Universität Hamburg, ZBH - Center for Bioinformatics, Research Group for Computational Molecular Design, Bundesstraße 43, 20146 Hamburg, Germany.
BioSolveIT GmbH, An der Ziegelei 79, 53757 Sankt Augustin, Germany.
J Chem Inf Model. 2022 Jun 13;62(11):2800-2810. doi: 10.1021/acs.jcim.2c00334. Epub 2022 Jun 2.
The distributions of physicochemical property values, like the octanol-water partition coefficient, are routinely calculated to describe and compare virtual chemical libraries. Traditionally, these distributions are derived by processing each member of a library individually and summarizing all values in a distribution. This process becomes impractical when operating on chemical spaces which surpass billions of compounds in size. In this work, we present a novel algorithmic method called SpaceProp for the property distribution calculation of large nonenumerable combinatorial fragment spaces. The novel method follows a combinatorial approach and is able to calculate physicochemical property distributions of prominent spaces like Enamine's REAL Space, WuXi's GalaXi Space, and OTAVA's CHEMriya Space for the first time. Furthermore, we present a first approach of optimizing property distributions directly in combinatorial fragment spaces.
理化性质值（如辛醇-水分配系数）的分布通常是通过计算来描述和比较虚拟化学文库的。传统上，这些分布是通过单独处理库中的每个成员并汇总分布中的所有值来得出的。当处理规模超过数十亿化合物的化学空间时，这个过程变得不切实际。在这项工作中，我们提出了一种名为 SpaceProp 的新算法方法，用于计算大型不可枚举组合片段空间的性质分布。该新方法采用组合方法，能够首次计算 Enamine 的 REAL Space、WuXi 的 GalaXi Space 和 OTAVA 的 CHEMriya Space 等重要空间的理化性质分布。此外，我们还提出了一种在组合片段空间中直接优化性质分布的初步方法。
