ChemCurator, Computer-Assisted Patent Curation and Analysis Tool

post · 4 years ago
by David Deng, Árpád Figyelmesi (ChemAxon)

Because of the significant investment associated with the development of a new drug, chemical patents are essential to pharmaceutical industry for intellectual property protection. For the same reason, patent analysis is an important but very labor intensive task. In addition to numerous examples and assay data scattered across the pages, complex Markush structures are often used to make the patent claim as broad as possible. Extracting useful information through hundreds of pages can be time consuming even for experienced information scientists.

If you have attended our webinar series before, you probably know that ChemAxon has long been developing informatics tools for IP and life science researchers. In the past, we have introduced Document to Structure, which can automatically extract chemical names from all types of documents; while Markush Technology can create, enumerate and search complex Markush structures. Now, by combining these two technologies, in this webinar we present a new product, ChemCurator, to help expedite patent curation.

ChemCurator can open a patent document (or any other documents), and extract chemical information, including exemplified structures, R-group fragments, reactions, assay data etc., from texts and images in a semi-automated way. The structures are displayed next to their original location in the document, which makes navigation easier and more intuitive. Users can drag-and-drop to populate and merge R-group fragments together, and quickly re-assemble the Markush structure according to the claim. Exemplified structure can also be searched against the Markush structure to see if they are within the chemical space or not. All these features will be introduced in this webinar.

We believe ChemCurator can be a very useful tool to help scientists analyze patents. Although it cannot completely eliminate human intervention, it can greatly reduce the processing time while confirming that the output Markush structure represents the chemical contents of the patent. A few new features are also in our develop plan, including automatic Markush structure creation from a structure library; and overlap analysis of Markush databases.

Open slides in pdf