Kalliokoski Tuomo, Ronkko Toni, Poso Antti
Department of Pharmaceutical Chemistry, University of Kuopio, Kuopio, Finland.
J Chem Inf Model. 2008 Jun;48(6):1131-7. doi: 10.1021/ci700216u. Epub 2008 May 20.
Algorithms were developed for ligand-based virtual screening of molecular databases. FieldChopper (FC) is based on the discretization of the electrostatic and van der Waals field into three classes. A model is built from a set of superimposed active molecules. The similarity of the compounds in the database to the model is then calculated using matrices that define scores for comparing field values of different categories. The method was validated using 12 publicly available data sets by comparing the method to the electrostatic similarity comparison program EON. The results suggest that FC is competitive with more complex descriptors and could be used as a molecular sieve in virtual screening experiments when multiple active ligands are known.
已开发出用于基于配体的分子数据库虚拟筛选的算法。FieldChopper（FC）基于将静电场和范德华场离散化为三类。从一组叠加的活性分子构建模型。然后使用定义不同类别场值比较分数的矩阵来计算数据库中化合物与模型的相似性。通过将该方法与静电相似性比较程序EON进行比较，使用12个公开可用数据集对该方法进行了验证。结果表明，FC与更复杂的描述符具有竞争力，并且在已知多个活性配体时可在虚拟筛选实验中用作分子筛。
