Visit us on Facebook Visit us on Linked In Visit us on Twitter Visit us on Instagram View our RSS Feed 
Mar
16
Posted by Alison Aldrich on March 16th, 2010
Posted in: Technology
The freely available PubMed API (application programming interface) makes it possible for programmers from outside of the National Library of Medicine to develop alternatives to PubMed.gov for searching NLM’s vast database of biomedical journal literature citations. In this guest post, super searcher Kristine Ogden describes two third-party PubMed tools, HubMed and Quertle, and explains how they can be useful for specialized searching.
by Kristine Ogden
ICON Clinical Research
I’m a researcher and PubMed is my first stop when I need to find relevant biomedical literature. It offers broad, international coverage and its MEDLINE subset is a well-regarded source for indexed citations. But recently, I’ve been keeping my eye on two third-party websites that offer the ability to search and interact with PubMed data: HubMed and Quertle. Although neither of these tools has knocked PubMed from its place as my preferred search site, they do feature several helpful functions that are not available on the NCBI-maintained PubMed website. As a consequence, I find myself visiting HubMed and Quertle to help augment my PubMed searches. Hopefully, this little summary of HubMed and Quertle will pique your interest and prompt some exploration of these two web sites. Happy searching!
HubMed accurately describes itself as an “alternative interface to the PubMed medical literature database.” HubMed’s easy-to-use site has a simplified interface for searching PubMed and reviewing resulting citations. Just like the PubMed website, users may search using keywords that are translated using automatic term mapping or opt to search PubMed fields manually using the same syntax as one would on the PubMed web site (e.g., search “Arthritis/epidemiology”[Mesh]). Citation data may be exported, including exports to file formats that support the upload of citation data to reference manager software. Unlike PubMed, HubMed does not offer access to Clinical Queries, Special Topic Queries or familiar PubMed search features like the Limits screen. Users cannot query other NCBI-maintained Entrez databases from HubMed and there is limited (and, not always intuitive) access to MeSH information.
The HubMed features that I find most helpful are:
Citation Finder
Oftentimes, my colleagues and I will scan the bibliography from an article to determine if any the references cited are applicable to our research. The Citation Finder feature in HubMed allows a user to cut and paste the bibliography from a report (in Word/PDF/ASCII format), then each citation is parsed and automatically matched to its PubMed record. If no match is found, users are given the opportunity to review unmatched citations and fix any incorrectly parsed fields. In the past, I used to manually type each citation into reference manager software or look up each cited article in PubMed one-by-one. Now, I simply use this HubMed feature to download the matched citations to my reference manager software.
Refine search terms
When you run a keyword search from the HubMed main page, HubMed will suggest a selection of MeSH terms, subheadings or text words[all fields] that you may want to use to further refine your search (look at the very bottom of your result set). It’s similar to using the Search Builder feature that is located on PubMed’s Advanced Search page.
Try this search elsewhere…
Once a result set appears for a HubMed search, users have the option to run the same search in other web sites. This feature is located at the very bottom of the results screen. For example, I ran a search for “liver neoplasms”[Mesh] “epidemiology”[sh]. Here are the links I could click to run this search at a different site: Google (Scholar), PubMed, OpenDOAR, CrossRef, PMC, XplorMed
Rank Relations
Similar to the Related Articles search feature in PubMed, HubMed will gather related articles for a set of citation saved to the HubMed clipboard and present the list of resulting citations weighted by relevance. Here’s the summary from HubMed’s FAQ page:
“Select those papers that are the most interesting and add them to the clipboard. Browse around related articles using the ‘Related’ links and have a look at other papers published by the main authors by clicking on their names. Add any more relevant papers that you find to the clipboard. Once there are a reasonable number of papers in the clipboard, open the clipboard using the ‘clipboard’ button in HubMed’s menu bar. Select all the articles (using the ‘All’ checkbox at the top or bottom of the page), then press the ‘Rank Relations’ button. This will take a little while, as it is fetching the ‘related articles’ list for each of the selected articles, comparing the lists, and giving scores to articles that appear high up in each of the lists. When finished, it will return a list of the articles that were determined to be most closely related to as many as possible of the original selected articles. From this list, you can then proceed to check additional articles and repeat the ‘Rank Relations’ process until you have collected all the articles that you consider to be relevant. Pressing the ‘Display Checked Abstracts’ button will then display all those collected articles on one page, from where they can be saved, printed out or tagged.”
Abstract terms
When reviewing HubMed citation results, users have the option of viewing the citation abstract (if available). HubMed will highlight the terms in blue that are considered when running a Related Articles search in PubMed or Rank Relations search in HubMed. Simply click the TERMS link located below the abstract to shade these words.
If I select several citations of interest, I will review the terms in these citation abstracts to help me further refine my search syntax and choice of search vocabulary.
In addition to these features, I appreciate HubMed’s simple, clean user interface. There’s no clutter and it’s very easy to interact with you data and export selected citation results.
Quertle is a new biomedical search engine that is designed to support the execution of natural language searches by exploring the relationships between topics and characterizing similarities among documents that cover similar concepts. Citations are culled from PubMed, BioMed Central, and a few other sources (white papers, news, etc.). Unlike PubMed, which executes searches using automatic term mapping or a simple keyword search, Quertle is designed to support both relationship-driven searches based on a sort of conceptual index framework that classifies reported findings and attempts to define relationships between records and keyword searches. In addition, the Quertle database includes a set of specialized terms (known as PowerTerms ™) for broadly defined entities that may represent a major concept in a user’s search. Though there may be some overlap between PowerTermsTM entities, Quertle seems to be veering away from MeSH-like indexing in which an extremely detailed hierarchical system of concept terms is used to classify individual citations. When Quertle executes a search, it employs both a keyword search and search of its conceptual relationship index with the goal of retrieving a set of highly-relevant citations.
Quertle is a young service, and in my opinion, not yet robust enough or documented well enough to support most research efforts. I primarily use this site to run searches that help me uncover 5-10 articles of interest about a research topic. Then, I will review those Quertle articles to help me consider how I might want to construct my search in PubMed.
Briefly, here are some of the pros and cons of the latest version (v2) of Quertle:
PROS
CONS
Quertle has some documentation on its web site, but not enough to assist a researcher that wants to query the system. Specifically, the structural framework for how information is indexed in Quertle and how searches are processed need to be better elucidated.
ABOUT Alison Aldrich