摘要

Molecular graph theory was used to design a unique and diverse, high-efficiency fragment screening collection. A data set retrieved from the annotated database AurSCOPE GPS was used as the reference set, and the GDB-13 database, a virtual library of enumerated organic molecules, was used as a source for the fragment selection. The data graph collection of Discngine as implemented in PipelinePilot was applied to perform the graph pharmacophore similarity matching between the reference and the GDB-13 data sets, leading to the ultimate fragment screening library. The relevance of this unique fragment collection was demonstrated by means of a virtual screening exercise using human trypsin as a test case. Several novel entities with high similarity to known trypsin inhibitors were identified in the in silico exercise. The application of this unique, high fragment efficiency collection to other protein targets in the framework of fragment-based drug discovery is warranted.

  • 出版日期2010-12