基于骨架结构分类法的递进式药物筛选(英文)

Progressive screening using scaffold-based classification approach

本文报告一种新颖的基于骨架结构分类法的递进式药物筛选(PS-SCA)技术。此技术在美国国立癌症研究所(NCI)的八组细胞水平上的高通量筛选实验数据上进行了测试,结果表明,基于拓扑结构数据的递进式筛选可以极大地降低药物筛选的代价,缩短筛选时间。模拟实验证明,PS-SCA递进式筛选技术包括三个阶段:(1)骨架多样性采样筛选试验:(2)活性化合物发现试验;(3)可忽略的多余筛选试验。运用PS-SCA递进式筛选技术,可以在只筛选20％的化合物情况下找到最有意义的70-80％的活性化合物。而且,这70％-80％的活性化合物中包含了关键的结构骨架。

Progressive screening using scaffold-based classification approach （PS-SCA） is presented in the paper. The process has been tested on eight NCI cell line HTS data sets. The results show that 2D structure based progressive screening can expedite hit and lead identification processes with dramatically reduced costs. It is discovered that the PS-SCA process only consists of three types of virtual screening trials： （1） Seeding trial, the first trial, which selects compounds based on scaffold diversity, and find initial seed-hits to discover more hits in further screening trials, （2） Fishing trial, the second trial, which discovers all promising hits based on the knowledge of the seed-hits, and （3） Rest trials, which can only find additional or random hits, and therefore can be skipped in screening process to reduce hit/lead identification costs. The PS-SCA process experiments on eight NCI screening data sets show that it can discover 70%-80% of hits by screening only 20% compounds in the inventory, and hop privileged scaffolds from HTS data.