Derek Lowe of In the Pipeline has previously commented on the fact that relying on the same old reliable chemical transformations may restrict medicinal chemists to particular regions of chemical space. In order to reach beyond these, it is necessary to use reactions that are “often not so general and reliable”.
As part of a poster presentation at the recent Bio-IT World meeting in Boston, Roger has analysed the Electronic Lab Notebook of a major pharmaceutical company using our NameRXN software to extract the major classes of reaction as well as the specific top five transformations:
Update 2013/04/22: See Derek Lowe’s analysis of these results over at In the Pipeline.
The full poster explains how the data was extracted from the PKI/CambridgeSoft ELN using HazELNut, and also shows how data quality issues in the reaction entry may be flagged up by combining the NameRXN information with atom-mapping algorithms (PDF version):