In silico ADMET tools: the dawn of a new generation?

During the late 1990s, the allure of in silico methods for absorption, distribution, metabolism, excretion and toxicity prediction led to the rapid development of what might be termed first-generation models. However, despite much initial enthusiasm, these largely proved disappointing when applied in real-world settings. Some of the reasons for this disappointment are examined in this perspective. The present status of the field, as second-generation models are beginning to emerge, is surveyed. Finally, the outstanding challenges in the field are delineated and some possible approaches to their resolution suggested.