Lhasa Limited, 22-23 Blenheim Terrace, Woodhouse Lane, Leeds, LS2 9HD, UK.
Chem Biodivers. 2009 Nov;6(11):2107-14. doi: 10.1002/cbdv.200900133.
Hepatotoxicity is a major cause of pharmaceutical drug attrition and is also a concern within other chemical industries. In silico approaches to the prediction of hepatotoxicity are an important tool in the early identification of adverse effects in the liver associated with exposure to a chemical. Here, we describe work in progress to develop an expert system approach to the prediction of hepatotoxicity, focussing particularly on the identification of structural alerts associated with its occurrence. The development of 74 such structural alerts based on public-domain literature and proprietary data sets is described. Evaluation results indicate that, whilst these structural alerts are effective in identifying the hepatotoxicity of many chemicals, further research is needed to develop additional structural alerts to account for the hepatotoxicity of a number of chemicals which is not currently predicted. Preliminary results also suggest that the specificity of the structural alerts may be improved by the combined use of applicability domains based on physicochemical properties such as log P and molecular weight. In the longer term, the performance of predictive models is likely to benefit from the further integration of diverse data and prediction model types.
肝毒性是药物淘汰的一个主要原因，也是其他化学工业关注的问题。药物肝毒性的计算预测方法是在早期识别与化学物质接触相关的肝脏不良影响的重要工具。在这里，我们描述了开发一种用于预测肝毒性的专家系统方法的进展，特别关注与肝毒性发生相关的结构警示的识别。描述了基于公共领域文献和专有数据集开发的 74 种此类结构警示。评估结果表明，虽然这些结构警示可有效识别许多化学物质的肝毒性，但需要进一步研究来开发更多的结构警示，以涵盖目前未预测到的一些化学物质的肝毒性。初步结果还表明，通过使用基于物理化学性质（如 log P 和分子量）的适用性域，可以提高结构警示的特异性。从长远来看，通过进一步整合不同的数据和预测模型类型，预测模型的性能可能会得到改善。
