One gap in current patent-mining practice is the lack of tools to build SAR knowledge. Here, we report a novel technique that enabled us to derive useful SAR information from the exemplified structures of a series of patents. In our approach, exemplified chemical structures were extracted from patent documents. They were grouped into structural series based on similarity and binding mode, after which the R-group table was generated. By analyzing R-group usages over time, we were able to build insights into SAR of a structural series, even though the biological activities were not available.
Author(s): Meihua Tu, Jeffrey A Pfefferkorn1, Angel Guzman-Perez, Kevin J Filipski
Organization(s): Pfizer Worldwide Research & Development
Source: Pharmaceutical Patent Analyst
Year: 2012
http://www.future-science.com/doi/abs/10.4155/ppa.12.60?journalCode=ppa