs) associated with acute toxicity, and (3) developed molecular profiling capabilities to bin compounds by MIE followed by development of prediction models using molecular docking or simple regression plots within regional compound classes. As a proof-of-concept, we focused on predicting acute toxicity for several MIEs. We are developing a workflow that integrates implementation of computational mechanistic screening with targeted in vitro methods to derive estimates of acute toxicity early in new-product development.