Department of Life Science Informatics, B-IT, LIMES Program Unit Chemical Biology and Medicinal Chemistry, Rheinische Friedrich-Wilhelms-Universität, Dahlmannstrasse 2, D-53113 Bonn, Germany.
J Chem Inf Model. 2013 Jun 24;53(6):1263-71. doi: 10.1021/ci400165a. Epub 2013 May 20.
We have searched for chemical transformations that improve drug development-relevant properties within a given class of active compounds, regardless of the compounds they are applied to. For different compound data sets, varying numbers of frequently occurring data set-dependent transformations were identified that consistently induced favorable changes of selected molecular properties. Sequences of compound pairs representing such transformations were determined that formed pathways leading from unfavorable to favorable regions of property space. Data set-dependent transformations were then applied to predict a series of compounds with increasingly favorable property values. By database searching the desired biological activity was detected for several designed molecules or compounds that were very similar to these molecules. Taken together our findings indicate that data set-dependent transformations can be applied to predict compounds that map to favorable regions of molecular property space and retain their biological activity.
我们一直在寻找能够改善特定类别的活性化合物的药物研发相关性质的化学转化，而不考虑它们所应用的化合物。对于不同的化合物数据集，确定了数量不等的经常出现的数据集相关转化，这些转化一致地诱导所选分子性质的有利变化。确定了代表这些转化的化合物对序列，这些序列形成了从不利区域到有利区域的途径。然后应用数据集相关的转化来预测一系列具有越来越好的性质值的化合物。通过数据库搜索，发现了几个设计分子或与这些分子非常相似的化合物具有所需的生物活性。总之，我们的研究结果表明，数据集相关的转化可以应用于预测那些映射到分子性质空间有利区域并保留其生物活性的化合物。
