A combination of OntheFly annotation software, the Reflect text-mining web service, and the ProMiner terminology-recognition tool has yielded a 50 percent increase in the productivity of Jackson Laboratory staff indexers for certain curation tasks. Researchers evaluated the automation tools within their Mouse Genome Informatics biocuration workflow and reported results in the journal Database. The mouse genome project is a model organism database for the mouse.
Some 1,000 articles enter the project's annotation pipeline each month, reports GenomeWeb, and 700 are selected for annotation. The project requires that the 30-member curator and support team index each paper to at least one gene entity before checking for additional characteristics.
To streamline the massive task, researchers evaluated available text-mining tools, focusing on named entity recognition software and tools that identify protein interactions and relations. A part-time curator operating manually indexes about two papers an hour, or 50 articles per week. Using the automation tools, the figure rises to 60 to 70 articles per week.
- here's the GenomeWeb article
- see the Database paper