Migrating the eMolecules Chemical Search Engine to ChemAxon Technology
A publicly accessible web site such as emolecules.com, which offers free searching of over ten million commercially available chemical compounds, presents unique technical challenges to Relational Database (RDB) technology. RDB systems have unpredictable (and often long) search times, poor support for pagination, no feedback on search progress, no support for incremental refinement of search results, and poor support for interrupt/cancel operations. In addition, text-search engines such as Google and Bing have acclimatized users to nearly instantaneous responses, further “raising the bar” of user expectations for a chemical-search web site. Due to these limitations, in 2005 eMolecules developed a custom non-relational cheminformatics search engine that operates in conjunction with an RDB system to meet our customers’ needs via a web-based user interface.
Recent improvements in cheminformatics and relational technology have made it feasible to use ChemAxon’s JChem-PSQL “cartridge,” fully embedded in a Postgres RDB system, for the eMolecules website. In this talk, we outline the challenges and solutions involved in adapting and revising eMolecules’ technology to this RDB system. The result is a web-based chemical search system that provides nearly instantaneous responses, efficient pagination, progress feedback, incremental refinement of search results, and an interrupt/cancel feature.