Plugin for concept-assisted search and navigation on PUBMED

Introduction: Search of MEDLINE using the PUBMED interface still remains a preferred approach for many scientists, even in the presence of alternatives. While PUBMED provides a rich interface for querying, it does not provide deep analysis of the query results to refine queries and build hypotheses. To provide an improved search experience in PubMed, we have developed a Firefox browser plugin called “TPX plus” that enables concept-assisted search and navigation of the MEDLINE database while remaining in the PubMed website. Some of the features of the plugin include identification and highlighting of biomedical concept types like genes, processes, diseases etc. in the PubMed results page; link-outs to dictionary sources for identified terms; a Concept Explorer to view ranked automatically identified concepts in the results as well as use them for filtering/refining the search; deriving related concepts using a statistical concept-associations network; bookmarking and managing articles, storing notes and sharing comments on articles. Implementation: The core of the plugin is written in JavaScript that uses Web 2.0 technologies such as jQuery, AJAXlJSON and XUL for handling and manipulating content of web pages within the PubMed domain. 'TPX plus' uses the DOM structure of PubMed's search interface for a) building input to Annotation server by extracting appropriate content such as title, abstract text and PubMed ID from the appropriate DOM elements and b) presenting the top ranked biomedical concepts based on their relevance to the search query under concept explorer and c) highlight biomedical concepts for each abstract in the search result under article view in the PubMed's search interface. Conclusion: Its three-pronged nature of relying on PubMed as the basic indexing engine, enhancing PubMed by performing customized offline and on-the-fly analysis of the search results and most importantly being available as a browser plugin allowing use of the PubMed user-interface makes this plugin an ideal PubMed search assistant for scientists.