I’ve just had a great conversation with a current OmniViz user about his use of the software in literature analysis. In his role he regularly needs to carry out a systematic survey of the literature to investigate the effects of chemicals on human disease, e.g. “What do we know about the effects of pesticide exposure on cancer?”  Gathering this information is an important step in estimating project costs, so it’s vital that searches are carried out both accurately and also to tight deadlines.

OmniViz uses clustering algorithms to sort documents from EndNote® and other sources into piles of thematically-related articles. These “themes” can be assigned automatically or dictated by the user. Thereafter, it offers a range of tools to examine the documents in more detail, such as the Galaxy view (see image). It also offers advanced text search capabilities, including a “Query by Example” option that will search for a specified number of documents that most closely resemble a chosen record or collection of records. Other tools offer different perspectives, e.g. CoMet, which produces a heat map of activity between any pair of variables, including dates, categorical values, major topics, or user-defined term lists. This allows questions to be posed such as:

• How do the reported cancer sites attributable to pesticide exposure change over time?
• Which chemicals are co-reported with different cancer types?
• What genes are implicated in the action of different carcinogens?

Using OmniViz, our customer was able to triage thousands of documents down to a manageable short list, in EndNote format, in around 2-3 hours. This compares with 2-3 weeks using conventional methods – clearly a big time saving and a more effective use of resources!

OmniViz Galaxy plot, showing an analysis of ~13,000 Medline abstracts (from author’s own collection). Records are aggregated into clusters of varying sizes; two clusters have been selected showing the top 3 topics within each cluster, in this case chemical names. Clustering can be carried out in an automated fashion or, as in this example, guided by user-defined topics.

Of course, OmniViz can also work with other data types, including numbers, categorical values and even chemical structures and gene/protein sequences. With the recent release of version 6.1 and our on-going development we’ll continue to improve the user experience by adding new functionality and support for different document formats. Let us know your thoughts at cip@instem.com.